Four-relationship integrated monitoring and water-fertilizer coordination management method for precision agriculture

Abstract

This invention belongs to the field of precision agriculture technology. It discloses a method for integrated monitoring of four conditions (earth, soil, and water) and coordinated water and fertilizer management for precision agriculture. First, when the top-layer spectral response after topdressing does not show an increase characteristic consistent with a preset growth pattern, a canopy penetration diagnostic mode is triggered, nutrient diagnostic channels are paused, and soil root water absorption rhythm signals and canopy upper transpiration rhythm signals are acquired. By analyzing the conduction synergy between the two, the existence of a stagnant layer in the middle and lower parts of the canopy is identified, and the vertical position of the stagnant layer is determined based on the deep distribution characteristics of the root water absorption signal. Canopy thermal infrared images are acquired, temperature heterogeneous regions are identified, and cross-validation with the location of the water and fertilizer stagnant layer is performed to generate a deficiency confirmation result in the middle and lower layers. Based on the confirmation result, a water and fertilizer decoupling control command is generated, and transpiration regulation is simultaneously implemented in the upper part of the canopy to reduce the water and fertilizer competitive advantage of the upper part. This invention improves the efficiency of water and fertilizer resource utilization and provides important technical support for modern precision agriculture.

Description

Technical Field

[0001] This invention relates to the field of precision agriculture technology, and more specifically, to a method for integrated monitoring of four conditions and coordinated water and fertilizer management for precision agriculture. Background Technology

[0002] A problem with existing technologies is that optical sensor-based nutrient diagnosis systems suffer from severe "visual blindness" when field crops enter the mid-to-late growth stages and the canopy gradually closes in. During the vigorous growth phase of tall crops such as corn, rice, and wheat, the dense upper leaves form a green barrier, causing UAV-borne spectrometers, ground-based multispectral cameras, and satellite remote sensing systems to only capture spectral information from the top of the canopy, completely obscuring the nutrient stress status of the lower leaves. Widely used vegetation indices such as NDVI and EVI encounter a "spectral saturation trap" when dealing with high biomass crops—when the canopy leaf area index exceeds a critical threshold, the sensitivity of these indices to changes in plant nutrient status decreases drastically. Even if farmers apply fertilizer according to conventional methods, the sensors cannot capture the corresponding increase in spectral response, thus giving erroneous signals of sufficient nutrition. This phenomenon is particularly prominent in high-yield fields, often causing farmers to lose effective monitoring methods during the critical period when crops need precise nutrient regulation the most, and they can only rely on experience to judge. As a result, the surface roots obtain excessive nutrients while the deep roots and middle and lower leaves are in a state of latent hunger for a long time, which ultimately affects grain filling and yield formation.

[0003] In view of this, the present invention proposes a method for integrated monitoring of four conditions and coordinated water and fertilizer management for precision agriculture to solve the above problems. Summary of the Invention

[0004] To overcome the aforementioned deficiencies of the existing technology and to achieve the above objectives, the present invention provides the following technical solution: a method for integrated monitoring of four conditions and coordinated water and fertilizer management for precision agriculture, comprising:

[0005] During the process of monitoring crops in a target area using an agricultural Internet of Things (IoT) system, the top-level spectral response sequence of the crops is obtained.

[0006] In response to the detection of a topdressing event, the top spectral response sequence did not produce an increase characteristic that conformed to the preset growth pattern, triggering the canopy penetration diagnostic mode;

[0007] In the canopy penetration diagnostic mode, the nutrient diagnostic channel based on the top spectral layer is paused to obtain the soil root water uptake rhythm signal and the upper canopy transpiration rhythm signal of the crop.

[0008] Compare the transduction coherence between soil root water uptake rhythm signals and canopy transpiration rhythm signals;

[0009] In response to the transmission synergy being lower than the preset synergy threshold, it is determined that there is a water and fertilizer blocking layer in the lower part of the crop canopy, and the vertical depth of the water and fertilizer blocking layer is determined based on the deep distribution characteristics of the soil root water absorption rhythm signal.

[0010] Acquire canopy thermal infrared images, identify vertical temperature heterogeneous regions in the canopy thermal infrared images, cross-validate the distribution depth of temperature heterogeneous regions with the vertical depth position of the water and fertilizer blocking layer, and generate the middle and lower layer deficiency confirmation results.

[0011] Based on the confirmation results of the deficit in the middle and lower layers, water and fertilizer decoupling control instructions are generated. These instructions are used to provide deep-seated, targeted water and fertilizer replenishment to vertical depths and to perform transpiration regulation on the upper canopy to reduce the competitive advantage of water and fertilizer in the upper layer.

[0012] The technical effects and advantages of this invention's integrated monitoring of four conditions and coordinated water and fertilizer management method for precision agriculture are as follows:

[0013] This invention overcomes the limitations of traditional agricultural monitoring systems in sensing the vigorous growth stages of crops, enabling comprehensive diagnostic analysis of the plant's internal nutritional status and significantly improving the spatiotemporal accuracy of water and fertilizer management. By establishing a multi-dimensional cross-validation mechanism, the risk of misjudgment in nutrient diagnosis is reduced, allowing agricultural producers to accurately identify and promptly respond to hidden nutrient stress issues in the lower layers of the crop. Regarding water and fertilizer resource allocation, this invention breaks through the spatial limitations of traditional surface fertilization, achieving on-demand distribution of nutrients in the vertical space of the plant, effectively solving the nutrient stratification phenomenon caused by competition imbalance within the canopy. This precise intervention capability not only significantly improves the absorption and utilization efficiency of water and fertilizer resources and reduces the risk of environmental runoff, but more importantly, it ensures a balanced nutrient supply to crops during critical growth stages, laying a solid foundation for grain filling and stable yield. Attached Figure Description

[0014] Figure 1 This is a schematic diagram of the integrated monitoring of four conditions and coordinated water and fertilizer management method for precision agriculture according to the present invention. Detailed Implementation

[0015] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0016] Please see Figure 1 , Figure 1This is a flowchart illustrating a method for integrated monitoring of four weather conditions and coordinated water and fertilizer management for precision agriculture, provided in an embodiment of this application. This method can be executed by an agricultural Internet of Things (IoT) system and may include:

[0017] S101: During the process of monitoring the four conditions of crops in the target area by the agricultural Internet of Things system, the top-level spectral response sequence of the crops is obtained.

[0018] In this embodiment, the agricultural Internet of Things system is a comprehensive monitoring and control platform that integrates multispectral sensors, soil moisture sensors, meteorological monitoring stations, and intelligent water and fertilizer machines. It has functions such as real-time data acquisition, remote diagnosis, and precise regulation and can be deployed in agricultural production scenarios such as fields and greenhouses.

[0019] The target area crops are specific crops within the monitoring range of the agricultural Internet of Things (IoT) system, including tall or densely planted crops such as corn, wheat, rice, and tomatoes. These crops form a closed canopy structure in the middle and late stages of growth, leading to significant differences in nutrient absorption between the top and lower layers. The "four conditions monitoring" refers to a technical system that comprehensively monitors crop growth conditions (seedling condition), soil moisture, pests and diseases, and disasters (meteorological disasters) during the crop growth process, providing data support for precision agricultural management.

[0020] Top-level spectral response sequences are sets of spectral curves collected from the top of the crop canopy in the visible, near-infrared, and short-wave infrared bands by multispectral sensors installed at a fixed height in the field (usually 30-50 cm above the top of the crop canopy) or hyperspectral cameras mounted on drones, arranged in chronological order. This sequence reflects the chlorophyll content, nitrogen level, and growth vigor of the leaves at the top of the crop canopy and is a primary data source for traditional remote sensing diagnostics.

[0021] In this embodiment, during the monitoring of crop conditions in a target area, the agricultural IoT system can automatically collect spectral reflectance data from the top of the crop canopy at preset time intervals (e.g., 10:00 AM and 3:00 PM daily), covering the red band (630-690 nm), red-edge band (690-730 nm), and near-infrared band (760-900 nm). This spectral data is then sorted by timestamp to generate a top-level spectral response sequence for subsequent nutrient status assessment.

[0022] For example, for maize crops, top-layer spectral data were collected twice daily for 30 consecutive days from the jointing stage to the tasseling stage, resulting in 60 sets of spectral reflectance records, forming a complete top-layer spectral response sequence. This sequence clearly shows the changing trends of spectral characteristics of the top leaves at different growth stages.

[0023] S102: In response to the detection of a top dressing event, the top spectral response sequence does not produce an increase characteristic that conforms to the preset growth pattern, triggering the canopy penetration diagnostic mode.

[0024] In this embodiment, a topdressing event refers to an agricultural operation during crop growth where fast-acting fertilizers such as nitrogen, phosphorus, and potassium are applied to the field according to agronomic management plans or system diagnostic recommendations. The time of occurrence, amount of fertilizer applied, and location of application are all recorded and marked by the system. Detection of a topdressing event can be achieved through the following methods: the system automatically identifies the work log of the fertilization equipment (such as a smart fertigation machine) or receives topdressing records input by management personnel to confirm the time of occurrence of the topdressing event.

[0025] The pre-defined growth pattern is a nutrient response model established based on crop physiology principles and historical field trial data. It describes the characteristic changes in the top spectral response sequence of crops under normal conditions within a specific time window after topdressing (usually 3-7 days after topdressing). Specifically, the normalized difference vegetation index (NDVI) or chlorophyll index (CI) should increase by at least 5%-15%, and the red edge position should shift 1-3 nm towards longer wavelengths, indicating an increase in nitrogen content and enhanced photosynthetic activity in the top leaves of the canopy.

[0026] The increase characteristics refer to the incremental features of spectral indices after topdressing relative to the baseline values ​​before topdressing in the top spectral response sequence, including the magnitude of the increase (e.g., NDVI increase), the rate of increase (e.g., daily NDVI growth rate), and the duration of the increase (e.g., the number of days the high value is maintained). Increase characteristics that conform to the preset growth pattern mean that these incremental indicators have all reached the preset threshold range, reflecting that nutrients have been effectively absorbed and transformed into canopy growth.

[0027] The canopy penetration diagnostic mode is a deep nutrient diagnostic mechanism designed specifically for crops with closed canopies. It is activated when the top-level spectral diagnostic mode fails. By analyzing the soil-plant hydrothermal conduction process and the temperature distribution of the canopy's vertical structure, it diagnoses whether there is water and fertilizer blockage in the middle and lower layers, avoiding the diagnostic blind spots caused by relying solely on the top-level spectrum.

[0028] In this embodiment, when a topdressing event is detected, the top-level spectral response sequences for 3 days before and 7 days after topdressing are automatically extracted, and the increment of NDVI at each time point after topdressing relative to the baseline value before topdressing is calculated. If the NDVI increment is consistently less than 5% over 7 consecutive days, and there is no significant shift in the red edge position, it is determined that the top-level spectral response sequence does not produce an increase characteristic that conforms to the preset growth pattern.

[0029] This anomaly may be caused by the following reasons: nutrients failed to reach the top of the canopy effectively after fertilization, or nutrients were intercepted in the middle and lower layers of the canopy, or insufficient soil moisture prevented nutrients from dissolving and transporting. In this case, the canopy penetration diagnostic mode is automatically triggered, pausing the diagnostic channel that relies solely on the top-level spectrum and instead initiating a deep diagnostic process based on the soil-plant hydrothermal conduction process to identify the root cause of the abnormal nutrient response.

[0030] For example, when 30 kg / mu of urea was applied to a cornfield during the large trumpet stage, the average NDVI before topdressing was 0.72. On the 3rd day after topdressing, the NDVI was only 0.73, and on the 7th day, it was 0.74, an increase of only 2.8%, far below the preset 5% threshold. An abnormal top-layer spectral response was detected, triggering the canopy penetration diagnostic mode to collect soil root water absorption rhythm signals and canopy transpiration rhythm signals to pinpoint the specific layers of nutrient blockage.

[0031] S103: In the canopy penetration diagnostic mode, the nutrient diagnostic channel based on the top-level spectrum is paused to obtain the soil root water absorption rhythm signal and the upper canopy transpiration rhythm signal of the crop.

[0032] In this embodiment, the nutrient diagnosis channel based on the top-level spectrum is the main technical approach of traditional remote sensing diagnosis, which infers the overall nutrient status of the crop by analyzing the spectral reflectance characteristics of the top of the canopy. Suspending this channel means that after the canopy penetration diagnosis mode is activated, the top-level spectral data is no longer used as the main basis for nutrient diagnosis, avoiding misjudgments caused by canopy closure, and instead relying on physiological indicators that better reflect the water and fertilizer transport process inside the canopy.

[0033] The soil root water absorption rhythm signal is obtained by monitoring the dynamic changes in soil moisture content in each layer in real time using soil moisture sensors buried at different soil depths (e.g., 10cm, 30cm, 50cm, 70cm), and combining this with root distribution density data to calculate the time-series curve of root layer water absorption rate. This signal reflects the rhythmic characteristics of crop roots absorbing water from each soil layer over a 24-hour period, including the peak water absorption time, water absorption rate, and total water absorption.

[0034] The canopy transpiration rhythm signal is generated by real-time monitoring of the transpiration rate of leaves in the upper canopy using leaf transpiration rate sensors (such as thermal diffusion stem flow meters or infrared leaf temperature sensors) installed at the top of the crop canopy, producing a time-series curve of transpiration intensity. This signal reflects the rhythm of water loss from the top leaves of the canopy throughout the day, including the peak transpiration time, transpiration rate, and total daily transpiration, and is a key indicator for assessing canopy water status and hydraulic conduction efficiency.

[0035] In this embodiment, when the canopy penetration diagnostic mode is triggered, the nutrient diagnostic channel based on the top-layer spectrum is immediately suspended, and spectral indices such as NDVI or CI are no longer used as the main basis for judging nutrient status. At the same time, the soil-plant hydrothermal conduction monitoring network is activated to continuously collect soil root water absorption rhythm signals and upper canopy transpiration rhythm signals.

[0036] Specifically, soil moisture data is read from soil moisture sensors buried at depths of 10cm, 30cm, 50cm, and 70cm, recorded every 10 minutes, and continuously monitored for 24 hours to generate time-series curves of soil moisture content for each layer. Combined with a pre-established root distribution density model (such as root length density data obtained through root drilling), the actual water absorption rate of the root system in each layer is calculated to form a rhythmic signal of soil root water absorption.

[0037] Simultaneously, stem sap flow rate data were read from a stem flow meter installed at the top of the canopy (approximately 2.5 meters above the ground, close to the spike-like leaves of the plant), recorded every 10 minutes, and continuously monitored for 24 hours to generate a transpiration rhythm signal from the upper canopy. This signal directly reflects the dynamics of water loss from the leaves at the top of the canopy and is direct evidence for assessing the efficiency of canopy water conduction.

[0038] For example, during the tasseling stage of maize, the soil root water uptake rhythm signals collected showed that: the water uptake peaked at 9:00 AM for the 10cm shallow root layer (0.8mm / h), at 10:00 AM for the 30cm layer (0.6mm / h), at 11:00 AM for the 50cm layer (0.4mm / h), and at 12:00 PM for the 70cm deep root layer (0.2mm / h). Simultaneously, the transpiration rhythm signal from the upper canopy showed that the transpiration rate peaked at 1:00 PM (4.5 mmol·m³). -2 ·s -1） These data provide a foundation for subsequent analysis of transmission synergy.

[0039] S104: Compare the transduction synergy between soil root water absorption rhythm signals and canopy transpiration rhythm signals.

[0040] In this embodiment, the conduction synergy is a comprehensive indicator reflecting the degree of temporal coupling between the soil root water uptake process and the canopy transpiration process, used to assess the conduction efficiency of water transport from the soil root layer to the top of the canopy. In a normal soil-plant-atmosphere continuum (SPAC) system, root water uptake and canopy transpiration should exhibit highly synchronized rhythmic characteristics, that is, the peak time point of root water uptake should be slightly earlier than or synchronous with the peak time point of canopy transpiration, with a time offset usually within 30 minutes, indicating that the water conduction path is unobstructed and without significant obstruction.

[0041] In this embodiment, the core method for comparing conduction synergy is to extract key time feature points from the soil root water uptake rhythm signal and the canopy transpiration rhythm signal, and calculate the time offset and correlation coefficient between the two. The specific steps are as follows:

[0042] First, from the 24-hour soil root water uptake rhythm signal, the global peak time point of the water uptake rate is identified and denoted as . Typically, the peak of root water absorption occurs between 9:00 AM and 12:00 PM during the daily cycle, when soil temperature is suitable, transpiration pull is strong, and root activity is high.

[0043] Secondly, the global peak time point of transpiration rate was identified from the 24-hour upper canopy transpiration rhythm signal, and denoted as _____. Typically, the peak of canopy transpiration occurs between 12:00 and 14:00 in the diurnal cycle, when solar radiation is strongest, temperature is highest, and saturated vapor pressure difference is greatest.

[0044] Then, calculate the time offset. .like Within a reasonable range (e.g., 0-30 minutes), it indicates that after the roots absorb water, the water can be promptly transported to the top of the canopy to support transpiration, demonstrating high synergistic transport; if A significant increase (e.g., exceeding 60 minutes or even several hours) indicates that the water absorbed by the roots is hindered in its transport to the canopy, resulting in low conduction synergy.

[0045] In addition, the cross-correlation coefficient R between the water absorption rhythm signal and the transpiration rhythm signal is calculated. An R value close to 1 indicates that the fluctuation trends of the two are highly consistent, while a low R value indicates that the two are decoupled, which also reflects a decrease in the degree of conduction coordination.

[0046] For example, in canopy penetration diagnosis of cornfields, the extracted soil root water uptake rhythm signals show: the comprehensive water uptake peak time point At 10:30 AM, the weighted average water absorption rate of the root system at each layer reached 0.65 mm / h. The transpiration rhythm signal from the upper canopy shows the peak transpiration time. At 2:00 PM, the peak transpiration rate was 4.8 mmol·m³. -2 ·s -1 The calculated time offset ΔT = 14:00 - 10:30 = 3.5 hours, far exceeding the normal range of 30 minutes. At the same time, the cross-correlation coefficient R = 0.42, which is significantly lower than the normal threshold of 0.75, indicating that the conduction coordination is seriously low.

[0047] S105: In response to the transmission synergy being lower than the preset synergy threshold, it is determined that there is a water and fertilizer blocking layer in the lower part of the crop canopy, and the vertical depth of the water and fertilizer blocking layer is determined based on the deep distribution characteristics of the soil root water absorption rhythm signal.

[0048] In this embodiment, the preset coordination threshold is a critical value pre-set based on extensive field trial data and crop water physiology principles to determine whether the conduction coordination is normal. This threshold typically includes two dimensions: a time offset threshold (e.g., ΔT ≤ 60 minutes) and a cross-correlation coefficient threshold (e.g., R ≥ 0.70). When any indicator of the actual conduction coordination is lower than the corresponding threshold, the conduction coordination is determined to be lower than the preset coordination threshold.

[0049] The water and fertilizer retardation layer refers to a specific spatial layer in the vertical structure of a crop canopy where the upward transport of water and nutrients is hindered due to dense foliage, overlapping stems, or local differences in physiological activity. This layer is usually located in the lower to middle part of the canopy (such as the middle leaf layer of a corn plant, 0.8-1.5 meters above the ground). It manifests as a paradoxical phenomenon where water and fertilizer absorbed by the roots below this layer cannot be transported upwards to the top of the canopy in a timely manner, resulting in abnormal spectral response at the top layer while the roots at the bottom continue to absorb nutrients normally.

[0050] The deep layer distribution characteristics refer to the spatiotemporal distribution patterns of root water uptake rhythm signals at different soil depths, including the differences in peak water uptake times at each layer, the vertical gradient of water uptake rates, and the response delays between water uptake signals and canopy transpiration signals. By analyzing these characteristics, it is possible to pinpoint the soil depth layer where the root water uptake signal and canopy transpiration signal have the greatest transmission delay, and thus infer the location of the corresponding vertical depth of the canopy, which is the location of the water and fertilizer retardation layer.

[0051] Vertical depth refers to the specific height range of the water and fertilizer retention layer in the vertical space of the crop canopy. It is usually expressed as the height from the ground (unit: meter or centimeter) or as a percentage of the plant height (such as 40%-60% of the total plant height), providing a precise spatial positioning basis for subsequent deep-layer targeted water and fertilizer replenishment.

[0052] In this embodiment, when the detected conduction synergy is lower than a preset synergy threshold (e.g., ΔT = 3.5 hours > 60 minutes, and R = 0.42 < 0.70), it is immediately determined that a water and fertilizer retention layer exists in the lower part of the crop canopy. To accurately locate the vertical depth of this retention layer, the deep distribution characteristics of the soil root water absorption rhythm signal are further analyzed.

[0053] Specifically, the peak time points of root water absorption at four depths of 10cm, 30cm, 50cm, and 70cm were extracted and denoted as T. 10 T 30 T 50 T 70 Then, the peak water absorption time points of each layer and the peak evaporation time point of the upper canopy were calculated. The response delays between them are denoted as ΔT. 10 ΔT 30ΔT 50 ΔT 70 .

[0054] According to the physical laws of soil-plant water conduction, the soil depth layer with the greatest response time delay corresponds to the root system that requires the longest conduction path or encounters the greatest conduction resistance to reach the top of the canopy. Therefore, the vertical spatial interval between this depth layer and the top of the canopy is the most likely location of the water and fertilizer retention layer.

[0055] The soil depth layer with the longest response time delay (e.g., the 50cm layer, ΔT) 50 =4.2 hours (maximum value) is used as the reference point for the lower boundary of the water and fertilizer barrier layer. Combined with the root-canopy correspondence model of the crop (e.g., the root system of corn at a depth of 50cm mainly supports the middle and lower leaves of the plant, corresponding to a height of about 0.8-1.2 meters), the vertical depth of the water and fertilizer barrier layer is determined to be in the range of 0.8-1.2 meters from the ground.

[0056] For example, in a cornfield, analysis revealed that the peak water absorption T at the 10cm layer... 10 =9:00, response delay ΔT 10 =5 hours; Peak water absorption T at a 30cm layer 30 =10:00, response delay ΔT 30 =4 hours; Peak water absorption T at 50cm layer 50 =11:00, response delay ΔT 50 =3 hours; Peak water absorption T at 70cm layer 70 =12:00, response delay ΔT 70 =2 hours. Although the 10cm layer had the largest response time delay, considering the high root density and large total water absorption in the shallow layer, the increased delay was likely due to competition from transpiration pull rather than physical blockage. Based on the combined data of root distribution density and water absorption rate, the vertical depth of the canopy (0.9-1.3 meters) corresponding to the 30-50cm soil layer was ultimately determined to be the core area of ​​the water and fertilizer blocking layer.

[0057] S106: Acquire canopy thermal infrared images, identify vertical temperature heterogeneous regions in the canopy thermal infrared images, cross-validate the distribution depth of temperature heterogeneous regions with the vertical depth position of the water and fertilizer blocking layer, and generate the middle and lower layer deficiency confirmation results.

[0058] In this embodiment, the canopy thermal infrared image is a two-dimensional thermal image reflecting the temperature distribution at different spatial locations of the crop canopy, obtained by vertically scanning the crop canopy with a thermal infrared camera mounted on a drone or a fixed support in the field. The grayscale value or pseudo-color of each pixel in the image corresponds to the surface temperature of the leaf or stem at that location, typically ranging from 25-40°C, with a resolution of up to 0.1°C.

[0059] Temperature heterogeneity refers to spatial regions in a canopy thermal infrared image where there are significant temperature differences between adjacent pixel rows in the vertical direction. Under normal circumstances, the temperature in the vertical direction of the canopy should show a smooth and gradual change, with the temperature gradually increasing from the bottom to the top (because the top receives greater light intensity). If the temperature of a certain height layer is suddenly lower than the layers above and below, forming a "cold spot," or suddenly higher than the layers above and below, forming a "hot spot," then that layer is a temperature heterogeneity region, which usually indicates abnormalities in the moisture state or physiological activity of that layer.

[0060] Distribution depth refers to the height of a temperature heterogeneous region in the vertical space of the canopy. It is converted from the pixel row coordinates of the thermal infrared image to the actual height above the ground (unit: meters) through image processing algorithms. Geometric correction is required by combining parameters such as the drone's flight altitude, camera field of view, and crop height.

[0061] Cross-validation is a diagnostic method that integrates multi-source data. It compares the vertical depth of the water and fertilizer retention layer determined by hydraulic conduction analysis with the depth of temperature heterogeneous regions identified by thermal characterization to determine whether they point to the same spatial layer. If the depth difference is less than a preset threshold (e.g., less than 0.3 meters), the cross-validation is considered successful, confirming the presence of water and fertilizer deficiency at that layer. If the difference is large, further analysis is needed to avoid misdiagnosis.

[0062] The deficiencies in the middle and lower layers are confirmed by cross-validation, resulting in a clear diagnostic conclusion indicating water and fertilizer deficiencies in specific layers in the middle and lower part of the canopy. This includes the vertical depth range of the deficient layer, the degree of deficiency (e.g., mild, moderate, severe), and recommended replenishment strategies, providing a basis for decision-making in subsequent water and fertilizer decoupling regulation.

[0063] In this embodiment, after determining the vertical depth of the water and fertilizer retention layer, the thermal infrared camera mounted on the UAV is immediately activated to perform a vertical thermal imaging scan of the crop canopy in the target area. The UAV hovers 5 meters above the field, with the camera lens pointing vertically downwards. The resolution is set to 640×512 pixels, and the field of view covers a 3m×2.4m canopy area. The shooting time is selected between 10:00 and 11:00 am on a clear, windless day (when solar radiation is stable and the temperature difference in the canopy is most obvious).

[0064] After the image is captured, import the thermal infrared image into the image processing module and perform the following recognition process:

[0065] First, geometric correction is performed on the image to convert pixel coordinates into actual spatial coordinates. Based on the drone's altitude, camera parameters, and crop height, a correspondence between pixel rows and their height above the ground is established. For example, if a cornfield has a crop height of 2.8 meters and the image has 512 rows of pixels, then the vertical height interval for each row of pixels is approximately 2.8 / 512 ≈ 0.0055 meters per row.

[0066] Secondly, extract the temperature difference sequence between adjacent pixel rows in the vertical direction. Scan the image row by row and calculate the average temperature T of the pixels in the i-th row. i The average temperature T of the pixels in the (i+1)th row (i+1) Temperature difference ΔT between i =T (i+1) -T i A temperature difference sequence is generated.

[0067] Then, temperature abrupt change points are identified in the temperature difference sequence. A preset temperature difference threshold of 2℃ is set (based on field experiments, the temperature difference between adjacent 5cm heights in a normal canopy is typically less than 1.5℃). When |ΔT i When the temperature exceeds 2℃, it is determined that there is a temperature abrupt change between the i-th row and the (i+1)-th row. The height of this pixel row is the distribution depth of the temperature heterogeneous region.

[0068] Next, the depth of the identified temperature heterogeneous region distribution is compared with the vertical depth of the water and fertilizer retardation layer obtained from the aforementioned hydraulic conduction analysis. The depth deviation between the two is calculated, and if the deviation is less than a preset deviation threshold (e.g., 0.3 meters), the cross-validation is considered successful.

[0069] Finally, the results confirming the deficiency in the middle and lower layers are generated, and an example report is output: "Cross-validation confirms that there is a moderate water and fertilizer deficiency in the 1.0-1.3 meter height layer of the canopy (accounting for 36%-46% of the plant height) of the target cornfield. This is manifested by the leaf temperature in this layer being 2.5℃ lower than that in the layers above and below, and the response time from root water absorption to transpiration in this layer being up to 4 hours. It is recommended to implement deep-layer targeted water and fertilizer supplementation."

[0070] For example, in a cornfield, acquired thermal infrared images showed that at a height of 1.1 meters above the ground (corresponding to row 320 pixels), the average temperature of this row was 28.5℃, while the average temperature at 1.15 meters above (row 330) was 31.2℃, and the average temperature at 1.05 meters below (row 310) was 30.8℃. The temperature abrupt changes were 2.7℃ and 2.3℃ respectively, both exceeding the 2℃ threshold, identifying this as a region of temperature heterogeneity. This distribution depth of 1.1 meters closely matches the aforementioned vertical depth of the water and fertilizer retention layer (0.9-1.3 meters) derived from hydraulic conduction analysis, with a depth deviation of only 0.1 meters. Cross-validation passed, confirming the presence of water and fertilizer deficit in this layer.

[0071] S107: Based on the confirmation results of the deficiency in the middle and lower layers, a water and fertilizer decoupling regulation instruction is generated. The water and fertilizer decoupling regulation instruction is used to carry out deep-seated directional water and fertilizer replenishment to the vertical depth position, and to perform transpiration regulation on the upper part of the canopy to reduce the water and fertilizer competitive advantage of the upper part.

[0072] In this embodiment, the water and fertilizer decoupling control command is a comprehensive control command set generated by the system based on the confirmation results of the deficiency in the middle and lower layers, taking into account the location of the blocking layer, the degree of deficiency, and the crop growth stage. It includes parameters for deep-layer targeted water and fertilizer recharge and parameters for transpiration inhibition in the upper canopy. This command set includes specific parameters such as the depth of the deep fertilization probe (e.g., 1.2 meters), the duration of the fertilization pulse (e.g., 30 seconds each time), the interval period (e.g., once every 2 hours), the water and fertilizer ratio (e.g., nitrogen concentration 200 mg / L), the spraying concentration of the transpiration inhibitor (e.g., 50 ppm of ABA analog), and the spraying time (e.g., 30 minutes before deep recharge), ensuring that the control action is precise and executable.

[0073] Deep-layer targeted fertilization is a precision fertilization technology that targets the water and fertilizer retention layer in the lower part of the canopy. By controlling the deep fertilization actuator installed in the field, the fertilization probe (similar to a soil drilling device) is lowered to a determined vertical depth (e.g., 50cm soil depth corresponding to a canopy height of 1.2 meters). A water-fertilizer solution containing dissolved fast-acting fertilizer is injected into this depth layer, allowing nutrients to directly reach the root distribution area below the retention layer, bypassing the competitive absorption by the upper roots, thus achieving "vertical stratified targeted delivery" of nutrients.

[0074] Transpiration regulation involves spraying transpiration inhibitors (such as abscisic acid ABA analogs, antitranspirants, etc.) onto the upper leaves of the canopy to temporarily reduce the stomatal conductance of the upper leaves and slow down the transpiration rate. This weakens the "siphon pull" of the upper leaves on water, allowing the deeply injected water and fertilizer solution to diffuse laterally to the middle and lower layers and slowly move upwards, rather than being quickly drawn away by the upper roots, thus avoiding the vicious cycle of "upper layer interception, lower layer still lacking".

[0075] The competitive advantage of upper-level water and fertilizer distribution refers to a physiological competition phenomenon in a closed canopy where, due to the strong light and vigorous transpiration of the top leaves, a powerful transpiration pull is generated. This causes water and fertilizer absorbed by the shallow roots to be preferentially supplied to the upper leaves, while the water and fertilizer absorbed by the middle and lower roots are also taken by the upper leaves. As a result, even if there are nutrients near the roots, the middle and lower leaves cannot utilize them. This advantage is reduced through transpiration regulation, with the aim of temporarily promoting a more balanced distribution of nutrients in the vertical direction of the canopy.

[0076] In this embodiment, based on the confirmation results of the deficiency in the middle and lower layers, water and fertilizer decoupling control instructions are automatically generated. The specific process is as follows:

[0077] First, based on the vertical depth of the water and fertilizer retention layer (e.g., 1.0-1.3 meters high), and combined with the crop root-soil depth correspondence model, the target depth of the deep fertilization probe is determined. For example, the leaf layer at a height of 1.2 meters in corn is mainly supported by roots at a depth of 40-60 cm, so the probe depth is set to 50 cm.

[0078] Secondly, based on the degree of deficiency (e.g., moderate deficiency) and soil texture (e.g., loam, with moderate water retention capacity), calculate the injection pulse parameters for deep-seated targeted water and fertilizer resupply. For example, set each injection to last 30 seconds, injecting approximately 500 mL of water and fertilizer solution (nitrogen concentration 200 mg / L, equivalent to 100 mg of nitrogen per injection), with an interval of 2 hours, for a total of 300 mg of nitrogen, approximately 0.3 g / plant.

[0079] Secondly, based on the peak intensity of the transpiration rhythm signal in the upper canopy (e.g., peak transpiration rate of 4.8 mmol·m⁻¹), -2 ·s -1） Calculate the concentration and timing of the transpiration inhibitor to be sprayed. For example, select an ABA analog at a concentration of 50 ppm and set the spraying time to 30 minutes before the start of deep nutrient supply to ensure that the inhibitor takes effect before injecting water and fertilizer, thus preventing the upper roots from being immediately pulled away.

[0080] Finally, the above parameters are integrated into a water and fertilizer decoupled control command, which is then sent to the deep fertilization actuator and spraying drone in the field through the Internet of Things communication module to start the automatic execution process.

[0081] For example, in a cornfield, the generated water and fertilizer decoupling control instructions are as follows: "Probe to a depth of 50 cm → Spray ABA 50 ppm to the top of the canopy (9:00 AM) → Wait 30 minutes → Initiate deep fertilization (9:30 AM), 30 seconds each time, 2-hour intervals, for 3 consecutive times → Monitor the transpiration rhythm signal to decrease to 3.0 mmol·m -2 ·s -1 The following will be confirmed to be effective. "After implementation, the NDVI of the lower and middle leaves of the canopy was monitored to increase from 0.65 to 0.72 within 48 hours, an increase of 10.8%, which is significantly higher than that of the control field with only top fertilizer application (an increase of only 3.2%), proving that the decoupled regulation of water and fertilizer effectively alleviated the deficiency in the lower and middle layers.

[0082] As can be seen from the above, this embodiment, through comprehensive analysis of soil root water absorption rhythm signals, canopy transpiration rhythm signals, and canopy thermal infrared images in the canopy penetration diagnostic mode, accurately locates the vertical depth of the water and fertilizer retention layer, and generates a deficiency confirmation result for the middle and lower layers based on cross-validation. Subsequently, through water and fertilizer decoupling control commands, the synergistic effect of deep-layer targeted water and fertilizer replenishment and upper canopy transpiration regulation is realized, effectively solving the technical problem of insufficient nutrient absorption in the middle and lower layers of crops in closed canopies, avoiding the resource waste of traditional topdressing with "excess in the upper layer and deficiency in the lower layer," significantly improving fertilizer utilization efficiency and the vertical growth balance of the crop canopy, and providing an innovative technical path for nutrient management in precision agriculture.

[0083] In one embodiment of this application, in response to the detection of a topdressing event and the failure of the top spectral response sequence to produce an increase characteristic conforming to a preset growth pattern, a canopy penetration diagnostic mode is triggered, including:

[0084] S201: After multiple consecutive historical topdressing events, extract the spectral increments after each topdressing event from the top-level spectral response sequence.

[0085] In this embodiment, historical topdressing events refer to all topdressing operations recorded during the current crop growing season, including the time of each topdressing, fertilizer type (e.g., urea, compound fertilizer), fertilizer application rate (e.g., 30 kg / mu), and application method (e.g., broadcasting, drip irrigation). "Successive times" means at least three or more topdressing events to ensure a sufficient sample size for trend analysis and avoid the randomness of single data points.

[0086] Spectral increment refers to the increase in a specific spectral index (such as NDVI, red-edge position REP, or chlorophyll index CI) in the top spectral response sequence relative to the baseline value before fertilization, after each topdressing event. The specific calculation method is: Spectral increment = mean spectral index on day 7 after topdressing - mean spectral index on days 3 before topdressing. This indicator directly reflects the stimulating effect of topdressing on the spectral characteristics of the top of the canopy.

[0087] In this embodiment, all historical topdressing event records for the current growing season are first retrieved from the database. For example, the cornfield was topdressed three times during the jointing stage, the large trumpet stage, and the tasseling stage, on June 10, June 25, and July 8, respectively, with a fertilizer application rate of 30 kg / mu of urea in each case.

[0088] Then, for each topdressing event, NDVI data for the 3 days before and 7 days after topdressing were extracted from the top-level spectral response sequence. For example, for the first topdressing (June 10): the mean NDVI for the 3 days before topdressing (June 7-9) was 0.58, and the mean NDVI for the 7 days after topdressing (June 11-17) was 0.66, with a spectral increment of 0.66 - 0.58 = 0.08 (an increase of 13.8%). For the second topdressing (June 25): the mean NDVI before topdressing was 0.68, and after topdressing it was 0.72, with a spectral increment of 0.04 (an increase of 5.9%). For the third topdressing (July 8): the mean NDVI before topdressing was 0.72, and after topdressing it was 0.73, with a spectral increment of 0.01 (an increase of 1.4%).

[0089] Arrange the spectral increments after these three topdressings in chronological order to form a spectral increment sequence: [0.08, 0.04, 0.01], providing a data basis for subsequent trend analysis.

[0090] S202: If the spectral increment shows a decreasing trend with each topdressing, and the crop biomass accumulation rate remains at the preset growth level, then it is determined that the top-level spectral response sequence has not produced an increase characteristic that conforms to the preset growth law.

[0091] In this embodiment, the successive decay trend refers to a situation in the spectral increment sequence where the spectral increment of a subsequent topdressing is significantly smaller than that of the previous one, and this trend is observed in multiple consecutive topdressings. This indicates that the stimulating effect of topdressing on the top-level spectrum gradually weakens or even becomes ineffective. The criterion is: if the (i+1)th spectral increment < the ith spectral increment × the decay coefficient (e.g., 0.70), and this condition is met for two or more consecutive times, then the decay trend is confirmed.

[0092] Biomass accumulation rate refers to the increase in aboveground dry matter weight of a crop per unit time (e.g., per week). It is estimated through periodic sampling (e.g., collecting 5 plant samples every 10 days, drying and weighing) or based on remote sensing inversion models using canopy height and leaf area index (LAI), and is expressed in g / (plant·day) or kg / (acre·day). This indicator reflects the overall growth rate of the crop.

[0093] The preset growth level is a biomass accumulation rate threshold set according to the standard growth curve of the crop variety at a specific growth stage. For example, for maize from the tasseling stage to the large tasseling stage, the normal biomass accumulation rate should be ≥8g / (plant·day). If the measured value reaches or exceeds this threshold, it indicates that the overall crop growth is normal and the nutrient supply is sufficient.

[0094] In this embodiment, trend analysis is first performed on the extracted spectral increment sequence. Taking the aforementioned example, the spectral increment sequence is [0.08, 0.04, 0.01]. The second increment 0.04 = the first 0.08 × 0.50 < 0.08 × 0.70, and the third increment 0.01 = the second 0.04 × 0.25 < 0.04 × 0.70. The attenuation condition is met twice consecutively, confirming a gradual attenuation trend.

[0095] Simultaneously, biomass accumulation rate data for the same period were retrieved. For example, through field sampling and measurement, the biomass accumulation rate of corn was 9.2 g / (plant·day) after the first topdressing (June 10-25), 8.8 g / (plant·day) after the second topdressing (June 25-July 8), and 8.5 g / (plant·day) after the third topdressing (July 8-20), all of which were ≥ the preset growth level of 8 g / (plant·day), indicating that the overall growth rate of the crop did not decrease, and the total amount of nutrients absorbed was still increasing.

[0096] A contradictory phenomenon arises: the spectral increase continues to decrease after topdressing, but the biomass accumulation rate remains normal. This indicates that the applied fertilizer was indeed absorbed by the crop and converted into biomass, but this was not reflected in the spectral characteristics of the top canopy. In other words, nutrients were trapped in the lower and middle layers of the canopy and not effectively transported to the top. Based on this, it is determined that the spectral response sequence of the top layer did not produce an increase characteristic consistent with the preset growth pattern (the preset pattern is that topdressing should stimulate a continuous increase in the top layer spectrum, rather than a decrease).

[0097] For example, in a tomato greenhouse, after four consecutive applications of potassium fertilizer, the spectral increment sequence was [0.12, 0.08, 0.05, 0.02], showing a significant decrease. However, the fruit enlargement rate (a measure of biomass accumulation) increased from 8 g / (plant·day) to 10 g / (plant·day), exceeding the preset level of 7 g / (plant·day). This indicated an abnormal spectral response in the top layer, triggering a canopy penetration diagnosis.

[0098] S203: Suspend the nutrient diagnostic pathway based on top-level spectra and activate the canopy penetration diagnostic mode based on soil-plant hydrothermal conduction processes.

[0099] In this embodiment, suspending the nutrient diagnosis channel based on the top-level spectrum means stopping the use of top-level spectral indices such as NDVI and CI as the main basis for judging the nutrient status of crops, and no longer generating fertilization recommendations based on spectral thresholds (such as NDVI < 0.60 indicating nitrogen deficiency), so as to avoid misdiagnosis caused by canopy closure (such as the top-level spectrum being normal but the middle and lower layers actually lacking fertilizer, or the top-level spectrum being low but the overall crop not lacking fertilizer).

[0100] Activating the canopy penetration diagnostic model based on soil-plant hydrothermal conduction processes refers to initiating a deep diagnostic process that does not rely on top-level spectra but on physiological and ecological indicators such as soil moisture dynamics, root water uptake rhythm, stem sap flow rate, and canopy temperature distribution. By analyzing the conduction efficiency of water and fertilizer in the soil-root-stem-leaf continuum, the specific layers of nutrient blockage can be located.

[0101] In this embodiment, once it is determined that the top-level spectral response sequence does not produce an increase characteristic that conforms to the preset growth law, the following operations are immediately performed:

[0102] First, stop automatically pushing fertilization suggestions based on indicators such as NDVI and CI to avoid making incorrect decisions based on outdated spectral data.

[0103] Second, the following monitoring devices are automatically activated: soil moisture sensor network (to collect water content in each layer), stem flow meter (to collect stem sap flow rate), canopy temperature sensor (to collect leaf temperature), and thermal infrared camera (to capture canopy thermal images), to begin continuously collecting relevant data on the soil-plant water and heat conduction process.

[0104] Third, the preset "hydrothermal conduction diagnostic algorithm" is invoked to perform real-time analysis on the collected data, including calculating indicators such as root water absorption rhythm signal, canopy transpiration rhythm signal, and conduction coordination degree, before proceeding to the subsequent retardation layer location process.

[0105] For example, after three topdressings at the jointing, booting, and heading stages in a wheat field, the spectral increments were 0.10, 0.06, and 0.02, respectively, showing a significant decrease. However, the number of grains per ear (reflecting biomass accumulation) increased from 35 grains / ear to 42 grains / ear, exceeding the preset level of 32 grains / ear. This indicates an abnormal spectral response. The top-level spectral diagnostic channel is immediately paused, and the canopy penetration diagnostic mode is activated. Data on soil moisture in the 0-60cm layers and canopy sap flow are then collected, initiating the retardation layer localization process.

[0106] In a watermelon cultivation example, after three consecutive historical topdressing events (days 8, 16, and 24, with 15 kg / mu of bio-organic fertilizer applied each time), the spectral increments after each topdressing were extracted from the top-level spectral response sequence of an organic watermelon planting base. After the first topdressing, the reflectance in the near-infrared band (850 nm) increased from 62% to 71%, with a spectral increment of 9%. After the second topdressing, the reflectance increased from 70% to 75%, with a spectral increment of 5%. After the third topdressing, the reflectance increased from 74% to 76%, with a spectral increment of only 2%. The spectral increment showed a decreasing trend with each topdressing (9%→5%→2%).

[0107] However, the cumulative biomass rates of watermelons measured during the same period were 86 g / plant / day, 89 g / plant / day, and 91 g / plant / day, respectively, still maintaining the preset growth level (85-100 g / plant / day), and even slightly increasing, indicating that the actual nutrient utilization efficiency of the plants was good. It was determined that the top-layer spectral response sequence did not produce the amplification characteristics consistent with the preset growth pattern (the amplification should be stable or increase with increasing biomass, but it actually decreased). Therefore, the nutrient diagnosis channel based on the top-layer spectrum was suspended, and the canopy penetration diagnosis mode based on the soil-plant hydrothermal conduction process was activated to avoid nutrient misjudgment due to spectral signal failure (such as mistakenly believing it to be nutrient deficiency and applying excessive topdressing).

[0108] As can be seen from the above, this embodiment extracts the spectral increments after multiple consecutive historical topdressing events, analyzes the attenuation trend with the number of topdressings, and combines this with whether the biomass accumulation rate remains normal to accurately identify the abnormal phenomenon of "nutrients being absorbed but not reflected in the top-layer spectrum." After confirming the failure of the top-layer spectrum diagnosis, this channel is promptly paused and the canopy penetration diagnosis mode is activated. This avoids the diagnostic blind spots caused by traditional methods relying solely on the top-layer spectrum, ensuring accurate location of the nutrient blocking layer even under canopy closure conditions. This lays a reliable diagnostic foundation for subsequent precise regulation and significantly improves the adaptability and accuracy of precision agriculture nutrient management.

[0109] In one embodiment of this application, comparing the transduction coherence of soil root water uptake rhythm signals and canopy transpiration rhythm signals includes:

[0110] S301: Within the preset daily monitoring window, extract the peak water absorption time of the soil root water absorption rhythm signal and the peak transpiration time of the canopy transpiration rhythm signal.

[0111] In this embodiment, the preset daily monitoring window refers to a continuous 24-hour time window pre-set by the system for collecting and analyzing soil root water absorption rhythm signals and canopy transpiration rhythm signals. To ensure data representativeness, the monitoring window is typically selected on a typical sunny day without rainfall (to avoid rhythm disruption caused by rainfall or cloudy weather), starting at 00:00 AM and ending at 00:00 AM the following day, covering the complete diurnal cycle.

[0112] The peak water absorption time point refers to the specific moment when the soil root water absorption rhythm signal reaches its maximum value within a 24-hour diurnal monitoring window. The specific identification method is as follows: peak values ​​are detected for the root water absorption rate time-series curves of each soil layer (e.g., 10cm, 30cm, 50cm, 70cm), and then the weighted average of the water absorption rate of each layer is calculated (the weight is the root length density of each layer), resulting in a comprehensive root water absorption rate curve. The time point corresponding to the global maximum value of this curve is the peak water absorption time point.

[0113] The peak transpiration time point refers to the specific moment when the transpiration rhythm signal in the upper canopy reaches its maximum value within a 24-hour diurnal monitoring window. The specific identification method is as follows: peak detection is performed on the time-series curve of the transpiration rate of the upper canopy leaves (collected by a stem flow meter or leaf temperature sensor) to find the time point corresponding to the global maximum value of the transpiration rate.

[0114] In this embodiment, a preset daily monitoring window is first determined. For example, July 15 (sunny, no rain, maximum temperature 32°C, relative humidity 60%) is selected as the monitoring day, and the monitoring window is from 00:00 on July 15 to 00:00 on July 16, a total of 24 hours.

[0115] Then, time-series data of soil moisture content in each layer within the monitoring window were read from the soil moisture sensor network, with one sampling point every 10 minutes, for a total of 144 data points / layer. Combined with the root distribution density model (e.g., 40% root length density in the 10cm layer, 30% in the 30cm layer, 20% in the 50cm layer, and 10% in the 70cm layer), the root water absorption rate of each layer was calculated, and a weighted average was obtained according to root length density to obtain the comprehensive root water absorption rate curve.

[0116] For example, the water absorption rate of the 10cm layer peaks at 0.8mm / h at 10:00, the 30cm layer at 0.6mm / h at 10:30, the 50cm layer at 0.4mm / h at 11:00, and the 70cm layer at 0.2mm / h at 11:30. After weighted averaging, the overall water absorption rate peaks at 0.62mm / h at 10:20, therefore the peak water absorption time is 10:20.

[0117] Simultaneously, time-series data of sap flow rate in the upper canopy stems within the monitoring window were read from the stem flow meter, with one sampling point every 10 minutes. The sap flow rate curve showed that it reached a peak of 5.2 mmol·m³ at 13:40. -2 ·s -1（ (corresponding to the peak transpiration rate), therefore the peak transpiration time is 13:40.

[0118] S302: Calculate the time offset between the peak water absorption time point and the peak evaporation time point.

[0119] In this embodiment, the time offset refers to the time difference between the peak transpiration time point and the peak water absorption time point, calculated as: Time offset ΔT = Peak transpiration time point - Peak water absorption time point. This indicator directly reflects the time delay required for water absorbed by the roots to reach the top of the canopy and participate in transpiration, and is a core parameter for evaluating water conduction efficiency.

[0120] In this embodiment, based on the previously extracted peak water absorption time point (10:20) and peak evaporation time point (13:40), the time offset is calculated as follows: ΔT = 13:40 - 10:20 = 3 hours 20 minutes = 200 minutes.

[0121] This time offset reflects the following physiological process: the root system reaches its maximum water absorption rate at 10:20. The absorbed water needs to pass through the xylem of the root system → xylem of the stem → xylem of the leaf → mesophyll cells → stomata, and is finally lost through transpiration at 13:40. Under normal circumstances, this conduction process should be completed within 30-60 minutes (for crops with a plant height of 2-3 meters), but the measured ΔT = 200 minutes, far exceeding the normal range, indicating a significant blockage in the conduction pathway.

[0122] S303: If the time offset is greater than the preset time offset tolerance, and the peak evaporation time point lags behind the peak water absorption time point, then the conduction synergy is determined to be lower than the preset synergy threshold.

[0123] In this embodiment, the preset time offset tolerance is a pre-set time threshold used to determine whether the conduction efficiency is normal, based on crop water physiology principles and field test data. For tall crops (such as corn and sorghum) with a plant height of 2-3 meters, the normal time offset tolerance is usually set to 60 minutes; for short crops (such as soybeans and cotton) with a plant height of less than 1 meter, the time offset tolerance can be set to 30 minutes. If the measured time offset exceeds this tolerance, it indicates that the conduction process is abnormally slow.

[0124] The transpiration peak time point lagging behind the water absorption peak time point is a logical condition for determining the correctness of the conduction direction. In a normal SPAC system, the direction of water conduction is "root water absorption → canopy transpiration," therefore, the water absorption peak should precede or be synchronous with the transpiration peak. If the transpiration peak is delayed, and the delay time is too long, it indicates that the water absorbed by the roots is blocked during the upward conduction process. If the transpiration peak is earlier than the water absorption peak, it may indicate that the canopy transpiration consumes the water stored the previous day, which needs to be analyzed separately.

[0125] The preset coordination threshold is a comprehensive threshold used to determine whether the conduction coordination meets the standard. It typically includes a time offset threshold (e.g., ≤60 minutes) and a cross-correlation coefficient threshold (e.g., ≥0.70). When any of the measured conduction coordination indicators fails to reach the corresponding threshold, the conduction coordination is determined to be below the preset coordination threshold.

[0126] In this embodiment, it is first determined whether the time offset exceeds the preset tolerance. For corn plants with a height of 2.8 meters, the preset time offset tolerance is 60 minutes, while the actual measured ΔT = 200 minutes > 60 minutes, which meets the condition of "greater than the preset tolerance".

[0127] Secondly, determine whether the peak transpiration time (13:40) lags behind the peak water absorption time (10:20). Obviously, 13:40 > 10:20, satisfying the "lag" condition, indicating that the conduction direction is correct but the efficiency is low.

[0128] Simultaneously, the cross-correlation coefficient R between the soil root water uptake rhythm signal and the canopy transpiration rhythm signal was calculated. Pearson correlation analysis of the two time-series curves yielded R=0.45, significantly lower than the preset threshold of 0.70, further confirming the decoupling between the two.

[0129] Based on the above analysis, it was determined that the conduction synergy was lower than the preset synergy threshold, indicating that there was water conduction obstruction in the crop canopy, thus triggering the subsequent obstruction layer location process.

[0130] For example, in a sorghum field with a plant height of 3.2 meters and a preset time deviation tolerance of 70 minutes, monitoring results showed that the peak water absorption time was 9:50, the peak transpiration time was 14:30, ΔT=280 minutes>70 minutes, and transpiration lagged behind water absorption. The cross-correlation coefficient R=0.38<0.70, indicating that the conduction synergy was lower than the preset synergy threshold, and that a water and fertilizer retardation layer existed in the lower part of the canopy.

[0131] In another exemplary scenario, in watermelon drip irrigation cultivation, within the daily monitoring window (5:30 AM - 7:00 PM, a longer monitoring period than other crops due to the long transpiration duration of watermelons in summer), the peak water absorption time point of the soil root water absorption rhythm signal was extracted at 7:50 AM (the peak soil volumetric water content decrease rate was detected as -2.3% / h by a 30cm depth TDR sensor), and the peak transpiration time point of the canopy transpiration rhythm signal was extracted at 10:20 AM (the lowest average leaf temperature of 25.6℃ was detected by an infrared thermal imager, corresponding to a peak transpiration rate of 9.1 mmol / m³). 2 / s).

[0132] The time offset between the peak water absorption time (7:50) and the peak transpiration time (10:20) was calculated to be 150 minutes. Since this time offset is greater than the watermelon-specific preset time offset tolerance of 85 minutes (considering that the watermelon leaf area index (LAI) can reach 4.5-5.2, resulting in high transpiration intensity, the normal conduction delay should be controlled within 85 minutes), and the peak transpiration time (10:20) is significantly later than the peak water absorption time (7:50), it was determined that the conduction synergy is lower than the preset synergy threshold, indicating that water and fertilizer transport is obstructed in the lower part of the watermelon canopy.

[0133] As can be seen from the above, this embodiment accurately extracts the peak water absorption time points of the soil root water absorption rhythm signal and the peak transpiration time points of the canopy transpiration rhythm signal within a preset daily monitoring window, calculates the time offset between the two, and, combined with preset time offset tolerance and conduction direction logic, accurately determines whether the conduction synergy is lower than a preset synergy threshold. This method avoids the one-sidedness of traditional methods that rely solely on water data at a single moment. By revealing the dynamic characteristics of the water conduction process through all-weather rhythm analysis, it provides reliable physiological evidence for the accurate location of the water and fertilizer retardation layer, significantly improving the scientific rigor and accuracy of canopy penetration diagnosis.

[0134] In one embodiment of this application, determining the vertical depth of the water and fertilizer retention layer based on the deep distribution characteristics of soil root water absorption rhythm signals includes:

[0135] S401: Obtain the root water absorption signals corresponding to different soil depth levels.

[0136] In this embodiment, different soil depth levels refer to several vertical soil layers pre-divided based on crop root distribution characteristics and soil profile structure, each with an independent depth range. For field crops, this is typically divided into 4-6 levels, for example: 0-15cm (shallow layer, mainly the area with dense fibrous roots), 15-35cm (medium-shallow layer, the area where secondary roots are distributed), 35-55cm (middle layer, the area where the taproot extends), and 55-80cm (deep layer, the area where deep roots explore water). The stratification needs to be combined with the crop root system architecture and soil texture to ensure that the root functions of each layer are significantly different.

[0137] Layered root water absorption signal refers to the dynamic rate signal of water absorption from soil by roots at each soil depth level. By embedding soil moisture sensors at each level, the rate of change in water content at that level is monitored in real time. Combined with root distribution density, the water absorption rate of the roots at that level is calculated (unit: mm / h or L / (m²)). 3 The signals in each layer are independent of each other, and together they constitute the vertical distribution characteristics of root water absorption.

[0138] In this embodiment, based on the root system survey results of the target cornfield (collected by root drilling at the beginning of the growing season), it was determined that the root system is mainly distributed in the 0-70cm soil layer, and thus divided into four depth levels: 10cm layer (representing 0-20cm), 30cm layer (representing 20-40cm), 50cm layer (representing 40-60cm), and 70cm layer (representing 60-80cm).

[0139] A TDR (Time Domain Reflectometer) soil moisture sensor is buried at the center depth of each layer, and the volumetric water content θ (unit: cm³) of the soil in that layer is recorded every 10 minutes. 3 / cm 3） Within a 24-hour daily monitoring window, the rate of change of water content dθ / dt in each layer was calculated, combined with the root length density Lr of that layer (measured by a root scanner, unit: cm / cm). 3） Calculate the water absorption rate of the stratified root system Q = dθ / dt × Lr.

[0140] For example, in a 10cm layer: root length density Lr = 1.2cm / cm² 3 Between 9:00 and 10:00 AM, the moisture content decreased from 0.28 to 0.26, dθ / dt = -0.02 / h, and the water absorption rate Q = 0.024 cm / h = 0.24 mm / h.

[0141] 30cm layer: Lr=0.8cm / cm 3 From 9:30 to 10:30, the moisture content decreased from 0.25 to 0.235, Q = 0.012 cm / h = 0.12 mm / h.

[0142] 50cm layer: Lr=0.5cm / cm 3 Between 10:00 and 11:00, the moisture content decreased from 0.23 to 0.225, and Q = 0.0025 cm / h = 0.025 mm / h.

[0143] 70cm layer: Lr=0.2cm / cm 3 From 10:30 to 11:30, the moisture content decreased from 0.21 to 0.209, Q = 0.0002 cm / h = 0.002 mm / h.

[0144] The water absorption rates of each layer are arranged by time to generate time-series curves of root water absorption signals, providing data for subsequent response delay analysis.

[0145] S402: Compare the response delay of root water absorption signals in each layer with the transpiration rhythm signal in the upper canopy.

[0146] In this embodiment, response delay refers to the time difference between the peak time of root water absorption at a certain soil depth and the peak time of transpiration at the top of the canopy. The calculation formula is: Response delay = Peak transpiration time - Peak water absorption time of that layer. This indicator reflects the time required for the water absorbed by the roots in that layer to be conducted to the top of the canopy and participate in transpiration, and is a key basis for locating the retardation layer.

[0147] In this embodiment, the peak water absorption time point of each layer is first extracted from the root water absorption signals of each layer. For example, the peak water absorption time point of the 10cm layer is 9:00 (water absorption rate 0.8mm / h), the peak water absorption time point of the 30cm layer is 10:00 (0.6mm / h), the peak water absorption time point of the 50cm layer is 11:00 (0.4mm / h), and the peak water absorption time point of the 70cm layer is 12:00 (0.2mm / h).

[0148] Then, the peak transpiration time point was extracted from the transpiration rhythm signal in the upper canopy, for example, 13:30 (transpiration rate 5.0 mmol·m³). -2 ·s -1） .

[0149] Next, the response delay of each layer is calculated:

[0150] Response time for a 10cm layer = 13:30 - 9:00 = 4.5 hours = 270 minutes;

[0151] Response delay for a 30cm layer = 13:30 - 10:00 = 3.5 hours = 210 minutes;

[0152] Response delay for a 50cm layer = 13:30 - 11:00 = 2.5 hours = 150 minutes;

[0153] Response delay for a 70cm layer = 13:30 - 12:00 = 1.5 hours = 90 minutes;

[0154] Arranging the response delays of each layer by depth: [270min, 210min, 150min, 90min], it was found that the response delay decreased with increasing soil depth, indicating that the water absorbed by the deep roots actually reached the top of the canopy faster. This contradicts the conventional understanding (deeper water conduction distance is longer and the delay should be greater), suggesting that there is conduction blockage between the shallow and middle layers.

[0155] S403: The vertical spatial range between the soil depth layer with the largest response delay and the upper part of the canopy is determined as the spatial range where the water and fertilizer blocking layer is located.

[0156] In this embodiment, the soil depth layer with the largest response delay refers to the layer with the largest time difference between the peak water absorption time and the peak transpiration time among all layered root water absorption signals. The water absorbed by the roots in this layer experiences the longest conduction delay before reaching the top of the canopy, indicating the most severe obstruction in the conduction path between this layer and the canopy.

[0157] The vertical spatial interval refers to the spatial range extending upwards from the canopy height (determined based on the root-canopy correspondence model) corresponding to the soil depth layer with the greatest response delay to the top of the canopy. This interval is the most likely location of the water and fertilizer retardation layer, because a certain layer within this interval hinders the upward transport of water and fertilizer, resulting in a significant increase in response delay.

[0158] In this embodiment, the soil depth layer with the longest response delay was identified as the 10cm layer (response delay 270 minutes). According to the maize root-canopy correspondence model (established through isotope tracing experiments): the 10cm deep root system mainly consists of fibrous roots, responsible for supplying the leaves and stems at the base of the plant (0-0.5 meters); the 30cm deep root system mainly supplies the area at 0.5-1.2 meters; the 50cm deep root system mainly supplies the area at 1.2-2.0 meters; and the 70cm deep root system mainly supplies the area at 2.0-2.8 meters (top of the canopy).

[0159] Since the 10cm root layer corresponds to the 0-0.5m height of the canopy, while the top of the canopy is at 2.8m, the vertical space between the two is 0.5-2.8m. It is determined that there must be a certain layer within this range (most likely in the lower middle part, 0.8-1.5m in height, because the leaves are dense and the stems are thick, which easily forms a physical barrier) that prevents the water absorbed by the 10cm root layer from being conducted upwards to the top, resulting in a response delay of 270 minutes (far exceeding the normal 30-60 minutes).

[0160] In contrast, the response delay of the 70cm deep root system was only 90 minutes, which is higher than the normal value but significantly less than that of the 10cm layer. This indicates that the path of water absorbed by the deep roots to the top (2.0-2.8 meters in height) is relatively smooth, and the blockage mainly occurs in the middle section of the water supply path of the shallow roots.

[0161] Based on the combined response delay data of each layer and the correspondence between the root system and the canopy, the spatial range of the water and fertilizer blocking layer is preliminarily determined to be the canopy height range of 0.8-1.5 meters.

[0162] S404: Determine the vertical depth of the water and fertilizer retention layer based on the spatial range.

[0163] In this embodiment, determining the vertical depth position based on the spatial range means that within the initially locked spatial range where the water and fertilizer blocking layer is located, further combining auxiliary information such as canopy structural characteristics (such as leaf density distribution and stem internode length) and historical disease records (such as local vascular tissue necrosis caused by stem base rot) to accurately locate the center depth and thickness of the blocking layer, and generate a specific vertical depth position description, such as "1.0-1.3 meters from the ground, 0.3 meters thick".

[0164] In this embodiment, after determining the spatial range of 0.8-1.5 meters, the canopy structure database is called to query the typical canopy structure of this maize variety during the tasseling stage: 0.8-1.0 meters in height corresponds to the 8th-10th leaves, with the leaves spread out and a leaf area index (LAI) of 2.5; 1.0-1.3 meters in height corresponds to the 11th-13th leaves (near the ear leaf), with the largest leaves, an LAI of 4.2, and the thickest stem (3.5 cm in diameter) and shortest internodes at this height; 1.3-1.5 meters in height corresponds to the 14th-16th leaves, with an LAI of 3.0.

[0165] Analysis suggests that the LAI (Leaf Intake) in the 1.0-1.3 meter height range is as high as 4.2, with extremely dense leaves and wide leaves at the ear position (leaf width can reach 12cm). The stems are thick but the internodes are short (only 8cm). This structural feature is most likely to cause physical obstruction of water conduction: dense leaves consume a lot of water, but the xylem conduction area of ​​the stem is relatively insufficient, forming a "supply and demand contradiction".

[0166] In addition, a review of historical disease records revealed that the field had experienced mild stem base rot during the jointing stage. Although it had been controlled, pathogen infection could lead to partial blockage of the vascular bundles inside the stems at a height of 1.0-1.2 meters, further exacerbating conduction blockage.

[0167] Based on the above analysis, the vertical depth of the water and fertilizer blocking layer was determined to be: 1.0-1.3 meters above the ground, 0.3 meters thick, and 1.15 meters deep at the center. This depth was mainly caused by the high LAI of the ear leaf layer and local damage to the vascular bundles of the stem.

[0168] For example, in a tomato greenhouse with a plant height of 2.0 meters, the root water absorption signal analysis showed the following response time delays: 180 minutes for the 5cm layer, 150 minutes for the 20cm layer, 120 minutes for the 40cm layer, and 90 minutes for the 60cm layer. The 5cm layer, with the longest response time, corresponds to a canopy height of 0-0.6 meters. This determined the retardation layer to be 0.6-2.0 meters in size. Considering the dense inflorescences of this variety at a height of 0.8-1.1 meters (resulting in high water consumption and limited stem transport), the final vertical depth was determined to be 0.8-1.1 meters, with a thickness of 0.3 meters.

[0169] As can be seen from the above, this embodiment acquires stratified root water absorption signals corresponding to different soil depths, compares the response delays of each layer with the canopy transpiration signals, identifies the soil depth layer with the largest response delay, and accurately determines the vertical depth of the water and fertilizer retention layer based on the root-canopy correspondence model and canopy structural characteristics. This method avoids the crude positioning that relies solely on overall root water absorption data in traditional methods. Through stratified refined analysis, it reveals the differences in the conduction efficiency of root water supply pathways at different depths, providing precise spatial targets for deep-layer targeted water and fertilizer replenishment, and significantly improving the targeting and effectiveness of water and fertilizer regulation.

[0170] In one embodiment of this application, the distribution depth of the temperature heterogeneous region is cross-validated with the vertical depth of the water and fertilizer retardation layer to generate a deficiency confirmation result in the middle and lower layers, including:

[0171] S501: In the canopy thermal infrared image, extract the temperature difference sequence of adjacent pixel rows in the vertical direction, and identify the depth of the pixel row where the temperature change is greater than the preset temperature difference threshold as the distribution depth of the temperature heterogeneous region.

[0172] In this embodiment, adjacent pixel rows refer to two rows of pixels arranged vertically (from bottom to top) in a canopy thermal infrared image. Each row of pixels corresponds to the temperature distribution of a certain horizontal layer of the canopy, and the height difference between two adjacent rows is typically 0.5-1 cm (depending on image resolution and shooting distance).

[0173] The temperature difference sequence is a numerical sequence reflecting the vertical temperature gradient of the canopy, generated by calculating the average temperature difference between adjacent pixel rows. The calculation formula is: ΔT i =T (i+1) -T i T i Let T be the average temperature of the pixels in the i-th row. (i+1) This represents the average temperature of the pixels in the (i+1)th row. This sequence can visually demonstrate the smooth, gradual, or abrupt changes in temperature along the vertical direction of the canopy.

[0174] A temperature abrupt change refers to the temperature difference |ΔT| at a certain position in a temperature difference sequence. iThe temperature fluctuation is significantly greater than the overall background fluctuation level of the sequence, indicating an abnormal temperature jump between the upper and lower layers at this location. This is usually caused by differences in water state (such as dry upper layer and wet lower layer, or wet upper layer and dry lower layer) or differences in physiological activity (such as vigorous photosynthesis in upper layer and senescence in lower layer).

[0175] The preset temperature difference threshold is a critical value pre-set to determine whether a temperature abrupt change is significant, based on the normal vertical gradient characteristics of crop canopy temperature. For closed canopies, the temperature difference between adjacent 5cm heights should normally be ≤1.5℃ (because the microenvironment inside the canopy is relatively uniform); if |ΔT i If the temperature exceeds 2.0℃, it is considered a significant mutation, indicating the presence of moisture or physiological abnormalities at that location.

[0176] Pixel row depth refers to the actual canopy height corresponding to a certain pixel row in a thermal infrared image. Through image geometric correction algorithms, combined with UAV flight altitude, camera field of view, image resolution and crop height, the pixel row coordinates are converted into height above the ground (unit: meters).

[0177] In this embodiment, the acquired canopy thermal infrared image has a resolution of 640×512 pixels, and the shooting range covers the corn canopy with a plant height of 2.8 meters. First, geometric correction is performed: the drone flies at an altitude of 5 meters, the camera's field of view is 45°×36°, and the vertical direction of the image has 512 rows of pixels corresponding to a canopy height of 2.8 meters. Therefore, the height interval of each row of pixels is 2.8 / 512≈0.0055 meters / row.

[0178] Then, pixel temperature data is extracted row by row. For example, the average temperature T of the pixels in row 100 (corresponding to a height of 0.55 meters) is... 100 =29.2℃, line 101 (corresponding height 0.56 meters) T 101 =29.5℃, temperature difference ΔT 100 =0.3℃. Calculations were performed sequentially to generate a temperature difference sequence, resulting in a total of 511 difference values.

[0179] Next, the temperature abrupt change points were searched in the temperature difference sequence. A preset temperature difference threshold of 2.0℃ was set, and the scan revealed that between rows 210 and 211, T... 210 =30.8℃, T 211 =28.2℃, ΔT 210 =-2.6℃, |ΔT 210 The value of 2.6℃ > 2.0℃, indicating a significant mutation. The pixel row depth corresponding to this mutation location is the height of the 210th row, which is approximately 210 × 0.0055 ≈ 1.16 meters.

[0180] The 1.16-meter depth was identified as the distribution depth of the temperature heterogeneous region, indicating that the temperature was abnormally low near the 1.16-meter height of the canopy (30.8℃ in the lower layer and 28.2℃ in the upper layer, cold spots). This suggests that the temperature in this layer may be due to water deficit leading to limited transpiration and abnormally low leaf temperature, or due to light shading leading to decreased photosynthetic activity.

[0181] S502: If the depth deviation between the distribution depth of the temperature heterogeneous region and the vertical depth of the water and fertilizer blocking layer is less than the preset deviation threshold, the cross-validation is deemed successful, and a confirmation result of water and fertilizer deficiency in the middle and lower layers is generated, indicating that there is water and fertilizer deficiency in the middle and lower layers.

[0182] In this embodiment, depth deviation refers to the spatial distance difference between the distribution depth of the temperature heterogeneous region (based on thermal infrared image recognition) and the vertical depth position of the water and fertilizer retardation layer (determined based on hydraulic conduction analysis). The calculation formula is: Depth deviation = |Distribution depth of temperature heterogeneous region - Center depth of water and fertilizer retardation layer|. This index is used to evaluate the consistency of results from two independent diagnostic methods.

[0183] The preset deviation threshold is a critical value used to determine whether two diagnostic results point to the same layer, and is usually set to 0.3 meters (considering the geometric correction error of thermal infrared images and the spatial resolution limitations of the hydraulic conduction model). If the depth deviation is less than this threshold, it indicates that the two methods are highly consistent and cross-validation is successful; if the depth deviation is large, further analysis of the cause is required to avoid misjudgment.

[0184] The results of the deficiency confirmation in the middle and lower layers are cross-validated to generate a formal diagnostic report, which clearly indicates that there is a water and fertilizer deficiency in specific layers in the middle and lower part of the canopy, including the vertical depth range of the deficient layer, the deficiency mechanism (such as hydraulic blockage preventing nutrients from reaching the canopy, insufficient local root activity, etc.), the degree of deficiency, and recommended control measures.

[0185] In this embodiment, the vertical depth of the water and fertilizer blocking layer determined by the aforementioned hydraulic conduction analysis is first retrieved: 1.0-1.3 meters, with a center depth of 1.15 meters. Then, the depth of the temperature heterogeneous region distribution based on thermal infrared image recognition is extracted: 1.16 meters.

[0186] Next, the depth deviation was calculated as |1.16-1.15| = 0.01 meters = 1 cm, which is much smaller than the preset deviation threshold of 0.3 meters. This indicates that the two diagnostic methods are highly consistent, and the cross-validation is successful.

[0187] Based on this, the results confirming the deficiency in the lower and middle layers are generated. An example report is shown below:

[0188] Confirmation report of water and fertilizer deficit in the middle and lower layers:

[0189] Validation status: Cross-validation passed, confirming a deficiency;

[0190] Deficit layer: 1.0-1.3 meters above the ground, corresponding to the 11th-13th leaves (ear-level leaf layer), with a thickness of 0.3 meters.

[0191] Deficiency mechanism: This layer has dense foliage (LAI=4.2) and high transpiration water demand, but the conduction response time from root water absorption (mainly supplied by the root system in the 30-50cm soil layer) to this layer is as high as 210 minutes, far exceeding the normal value of 60 minutes, indicating the presence of hydraulic blockage; at the same time, the leaf temperature in this layer is 2.6℃ lower than that in the layers above and below, showing cold spot characteristics, further confirming that insufficient water supply leads to restricted transpiration;

[0192] Deficiency level: Moderate deficiency (NDVI increase of only 1.4%, lower than the preset 5% threshold; estimated relative leaf water content of about 68%, lower than the normal value of 75%).

[0193] Recommended measures: Implement deep-layer targeted water and fertilizer replenishment. Depth probe to a depth of 50cm and inject 500mL of a nitrogen-containing fertilizer solution at a concentration of 200mg / L each time, with an interval of 2 hours, for a total of 3 times. At the same time, spray 50ppm of ABA to the top of the canopy to reduce transpiration competition from the upper layers and promote the transport of water and fertilizer to the middle and lower layers.

[0194] For example, in a wheat field, the vertical depth of the retardation layer was determined to be 0.6-0.9 meters (0.75 meters at the center) based on hydraulic conduction analysis. The depth of the temperature heterogeneous region was identified to be 0.78 meters based on thermal infrared image identification. The depth deviation was 0.03 meters (<0.3 meters). The cross-validation was successful, and a confirmation report was generated, confirming that there was a water and fertilizer deficit in this layer and recommending deep recharge.

[0195] As can be seen from the above, this embodiment extracts the temperature difference sequence in the vertical direction of the canopy thermal infrared image to identify the distribution depth of the temperature heterogeneous region corresponding to the location of temperature abrupt changes. This is then cross-validated with the vertical depth location of the water and fertilizer retardation layer obtained based on hydraulic conduction analysis. When the depth deviation is less than a preset threshold, water and fertilizer deficit in the middle and lower layers is confirmed. This method integrates two independent chains of evidence: hydraulic conduction analysis (reflecting physiological processes) and thermal characterization analysis (reflecting temperature anomalies). This significantly improves the reliability and persuasiveness of the diagnostic results, avoids potential misjudgments from single methods, and provides a solid decision-making basis for subsequent precise regulation.

[0196] In one embodiment of this application, it further includes:

[0197] S601: If the depth deviation between the distribution depth of the temperature heterogeneous region and the vertical depth of the water and fertilizer retardation layer is greater than or equal to the preset deviation threshold, then the hydraulic retardation and thermal characterization are decoupled.

[0198] In this embodiment, the decoupling of hydraulic retardation and thermal characterization refers to the spatial inconsistency (depth deviation ≥ 0.3 meters) between the location of the water and fertilizer retardation layer identified based on hydraulic conduction analysis and the location of the temperature heterogeneous region identified based on thermal infrared images. This indicates that the two diagnostic methods point to different layers. This decoupling phenomenon is usually caused by the following reasons: the hydraulic retardation layer exists but does not cause significant temperature anomalies (e.g., the degree of retardation is mild, and the leaves maintain temperature by regulating stomata), or the temperature anomaly is caused by non-water factors (e.g., pests and diseases, shading), rather than hydraulic retardation.

[0199] In this embodiment, when a depth deviation ≥ a preset deviation threshold is detected, the system does not directly generate a confirmation result for the deficiency in the middle and lower layers. Instead, it determines that the hydraulic hindrance and thermal characterization are decoupled and initiates a further analysis process to avoid misdiagnosis.

[0200] For example, in a cornfield, hydraulic conduction analysis determined the vertical depth of the stagnant layer to be 1.0-1.3 meters (1.15 meters at the center). However, thermal infrared imagery identified the depth of the temperature heterogeneous region at 0.65 meters. The depth deviation |0.65-1.15|=0.5 meters ≥ 0.3 meters, indicating decoupling. Further analysis revealed that at the 0.65-meter height, there was localized warming (hot spots) caused by corn borer larvae boring into the stalks, rather than cold spots caused by water deficit. Therefore, the temperature anomaly was unrelated to hydraulic stagnant layer.

[0201] S602: In response to the decoupling of hydraulic hindrance and thermal characterization, it generates a mid-to-lower layer deficiency early warning result based solely on hydraulic hindrance, and reduces the preset replenishment level for subsequent deep-layer targeted water and fertilizer replenishment.

[0202] In this embodiment, the mid-to-lower layer deficit early warning result based solely on hydraulic hindrance refers to a preliminary early warning report generated only based on the results of hydraulic conduction analysis (such as abnormal response delay or low conduction synergy) when cross-validation fails, rather than a confirmatory report. The reliability of this early warning result is lower than that of the confirmatory result; therefore, a conservative control strategy is recommended to avoid excessive intervention.

[0203] Reducing the preset replenishment level means that when performing deep-layer targeted water and fertilizer replenishment, the original planned water and fertilizer injection volume (such as 500 mL each time, nitrogen concentration 200 mg / L) is reduced proportionally (such as reduced to 300 mL each time, nitrogen concentration 150 mg / L) to reduce the risk of over-fertilization due to misdiagnosis, while still being able to replenish the suspected stagnant layer appropriately.

[0204] In this embodiment, after determining that the hydraulic hindrance and thermal characterization are decoupled, an example of generating an early warning report is as follows:

[0205] Early warning report of water and fertilizer deficiency in the middle and lower layers (unconfirmed):

[0206] Warning status: Cross-validation failed; the warning is based solely on hydraulic conduction analysis and may be incomplete.

[0207] Suspected deficient layers: 1.0-1.3 meters above the ground, response delay 210 minutes, conduction synergy R=0.45, lower than normal level;

[0208] Thermal verification results: The temperature anomaly region is located at a height of 0.65 meters, which deviates from the suspected missing layer by 0.5 meters, and they cannot corroborate each other;

[0209] Possible causes: The hydraulic obstruction was minor and did not cause significant temperature anomalies; or the temperature anomalies were caused by non-moisture factors such as pests and diseases.

[0210] Regulation recommendations: Adopt a conservative strategy, reduce the amount of deep-layer targeted water and fertilizer replenishment to 60% of the original plan, probe down to a depth of 50cm, inject 300mL of water and fertilizer solution with a nitrogen concentration of 150mg / L each time, with an interval of 2 hours, for 2 consecutive times; at the same time, strengthen field inspections to check for non-water factors such as diseases and pests, and adjust the strategy after further confirmation.

[0211] At the same time, the control parameters for deep-seated targeted water and fertilizer replenishment are automatically adjusted: the original plan of injecting 500 mL (100 mg of nitrogen) each time is reduced to 300 mL (45 mg of nitrogen) each time, a reduction of 55%; the original plan of continuous replenishment of 3 times is adjusted to 2 times, and the total nitrogen replenishment is reduced from 300 mg to 90 mg, which is about 30% of the original plan.

[0212] For example, in a tomato greenhouse, hydraulic analysis showed stagnation at a depth of 0.8-1.1 meters, but thermal infrared imaging revealed a temperature heterogeneity at 1.5 meters (depth deviation ≥ 0.3 meters, decoupled). An early warning was generated, reducing the fertilization amount from the planned 400 mL to 250 mL per application to prevent excessive vegetative growth due to misdiagnosis and over-fertilization.

[0213] As can be seen from the above, this embodiment determines whether hydraulic hindrance and thermal characterization are decoupled by setting a depth deviation threshold. When decoupled, it does not blindly confirm deficiency, but instead generates an early warning result based solely on hydraulic hindrance and reduces the subsequent recharge amount, adopting a conservative strategy. This mechanism effectively avoids the risk of over-intervention caused by misjudgment due to a single diagnostic method, embodying the precision agriculture principle of "better to recharge less than to recharge excessively." While ensuring the safety of regulation, it can still provide appropriate recharge to suspected deficient layers, reserving space for further confirmation and strategy adjustment.

[0214] In one embodiment of this application, deep-directed water and fertilizer replenishment to a vertical depth location includes:

[0215] S701: Based on the vertical depth position of the water and fertilizer blocking layer, control the fertilizer injection probe of the deep fertilizer injection actuator to descend to the vertical depth position.

[0216] In this embodiment, the deep fertilization actuator is an automated fertilization device, mainly composed of an electrically or hydraulically driven probe lowering mechanism, a water-fertilizer mixing and injection pump, a control unit, and a positioning sensor. This device can precisely lower the fertilization probe to a designated soil depth according to system instructions, and inject a pre-mixed water-fertilizer solution at that depth, achieving vertical, layered, and directional delivery of nutrients.

[0217] The fertilizer injection probe is the core component of the deep fertilizer injection actuator, similar to a soil drilling device. It is usually a stainless steel tube with a diameter of 2-3cm. The front end is tapered to facilitate penetrating the soil, and there are several liquid outlet holes (diameter of 2-3mm) distributed on the side wall. The probe can penetrate vertically to a depth of 80cm (for field crops), with a penetration accuracy of ±2cm.

[0218] The vertical depth refers to the soil depth corresponding to the water and fertilizer retention layer. The target fertilization depth (e.g., 50cm soil layer) is derived by the system based on the height of the canopy retention layer (e.g., 1.0-1.3 meters) and the root-soil depth correspondence model. This depth is the target position that the fertilization probe needs to reach to ensure that the injected water and fertilizer can directly reach the active root distribution area below the retention layer.

[0219] In this embodiment, based on the aforementioned determined vertical depth position of the water and fertilizer barrier layer (canopy height 1.0-1.3 meters, corresponding to soil depth 40-60 cm, center depth 50 cm), a probe insertion command is generated and sent to the deep fertilization actuator in the field via the Internet of Things communication module.

[0220] After receiving the instruction, the deep fertilization actuator initiates the following execution process:

[0221] First, the positioning sensor (GPS+RTK differential positioning, accuracy ±5cm) confirms the current position of the device and moves it to the target row (e.g., 10cm to the side of the 8th plant in the 15th row of corn to avoid directly damaging the roots).

[0222] Secondly, the electric drive mechanism starts, driving the fertilizer injection probe vertically downward to penetrate the soil. The conical structure at the tip of the probe compresses the soil to form a channel, and the depth sensor (encoder) provides real-time feedback on the penetration depth. When it reaches 50cm, the drive mechanism automatically stops, and the probe is fixed at the target depth.

[0223] For example, in a cornfield, the system instructs the probe to descend to a depth of 50cm. After the device locates the target plant, the probe descends at a constant speed of 10cm / s. The depth sensor provides feedback: 10cm→20cm→30cm→40cm→50cm, stopping once the target depth is reached. The entire descent process takes 5 seconds, with a positioning error of ±1.5cm, meeting the accuracy requirements.

[0224] S702: Obtain the soil texture and water holding characteristics at the vertical depth position, and determine the duration and interval of the fertilization pulse based on the soil texture and water holding characteristics, so that the injected water and fertilizer form a lateral infiltration front at the vertical depth position and inhibit the longitudinal migration of water and fertilizer to the shallow root distribution area.

[0225] In this embodiment, soil texture water-holding characteristics refer to the physical water-holding capacity of the soil layer at the target depth, which is mainly determined by soil texture (sand, loam, clay) and pore structure, including saturated water content θ. s Field water holding capacity θ fc and wilting coefficient θ wp Parameters such as water retention capacity and water transport rate vary significantly among soils of different textures: sandy soils have large pores, weak water retention capacity, and rapid water infiltration; clay soils have small pores, strong water retention capacity, and slow lateral water diffusion.

[0226] A fertilizer injection pulse refers to a single operation by which a deep fertilizer injection actuator injects a water-fertilizer solution into a target depth. It is performed in an intermittent manner of "short injection-stop-re-injection" rather than continuous injection. The volume of water and fertilizer injected each time is called the single pulse dose (e.g., 200 mL), the injection duration is called the pulse duration (e.g., 15 seconds), and the stop time between two pulses is called the pulse interval period (e.g., 1 hour).

[0227] A lateral infiltration front refers to the boundary of a moistened zone formed by the horizontal diffusion of a water-fertilizer solution injected into the soil at the target depth, centered on the injection point. An ideal lateral infiltration front is disc-shaped with a radius of 20-40 cm (depending on soil texture and injection volume), covering most of the root system at that depth to ensure that nutrients can be fully absorbed by the roots.

[0228] Vertical migration refers to the phenomenon where water and fertilizer solutions injected into deep soil layers, under the combined influence of gravity and the transpiration pull of upper-layer roots, move upwards along soil pores to the shallow root distribution area, where they are preferentially absorbed by the shallow roots, resulting in the failure of deep recharge to reach the target area. Suppressing vertical migration is key to deep-layer targeted recharge, which requires optimizing fertilization pulse parameters to ensure that water and fertilizer primarily diffuse laterally rather than move upwards vertically.

[0229] In this embodiment, the soil texture and water-holding characteristics at a vertical depth (50cm soil layer) are first obtained. By querying a field soil database (established based on previous soil profile surveys) or by real-time access to a soil texture sensor (such as a TDR combined with a soil compaction sensor) buried at a depth of 50cm, the soil parameters for this layer are obtained: texture is loam, saturated water content θ. s =0.45cm 3 / cm 3 Field water holding capacity θ fc =0.32cm 3 / cm 3The current water content θcurrent = 0.26 cm³ 3 / cm 3（ Slightly below field capacity, indicating room for water replenishment; soil porosity n=0.48, hydraulic conductivity K=0.8cm / h (moderate).

[0230] Then, based on the soil texture and water retention characteristics, the fertilizer injection pulse parameters are calculated:

[0231] (1) Single pulse dose calculation:

[0232] The goal is to increase the soil moisture content at a depth of 50 cm from 0.26 to approximately 0.32, close to field capacity, by an increase of Δθ = 0.06. Assuming a lateral infiltration front radius R = 30 cm and a thickness (vertical direction) h = 10 cm (the distribution range of the probe's pores), the infiltration volume V = π × R. 2 ×h=π×30 2 ×10≈28,300cm 3 The required water volume = V × Δθ = 28,300 × 0.06 ≈ 1,700 mL. Considering the concentration of the fertilizer solution (nitrogen 200 mg / L, close to pure water), the single-pulse dose is set at 200 mL, requiring 9 pulses to reach the target water content. However, to avoid excessive injection at one time leading to vertical migration, a multi-pulse injection strategy is adopted.

[0233] (2) Calculation of pulse duration:

[0234] The fertilizer injection pump flow rate is set to 800 mL / min. The time required to inject 200 mL is 200 / 800 = 0.25 minutes = 15 seconds. Therefore, the pulse duration is set to 15 seconds, which ensures accurate injection dosage while avoiding local saturation caused by excessive injection time.

[0235] (3) Calculation of pulse interval period:

[0236] The key is to ensure that each injection of water and fertilizer has fully diffused laterally and has not migrated upwards before the next injection. Based on the soil's hydraulic conductivity K = 0.8 cm / h, the lateral diffusion rate of water is approximately K / 2 = 0.4 cm / h (lateral diffusion is slower than vertical infiltration). To achieve a lateral diffusion of 10 cm (from the injection point to the surrounding root system), the time required is 10 / 0.4 = 25 hours. However, considering that root absorption accelerates water consumption, the actual diffusion of 10 cm only takes about 2 hours. Simultaneously, to inhibit vertical migration, it is necessary to ensure that the transpiration pull of the upper root system has been weakened by the transpiration inhibitor (effective 30 minutes after ABA application, lasting 4-6 hours). Taking all factors into consideration, a pulse interval of 2 hours is set, allowing for sufficient lateral diffusion of water and fertilizer while ensuring replenishment within the effective period of the transpiration inhibitor.

[0237] Finally, the fertilization pulse parameters were determined as follows: single pulse dose 200mL (nitrogen concentration 200mg / L, containing 40mg nitrogen), pulse duration 15 seconds, pulse interval 2 hours, continuous replenishment 3 times, totaling 600mL of water-fertilizer solution, containing 120mg nitrogen, approximately 0.12g / plant.

[0238] For example, the first pulse is executed: the fertilizer pump injects 200 mL of fertigation solution into a depth of 50 cm within 15 seconds. The solution sprays out from the outlet hole on the side wall of the probe, seeping laterally into the soil pores to form an initial infiltration front. After the injection stops, the soil moisture content at this depth is monitored, rising from 0.26 to 0.28, an increase of 2 percentage points. After waiting for 2 hours, the moisture content remains at 0.27 (only a decrease of 1 percentage point, indicating that the water is mainly absorbed by the roots or diffused laterally, with little vertical migration). A second pulse is executed, and the moisture content rises again to 0.29. After waiting another 2 hours, a third pulse is executed, and the moisture content finally reaches 0.31, close to the field capacity of 0.32, with the lateral infiltration front radius reaching 28 cm, covering more than 80% of the root distribution area at this depth.

[0239] Meanwhile, the soil moisture content in the 30cm shallow layer was monitored, and it only increased from 0.24 to 0.25 during the entire recharge process (an increase of 1 percentage point, which is far less than the 5 percentage point increase at the 50cm depth), indicating that the vertical migration was effectively suppressed and the water and fertilizer injected from the deep layer mainly stayed in the target depth layer and were not taken away by the shallow root system.

[0240] As can be seen from the above, this embodiment, by controlling the fertilization probe of the deep fertilization actuator to accurately probe to the vertical depth corresponding to the water and fertilizer retention layer, and optimizing the duration and interval of the fertilization pulse based on the soil texture and water-holding characteristics at this depth, achieves lateral directional diffusion of water and fertilizer at the target depth layer, effectively suppressing vertical migration back to the shallow layer. This method breaks through the technical bottleneck of traditional surface application or drip irrigation fertilization, where "upper-layer roots preferentially absorb nutrients, while middle and lower-layer roots have difficulty obtaining them," ensuring that nutrients supplied from deep layers can accurately reach the root distribution area below the retention layer. This significantly improves the balance and utilization efficiency of fertilizer distribution in the vertical direction of the canopy, providing an innovative technical means to solve the problem of nutrient deficiency in the middle and lower layers of closed-canopy crops.

[0241] In one embodiment of this application, transpiration regulation is performed on the upper canopy to reduce the competitive advantage of upper water and fertilizer, including:

[0242] S801: Before implementing deep-layer targeted water and fertilizer replenishment, spray an transpiration inhibitor onto the upper canopy.

[0243] In this embodiment, transpiration inhibitors are chemical agents that can temporarily reduce stomatal conductance and slow down transpiration rates in plant leaves. These include plant hormones (such as abscisic acid (ABA) and its analogues), film-forming agents (such as high molecular weight polymers that form a breathable and semi-permeable membrane on the leaf surface), and metabolic regulators (such as salicylic acid and chitosan). This embodiment preferably uses ABA analogues (such as S-ABA) because of their clear mechanism of action (promoting stomatal closure), rapid onset of action (20-30 minutes after spraying), moderate duration of action (4-6 hours), non-toxic side effects on crops, and the ability of stomata to recover naturally after the effective period, without affecting subsequent photosynthesis.

[0244] The upper canopy refers to the leaf layer located in the top third of the crop canopy's vertical structure. This part of the leaves receives ample sunlight and has high transpiration rates, making it the primary source of transpiration pull. Spraying this part with transpiration inhibitors can effectively weaken its "siphoning effect" on water, reducing the water and fertilizer competition advantage of the upper part.

[0245] The timing of spraying before implementing deep-layer targeted water and fertilizer replenishment is crucial: if replenishment is done before spraying, the water and fertilizer injected into the deep layer will be immediately drawn away by the strong transpiration pull of the upper root system, greatly reducing the replenishment effect; if spraying is done in advance, and replenishment is done after the transpiration inhibitor takes effect and the transpiration pull of the upper layer weakens, the deep water and fertilizer will have more time to diffuse laterally in the target depth layer and be absorbed by the middle and lower root system.

[0246] In this embodiment, after generating the water and fertilizer decoupling control command, the transpiration inhibitor spraying operation is executed first.

[0247] For example, in a cornfield, the instruction is to spray S-ABA at 50ppm onto the upper canopy at 9:00 AM. The agricultural drone hovers 3.5 meters above the field, with the nozzle tilted downwards at 30°, aiming at the 1.9-2.8 meter height layer, flying at a constant speed of 1.5 m / s, atomizing and spraying. The spraying task of 1 acre is completed in 10 minutes, with even adhesion to the leaf surface and no obvious loss.

[0248] S802: Monitor the decrease in the transpiration rhythm signal in the upper canopy during a preset delay period after spraying the transpiration inhibitor.

[0249] In this embodiment, the preset delay time period refers to the time window from the completion of spraying the transpiration inhibitor to the expected onset and stable inhibitory effect of the transpiration inhibitor. Based on the pharmacological characteristics of S-ABA, it is usually set to 30-60 minutes. During this time period, the transpiration rhythm signal in the upper canopy needs to be continuously monitored to verify whether the transpiration inhibitor takes effect as expected.

[0250] The decrease refers to the percentage reduction in the transpiration rate of the upper canopy relative to the baseline value before spraying after the application of a transpiration inhibitor. The calculation formula is: Decrease = (Transpiration rate before spraying - Transpiration rate after spraying) / Transpiration rate before spraying × 100%. This indicator directly reflects the actual inhibitory effect of the transpiration inhibitor.

[0251] In this embodiment, after spraying is completed at 9:00 a.m., continuous monitoring of the transpiration rhythm signal of the upper canopy is immediately started. The time window is 9:00-10:00 a.m. (preset delay time period of 60 minutes). The sampling frequency is once every 5 minutes, with a total of 12 data points.

[0252] Before spraying (9:00 AM), the baseline transpiration rate in the upper canopy was 4.8 mmol·m³. -2 ·s -1 (At this time, the air temperature is 28℃, the relative humidity is 55%, and the photosynthetically active radiation PAR = 1200 μmol·m⁻¹) -2 ·s -1 ).

[0253] The changes in transpiration rate after spraying are as follows:

[0254] At 9:05, the transpiration rate was 4.7 mmol·m³. -2 ·s -1 The decrease was 2.1% (the medicine had just adhered and had not yet been absorbed).

[0255] At 9:15, the transpiration rate was 4.3 mmol·m³. -2 ·s -1 The decrease was 10.4% (ABA began to be absorbed by the leaves, and stomatal conductance began to decrease).

[0256] At 9:25, the transpiration rate was 3.8 mmol·m³. -2 ·s -1 The decrease was 20.8% (accelerated stomatal closure);

[0257] At 9:35, the transpiration rate was 3.3 mmol·m³. -2 ·s -1 The decrease was 31.3% (close to a stable inhibitory effect).

[0258] At 9:45, the transpiration rate was 3.2 mmol·m³. -2 ·s -1 The decline was 33.3% (reaching a stable level);

[0259] At 10:00, the transpiration rate was 3.1 mmol·m³. -2 ·s -1 The decline was 35.4% (remaining stable);

[0260] The decrease was recorded at 35.4% 60 minutes after spraying (10:00), indicating that the transpiration inhibitor had taken full effect.

[0261] S803: In response to the decrease reaching a preset inhibition ratio, the deep-layer targeted water and fertilizer recharge is triggered to allow the deep-layer water and fertilizer recharge to move to the middle and lower layers under the condition that the transpiration pull of the upper canopy is weakened.

[0262] In this embodiment, the preset inhibition ratio is based on the theoretical model of water and fertilizer decoupling regulation and field test data. It is the minimum inhibition effect threshold that the transpiration inhibitor must achieve, which is usually set at 25%-30%. When the decrease reaches this threshold, it indicates that the transpiration pull of the upper canopy has been significantly weakened. At this time, deep water and fertilizer replenishment can ensure that the injected water and fertilizer are mainly transported to the middle and lower layers, rather than being quickly drawn away by the upper root system.

[0263] The triggering action for deep-layer targeted water and fertilizer replenishment refers to the system automatically sending a start command to the deep-layer fertilizer injection actuator after detecting that the evaporation rate has reached the target, and starting to execute the aforementioned probe probing, fertilizer injection pulse, and other operations to achieve precise timing coordination between evaporation inhibition and deep-layer replenishment.

[0264] In this embodiment, the preset inhibition ratio is set to 30%. After monitoring that the decrease reaches 33.3% > 30% at 9:45, it is determined that the transpiration inhibitor has taken full effect, and the deep-seated targeted water and fertilizer replenishment action is immediately triggered.

[0265] The specific process is as follows:

[0266] First, through the Internet of Things communication module, a start command is sent to the deep fertilization actuator in the field: "Probe descends to a depth of 50cm → execute fertilization pulse (200mL / time, 2 hours apart, 3 times in a row)".

[0267] Second, the deep fertilization actuator began its first pulse at 9:50, injecting 200 mL of a water-fertilizer solution (nitrogen concentration 200 mg / L, containing 40 mg of nitrogen) to a depth of 50 cm. At this time, the transpiration rate in the upper canopy had decreased to 3.2 mmol·m³. -2 ·s -1 Compared to 4.8 mmol·m before spraying -2 ·s -1 The water content decreased by 33.3%, indicating a significant weakening of the "siphoning pull" of the upper root system on water.

[0268] Third, the injected water and fertilizer solution diffused laterally in the 50cm depth layer. The soil moisture content at this depth increased from 0.26 to 0.28, while the moisture content in the 30cm shallow layer only increased slightly from 0.24 to 0.245 (the increase was only 1 / 4 of that at the 50cm depth). This indicates that the deep water and fertilizer solution mainly remained in the target depth layer, and the vertical migration to the shallow layer was effectively suppressed.

[0269] Fourth, a second pulse was administered 2 hours later (11:50); a third pulse was administered 2 hours later (13:50). Throughout the resupply process, the transpiration rate in the upper canopy remained between 3.0 and 3.3 mmol / m³. -2 ·s -1 The low position (transpiration inhibitors continue to act for 4-6 hours) ensures that deep water and fertilizer can be fully transported to the middle and lower layers and absorbed by the middle and lower root system.

[0270] For example, in wheat fields, after spraying with 50 ppm of ABA, the transpiration rate decreased from 5.5 mmol·m³. -2 ·s -1 Reduced to 3.8 mmol·m -2 ·s -1 The decrease was 30.9%, reaching the preset inhibition ratio of 30%, triggering deep replenishment. Water and fertilizer were injected to a depth of 40cm. Monitoring showed that the NDVI of the middle and lower leaves (0.6-0.9 meters in height) increased from 0.68 to 0.75 within 48 hours, an increase of 10.3%, which was significantly higher than the control without transpiration inhibition (an increase of only 3.1%).

[0271] As can be seen from the above, this embodiment achieves precise temporal coordination between transpiration regulation and deep fertilization by spraying transpiration inhibitors onto the upper canopy before implementing deep-layer targeted water and fertilizer resupply and monitoring the decline in transpiration rhythm signals within a preset delay period. When the decline reaches a preset inhibition ratio, the deep fertilization action is triggered. This method effectively weakens the water and fertilizer competition advantage of the upper canopy, ensuring that the deeply injected water and fertilizer can be fully transported to the middle and lower layers and absorbed by the root system of the target layer under the condition of reduced transpiration pull. It avoids the technical problem of "deep injection, upper layer grabbing" in traditional deep fertilization, significantly improves the actual effect of water and fertilizer decoupling regulation, and provides key supporting technology for solving the problem of uneven vertical nutrient distribution in closed canopy crops.

[0272] In one embodiment of this application, before triggering the coronal penetration diagnostic mode, the method further includes:

[0273] When the crop canopy closure is in the transitional range of partial canopy closure, the nutrient diagnosis channel based on the top spectrum and the auxiliary diagnosis channel based on soil-plant hydrothermal conduction are maintained simultaneously.

[0274] Based on the increasing trend of canopy closure, the decision weight of the nutrient diagnosis channel based on the top spectrum is gradually reduced, while the decision weight of the auxiliary diagnosis channel is increased simultaneously, until the canopy penetration diagnosis mode is determined.

[0275] In existing technologies, a single diagnostic mode is usually adopted, and after the spectral diagnosis fails, it directly switches to other diagnostic methods. The lack of a dual-channel collaborative mechanism during the transition period can easily lead to a break in diagnostic accuracy during the switching process, affecting the continuity of water and fertilizer decision-making.

[0276] In this embodiment, canopy closure is a quantitative indicator characterizing the degree to which crop canopy covers the ground. It is defined as the percentage of the vertical projection area of ​​the canopy to the total ground area, with a value range of 0-100%. It can be calculated by acquiring a top-down image of the crop canopy using a high-resolution camera installed in an agricultural IoT system, and then extracting the proportion of green vegetation pixels using an image segmentation algorithm. The partial canopy closure transition zone is a specific stage in which canopy closure transitions from sparse growth to complete closure. In this embodiment, it is specifically defined as the range where the canopy closure is between 60% and 85%. At this stage, the crop canopy is not yet completely closed, and the spectrum of the top layer can still partially reflect the nutrient status of the plant. However, as the canopy closure increases, the ability of the spectral signal to characterize the nutrient status of the middle and lower layers gradually weakens, requiring the introduction of auxiliary diagnostic methods.

[0277] Nutrient diagnostic channels based on top-level spectra utilize multispectral or hyperspectral sensors to collect the reflectance spectrum of the top of the crop canopy. By analyzing the reflectance of specific bands (such as the red band at 650nm and the near-infrared band at 850nm) and their combined indices (such as the Normalized Difference Vegetation Index (NDVI) and the Chlorophyll Index (CI), the channel diagnoses the abundance or deficiency of nutrients such as nitrogen and phosphorus in crops. This channel has a fast response speed and non-contact acquisition, but its ability to obtain nutrient information in the middle and lower layers after the canopy has closed is limited.

[0278] The auxiliary diagnostic channel based on soil-plant hydrothermal conduction is a technical approach that uses moisture sensors (such as time-domain reflectometers) deployed at different soil depths to collect root water absorption rhythm signals, combined with infrared thermal imagers installed on the upper part of the canopy to monitor transpiration rhythm signals, and analyzes the water conduction process from soil to roots to stems to leaves to indirectly diagnose water and fertilizer transport blockages. This channel can penetrate the canopy to obtain deep information, but the data processing complexity is high and it needs to be analyzed in conjunction with a hydraulic model.

[0279] Decision weight is the degree of influence of different diagnostic channels on the final control instructions in water and fertilizer management decision-making, expressed as a percentage. The sum of the weights of all channels is always 100%. The higher the weight, the greater the proportion of the diagnostic result of that channel in the decision-making.

[0280] In this embodiment, when the crop canopy closure is detected to have entered the partial canopy closure transition zone (60%-85%), the agricultural IoT system no longer relies solely on the nutrient diagnosis channel based on the top-layer spectrum. Instead, it simultaneously activates the auxiliary diagnosis channel based on soil-plant hydrothermal conduction, forming a dual-channel collaborative working mode. Specifically, the nutrient diagnosis channel based on the top-layer spectrum continues to acquire multispectral images of the top of the canopy in real time, calculating indices such as NDVI and CI to assess the nutrient status of the top layer. Simultaneously, the auxiliary diagnosis channel based on soil-plant hydrothermal conduction simultaneously acquires root water absorption signals at different soil depths (e.g., 10cm, 20cm, 40cm) and transpiration rhythm signals from the upper part of the canopy, calculating the time offset between the two to assess whether water and fertilizer conduction is smooth.

[0281] During the dual-channel collaborative operation, the agricultural IoT system continuously monitors the dynamic changes in canopy closure and extracts its increasing trend. This increasing trend can be quantified by the daily canopy closure increment (e.g., from 65% to 68%, a daily increment of 3%) or the slope of a linear fit over seven consecutive days (e.g., a slope of 2.5% / day). Based on this increasing trend, a dynamic adjustment strategy for decision weights is implemented.

[0282] First, a mapping relationship between canopy closure and decision weights is established. For example, when the canopy closure is 60%, the initial decision weight of the nutrient diagnosis channel based on the top-level spectrum is set to 70%, and the initial decision weight of the auxiliary diagnosis channel based on soil-plant hydrothermal conduction is set to 30%. As the canopy closure increases to 70%, the weight of the top-level spectrum channel decreases to 50%, while the weight of the auxiliary diagnosis channel increases to 50%. When the canopy closure reaches 85%, the weight of the top-level spectrum channel decreases to 20%, while the weight of the auxiliary diagnosis channel increases to 80%.

[0283] Secondly, linear interpolation or nonlinear functions (such as S-curves) are used to achieve a smooth transition of weights, avoiding decision fluctuations caused by abrupt changes in weights.

[0284] For example: using the Sigmoid function W 光谱 (C)=100 / (1+e ((C-72.5) / 5) ), where C is the current canopy closure, W 光谱 The top-level spectral channel weight percentage, and the auxiliary diagnostic channel weight W. 辅助 =100-W 光谱 This function ensures that the weights switch quickly around a canopy closure of 72.5%, but maintains a smooth transition at both ends of the interval.

[0285] Finally, when the canopy closure reaches the upper limit (85%) of the partial canopy closure transition zone and remains stable or continues to increase for 3 consecutive days, it is determined that the crop canopy has entered a high canopy closure state. At this time, the decision weight of the nutrient diagnosis channel based on the top spectrum has dropped to below the preset threshold (e.g., 20%), and the system automatically triggers a complete switch to the canopy penetration diagnosis mode. That is, the nutrient diagnosis channel based on the top spectrum is suspended, and the decision weight is allocated 100% to the diagnosis channel based on soil-plant hydrothermal conduction (at this time, it is officially named the canopy penetration diagnosis mode) to ensure that the diagnosis method is accurately matched with the crop growth stage.

[0286] The key reasons for this design are: to avoid diagnostic gaps, during the dual-channel collaborative operation, the two diagnostic results mutually verify each other, the spectral channel provides a snapshot of top-level nutrients, and the auxiliary channel provides information on deep water and fertilizer transport. The fusion decision reduces the risk of misjudgment by a single channel, ensuring that the accuracy of water and fertilizer regulation is not affected during the transition period; to smooth the weight transition, by dynamically adjusting the decision weights rather than directly switching the diagnostic mode, it can gradually adapt to changes in canopy structure, avoiding drastic fluctuations in water and fertilizer management strategies due to sudden mode changes, and ensuring the stability of the crop growth environment; and to accurately trigger the timing, using 85% canopy closure and three consecutive stable days as dual judgment conditions, ensuring that the spectral diagnosis has failed (excessive canopy closure completely blocks information in the middle and lower layers), and avoiding false triggers caused by short-term fluctuations in canopy closure (such as local leaf drop), thus improving the reliability of mode switching.

[0287] For example, in a maize planting area during the late jointing stage, the agricultural IoT system monitored that the crop canopy closure increased from 58% to 63%, entering the partial canopy closure transition range (60%-85%). The dual-channel synergistic mode was immediately activated: the nutrient diagnosis channel based on the top-level spectrum collected the canopy reflectance spectrum, calculating NDVI=0.72, corresponding to a nutrient status of "moderate to high"; simultaneously, the auxiliary diagnosis channel based on soil-plant hydrothermal conduction collected data showing that the root water uptake peak occurred at 9:15 AM, and the canopy transpiration peak occurred at 10:45 AM, with a time offset of 90 minutes, corresponding to a conduction synergy of "good". At this point, based on a canopy closure of 63%, the decision weight of the top-level spectrum channel was calculated to be 65% using the Sigmoid function, and the auxiliary diagnosis channel weight was 35%. The results from both channels were then combined to generate a water and fertilizer regulation recommendation: "Maintain the current irrigation frequency and apply an appropriate amount of nitrogen fertilizer."

[0288] Over the next 7 days, canopy closure increased at a rate of 2.8% per day, reaching 82.6% on the 7th day. Daily dynamic weight adjustments were made: on the 3rd day when canopy closure was 71%, the spectral weight decreased to 52%, and the auxiliary weight increased to 48%; on the 5th day when canopy closure was 77%, the spectral weight decreased to 35%, and the auxiliary weight increased to 65%; on the 7th day when canopy closure was 82.6%, the spectral weight decreased to 23%, and the auxiliary weight increased to 77%. During this stage, the spectral channel detected a slow decrease in NDVI to 0.68, but the auxiliary channel showed that the time shift between root water uptake and canopy transpiration increased to 120 minutes, and the conduction synergy decreased. Based on the 77% auxiliary weight, a potential risk of water and fertilizer retention was identified, and the irrigation strategy was adjusted in advance to "increase irrigation volume by 20% and extend irrigation interval by 12 hours."

[0289] On day 8, the canopy closure reached 85.2% and remained above 85% for three consecutive days, indicating a high canopy closure state. This officially triggered the canopy penetration diagnostic mode: the nutrient diagnosis channel based on the top-layer spectrum was completely suspended, and the decision weight was transferred 100% to the diagnosis channel based on soil-plant hydrothermal conduction. Thereafter, water and fertilizer diagnosis was performed solely based on root water uptake rhythm, canopy transpiration rhythm, and canopy thermal infrared images. This successfully identified a water and fertilizer retardation layer at a depth of 40cm in the lower half of the canopy, enabling deep-layer targeted water and fertilizer replenishment and avoiding misjudgments caused by continued reliance on ineffective spectral signals.

[0290] Throughout the transition period, the dual-channel collaborative operation ensured that the diagnostic accuracy did not decrease due to changes in the canopy structure. Dynamic weight adjustment enabled a smooth switch of the diagnostic mode, ultimately accurately triggering the canopy penetration diagnostic mode, which guaranteed the effectiveness of water and fertilizer management in the subsequent growth period of maize. Compared with the traditional single-channel diagnostic method, the yield increased by 8.3% and the water and fertilizer utilization efficiency increased by 12.7%.

[0291] In one embodiment of this application, after generating and executing water and fertilizer decoupling control instructions based on the mid-to-lower layer deficit confirmation results, the method further includes:

[0292] During the water and fertilizer transport recovery period after implementation, the degree of synergy between the renewal and transport of soil root water uptake rhythm signals and canopy transpiration rhythm signals was continuously monitored.

[0293] If the update transmission coordination degree rises above the preset coordination threshold, the current water and fertilizer replenishment status will be maintained.

[0294] If the renewal conduction synergy does not improve, the injection pulse frequency in deep-seated targeted water and fertilizer replenishment will be dynamically adjusted based on the deterioration trend of the renewal conduction synergy.

[0295] In existing technologies, water and fertilizer regulation commands usually lack a closed-loop feedback mechanism after execution, making it impossible to dynamically adjust management strategies based on the actual crop response. This can easily lead to excessive replenishment causing resource waste or insufficient replenishment causing persistent deficits, thus affecting the accuracy and economy of water and fertilizer management.

[0296] In this embodiment, the water and fertilizer conduction recovery period refers to the time window required from the execution of the water and fertilizer decoupling control command to the restoration of the normal water conduction path of the crop roots-stem-leaf. This duration is affected by factors such as crop type, growth stage, soil texture, and environmental temperature and humidity. In this embodiment, for field crops such as corn and wheat, the water and fertilizer conduction recovery period is usually set to 24-72 hours, which can be pre-configured based on historical monitoring data and crop physiological characteristics.

[0297] The updated conduction synergy is an index of the conduction synergy between the soil root water uptake rhythm signal and the canopy transpiration rhythm signal, which is recalculated after the water and fertilizer decoupling control command is executed. The calculation method is completely consistent with the conduction synergy before execution, that is, it is quantified by comparing the time offset between the peak water uptake time point and the peak transpiration time point. The smaller the time offset, the higher the updated conduction synergy, indicating that the water and fertilizer barrier layer has been effectively improved and the water and nutrient transport path tends to be smooth.

[0298] The degradation trend refers to the dynamic change characteristic of the updated transmission coordination degree not only failing to improve (i.e., the time offset not decreasing) compared to the transmission coordination degree before execution, but instead showing a continuous deterioration. Specifically, the time offset increases day by day during the recovery period or the coordination degree value decreases day by day. The degradation trend can be determined by linear regression analysis of monitoring data for 3-5 consecutive days. If the regression slope is negative and the absolute value is greater than the preset degradation threshold (e.g., -0.05 / day), then a degradation trend is determined to exist.

[0299] The injection pulse frequency is the reciprocal of the time interval between injections of the fertilizer solution into the target vertical depth position by the deep fertilizer injector during the deep directional fertilizer replenishment process. The unit is injections per hour. A higher frequency indicates more injections per unit time, and the total amount of fertilizer input increases accordingly. A lower frequency indicates a longer injection interval and a decrease in the total amount of fertilizer input. By dynamically adjusting the injection pulse frequency, the total amount and rhythm of deep fertilizer replenishment can be flexibly controlled without changing the amount of fertilizer injected per injection.

[0300] In this embodiment, once the water and fertilizer decoupling control command is generated based on the confirmation of deficit in the middle and lower layers and executed (i.e., deep-layer targeted water and fertilizer replenishment has been initiated and the transpiration inhibitor in the upper canopy has been sprayed), the agricultural IoT system immediately enters the closed-loop monitoring phase of the water and fertilizer transport recovery period. The core task of this phase is to continuously track the actual response of the crop to water and fertilizer regulation, avoiding management blind spots caused by "terminating monitoring after a one-time regulation".

[0301] Specifically, throughout the entire water and fertilizer transport recovery period (e.g., 24-72 hours after implementation), high-frequency acquisition of soil root water uptake rhythm signals and canopy transpiration rhythm signals is maintained. The acquisition frequency is typically set to once per hour or once every 2 hours to ensure that subtle changes in the water and fertilizer transport pathway recovery process can be captured. After each acquisition, the transport synergy is immediately calculated and updated. The specific calculation method is as follows: extract the peak water uptake time point T of the soil root water uptake rhythm signal within the current daily monitoring window (e.g., 6:00-18:00 on the same day). 吸水 And the peak transpiration time T of the transpiration rhythm signal in the upper canopy 蒸腾 Calculate the time offset ΔT = |T 蒸腾 -T 吸水 |, compare ΔT with the preset time offset tolerance T 容差 Compare the results; if ΔT≤T 容差 If the update transmission coordination degree is higher than the preset coordination threshold, then the update transmission coordination degree is higher than the preset coordination threshold; if ΔT>T 容差 If the update transmission coordination degree is lower than the preset coordination threshold, then the update transmission coordination degree will be lower than the preset coordination threshold.

[0302] During continuous monitoring, differentiated management strategies are implemented based on changes in the degree of synergy in the update transmission process:

[0303] If the update propagation coordination degree is detected to rise above the preset coordination threshold, that is, the time offset ΔT rises from the pre-execution state above the threshold (e.g., ΔT... 执行前 =150 minutes, T 容差 =90 minutes) shrink to the threshold range (e.g., ΔT) 执行后 =75 minutes <90 minutes) indicates that the decoupling regulation strategy of deep-layer targeted water and fertilizer supply and canopy transpiration inhibition has been effective, the water and fertilizer retardation layer has been effectively relieved, the soil-plant water conduction pathway has been restored to smoothness, and the water and nutrients absorbed by the roots can be transported to the upper part of the canopy for transpiration utilization in a timely manner.

[0304] At this point, it is determined that the current water and fertilizer replenishment status (including the parameter configurations such as the duration of the fertilizer injection pulse, the interval period, and the frequency of the fertilizer injection pulse of the deep fertilizer injection actuator) meets the needs of the crop. No additional adjustments are required. The current parameter configuration is automatically locked, and the deep directional water and fertilizer replenishment continues to be maintained until the preset replenishment cycle ends (e.g., it automatically stops after 7 days of continuous replenishment) or the crop enters the next growth stage, triggering a new diagnostic process.

[0305] The purpose of the maintenance strategy is to avoid excessive intervention: once water and fertilizer transport returns to normal, continuing to increase the amount of replenishment or increase the frequency of fertilization will not only fail to further improve crop growth, but may also lead to oversaturation of deep soil moisture, causing root hypoxia or nutrient leaching loss, resulting in resource waste and environmental risks.

[0306] If the monitoring shows that the synergy of the update transmission has not improved, or even shows a deteriorating trend, that is, the time offset ΔT does not shrink or continues to expand during the recovery period (e.g., ΔT=145 minutes on the first day after implementation, ΔT=155 minutes on the second day, and ΔT=168 minutes on the third day, showing a deteriorating trend of increasing day by day), it indicates that the current water and fertilizer decoupling regulation strategy is not strong enough, the deep water and fertilizer supply has failed to effectively break through the water and fertilizer stagnation layer, or the transpiration inhibition effect of the upper canopy has weakened, resulting in the upper part still having a competitive advantage over water and fertilizer.

[0307] At this point, based on the deterioration trend of the update conduction synergy, a dynamic adjustment strategy for the fertilizer injection pulse frequency is implemented. First, the update conduction synergy data sequence (or the corresponding time offset sequence) for 3-5 consecutive days during the recovery period is extracted. Linear regression fitting is performed using the least squares method to calculate the slope k of the deterioration trend. 劣化 For example, if the time offset sequence is [145, 155, 168, 182, 198] (unit: minutes), the fitted slope k is... 劣化 =13.2 minutes / day, indicating that the average time offset increases by 13.2 minutes per day, and the coordination continues to decline.

[0308] Secondly, the slope k 劣化 absolute value |k 劣化 | and the preset degradation threshold k 阈值 Perform a comparison; if |k 劣化 |>k 阈值 (e.g., k) 阈值 If the degradation trend is confirmed to be significant (e.g., 5 minutes / day), adjustments are required; simultaneously, based on |k 劣化 The magnitude of | determines the adjustment range Δf of the fertilizer injection pulse frequency. The adjustment range is positively correlated with the degree of degradation. For example, a linear mapping relationship Δf = α × |k is used. 劣化 |, where α is an adjustment coefficient (e.g., α = 0.1 times / hour / minute / day), when |k 劣化 When |=13.2 minutes / day, Δf=0.1×13.2=1.32 times / hour, rounding up gives Δf=1.4 times / hour.

[0309] Finally, at the original injection pulse frequency f 原 Based on this, the adjustment range Δf is increased to generate a new fertilizer injection pulse frequency f. 新 =f 原 +Δf, and then send it to the deep fertilization actuator for execution. For example: if the original fertilization pulse frequency f 原 = 2 times / hour (i.e., fertilize once every 30 minutes), after adjustment f 新=2+1.4=3.4 times / hour (i.e., fertilize once every 17.6 minutes). The fertilization interval is shortened from 30 minutes to 17.6 minutes, the total amount of water and fertilizer input per unit time increases by 70%, the intensity of deep water and fertilizer replenishment is significantly improved, which helps to break through the stubborn water and fertilizer blockage layer.

[0310] The key advantages of a dynamic adjustment strategy are: closed-loop feedback, where the system can evaluate the control effect in real time by continuously monitoring and updating the transmission synergy, and triggering adjustments as soon as the strategy fails, avoiding the persistent deficit problem caused by the traditional "one-time control without feedback" model; and quantitative adjustment, based on the slope k of the deterioration trend. 劣化 The quantitative calculation of the adjustment range Δf ensures that the adjustment intensity is precisely matched with the actual degree of degradation, avoiding both the waste of resources caused by over-adjustment and the recurrence of problems caused by under-adjustment. It flexibly adapts the fertilizer injection pulse frequency as the adjustment target, without changing physical parameters such as the depth of the fertilizer injection probe or the amount of fertilizer injected at one time. It can be achieved simply through time control, making it easy to execute and responding quickly.

[0311] It should be noted that after adjusting the fertilizer injection pulse frequency, monitoring of the synergy of the transmission should continue. If subsequent monitoring shows that the synergy rises above the preset synergy threshold, the system should immediately switch to Strategy 1 (maintain the current state) and lock the adjusted fertilizer injection pulse frequency. If the synergy continues to deteriorate, dynamic adjustment should be performed again to further increase the fertilizer injection frequency until the maximum frequency limit of the equipment (e.g., 5 times / hour) is reached or a manual intervention warning is triggered (if there is no improvement after 3 consecutive adjustments, it indicates that there may be non-water and fertilizer factors such as soil compaction and root diseases, which require manual on-site investigation).

[0312] For example, in a wheat-growing area, a water and fertilizer blocking layer was detected at a depth of 35cm in the lower part of the canopy during the grain-filling stage. After the deficiency in the lower layer was confirmed, the water and fertilizer decoupling control command was immediately executed: the deep fertilizer injection actuator descended to a depth of 35cm and injected fertilizer at the initial injection pulse frequency f. 原 Deep, targeted water and fertilizer replenishment is performed twice per hour (once every 30 minutes), with each injection of 500 mL of nitrogen, phosphorus, and potassium compound fertilizer solution lasting for 8 seconds; at the same time, transpiration inhibitors (mainly abscisic acid ABA, concentration 50 mg / L) are sprayed onto the upper canopy to reduce the transpiration pull of the upper part.

[0313] Before the control command was executed, monitoring data showed that the peak soil root water absorption occurred at 8:45 AM, the peak canopy transpiration occurred at 11:20 AM, and the time offset ΔT 执行前 =155 minutes, far exceeding the preset time offset tolerance T 容差 =90 minutes, the conduction synergy is lower than the preset synergy threshold, confirming the existence of water and fertilizer blockage.

[0314] After entering the water and fertilizer transport recovery period (set to 72 hours), root water uptake rhythm signals and canopy transpiration rhythm signals were collected every 2 hours to calculate and update the transport synergy.

[0315] Day 1 (within 24 hours after execution): Peak water absorption time 8:50, peak transpiration time 11:05, ΔT = 135 minutes, 20 minutes less than before execution, but still exceeding T. 容差 =90 minutes, the update transmission coordination degree is still lower than the preset coordination threshold, showing an initial trend of improvement but not reaching the target.

[0316] Day 2 (24-48 hours after execution): peak water absorption time 8:55, peak evaporation time 11:15, ΔT=140 minutes, which is 5 minutes longer than Day 1, indicating signs of deterioration.

[0317] Day 3 (48-72 hours after execution): Peak water absorption time 9:00, peak evaporation time 11:35, ΔT=155 minutes, the same as before execution, confirming the deterioration trend.

[0318] Extract the 3-day time offset sequence [135, 140, 155], perform linear regression fitting, and calculate the slope k. 劣化 =10 minutes / day, absolute value |k 劣化 |=10 minutes / day>k 阈值 =5 minutes / day, indicating a significant deterioration trend. Based on the adjustment coefficient α = 0.1 times / hour / minute / day, the adjustment amplitude Δf = 0.1 × 10 = 1.0 times / hour is calculated, generating a new fertilizer injection pulse frequency f. 新 =2+1.0=3.0 times / hour (i.e., fertilizer is injected once every 20 minutes). Immediately issue an adjustment instruction to the deep fertilizer injection actuator, and start executing at the new frequency from the 4th day.

[0319] Day 4 (Day 1 after adjustment): Peak water absorption time 9:05, peak transpiration time 10:50, ΔT = 105 minutes, 50 minutes less than Day 3, the deterioration trend has been curbed, but it still exceeds T. 容差 .

[0320] Day 5 (Day 2 after adjustment): Peak water absorption time 9:10, peak transpiration time 10:35, ΔT = 85 minutes, first time below T 容差 =90 minutes later, the transmission coordination level rises above the preset coordination threshold, and the water and fertilizer transmission path is restored to smooth operation.

[0321] Upon detecting that the synergy level meets the standard, immediately switch to the maintenance strategy: lock the current fertilizer injection pulse frequency f. 新=3.0 times / hour, continue to implement deep-layer targeted water and fertilizer replenishment, no further adjustments will be made. Subsequent monitoring showed that ΔT remained stable in the 75-85 minute range, the wheat grain-filling rate increased by 15.2% compared with before the regulation, the thousand-grain weight increased by 3.8 grams, and the water and fertilizer use efficiency increased by 18.5%. No deep-layer water oversaturation or nutrient leaching was observed, verifying the effectiveness of the closed-loop feedback and dynamic adjustment strategy.

[0322] The control group adopted the traditional "one-time control without feedback" mode, performing the same initial parameters (f) 原 After monitoring was stopped (twice per hour), the time offset was not detected in time when it increased to 155 minutes on the third day. The low-frequency replenishment continued until the seventh day, and the final ΔT only decreased to 130 minutes, which still did not meet the standard. The grouting rate increased by only 8.7%, and the thousand-grain weight increased by 1.9 grams, which was significantly lower than the effect of this embodiment, highlighting the necessity of continuous monitoring and dynamic adjustment.

[0323] As can be seen from the above, this embodiment continuously monitors and updates the synergy of water and fertilizer transport during the recovery period, and executes differentiated strategies based on changes in synergy: when the synergy meets the standard, the current water and fertilizer supply status is maintained to avoid resource waste caused by excessive intervention; when the synergy deteriorates, the frequency of fertilizer injection pulses is dynamically adjusted based on the deterioration trend to quantitatively increase the supply intensity, effectively solving the problem of strategy failure or resource redundancy caused by "one-time regulation without feedback" in traditional technologies. This achieves closed-loop optimization of water and fertilizer management, significantly improving the accuracy and adaptability of deep-level targeted water and fertilizer supply, providing intelligent and dynamic technical support for precision agriculture's four-condition monitoring and water and fertilizer synergistic management, ensuring that the water and fertilizer needs of crops in complex growth environments are met in a timely and accurate manner, and promoting a dual improvement in crop yield and resource utilization efficiency.

[0324] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

[0325] It should be noted that all formulas in this manual are calculated by removing dimensions and taking their numerical values. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters and thresholds in the formulas are set by those skilled in the art according to the actual situation.

[0326] Although embodiments of the invention have been shown and described, those skilled in the art will understand that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the claims and their equivalents.

Claims

1. A method for integrated monitoring of four conditions and coordinated water and fertilizer management in precision agriculture, characterized by: include: During the process of monitoring crops in a target area using an agricultural Internet of Things (IoT) system, the top-level spectral response sequence of the crops is obtained. In response to the detection of a topdressing event, the top spectral response sequence does not produce an increase characteristic that conforms to the preset growth pattern, triggering the canopy penetration diagnostic mode; In the canopy penetration diagnostic mode, the nutrient diagnostic channel based on the top-level spectrum is paused, and the soil root water absorption rhythm signal and the upper canopy transpiration rhythm signal of the crop are obtained. Compare the transduction coordination degree between the soil root water absorption rhythm signal and the canopy transpiration rhythm signal; In response to the transmission coordination degree being lower than a preset coordination threshold, it is determined that there is a water and fertilizer blocking layer in the lower part of the crop canopy, and the vertical depth position of the water and fertilizer blocking layer is determined based on the deep distribution characteristics of the soil root water absorption rhythm signal. Acquire a canopy thermal infrared image, identify vertical temperature heterogeneous regions in the canopy thermal infrared image, cross-validate the distribution depth of the temperature heterogeneous regions with the vertical depth position of the water and fertilizer blocking layer, and generate a middle and lower layer deficiency confirmation result. Based on the confirmation results of the deficiency in the middle and lower layers, a water and fertilizer decoupling control instruction is generated. The water and fertilizer decoupling control instruction is used to provide deep and directional water and fertilizer replenishment to the vertical depth position, and to perform transpiration regulation on the upper part of the canopy to reduce the water and fertilizer competitive advantage of the upper part.

2. The integrated monitoring of four conditions and coordinated water and fertilizer management method as described in claim 1, characterized in that, The response, upon detecting a topdressing event, is that the top-layer spectral response sequence does not exhibit an amplification characteristic consistent with a preset growth pattern, triggering the canopy penetration diagnostic mode, including: After multiple consecutive historical topdressing events, the spectral increments after each topdressing event are extracted from the top-level spectral response sequence. If the spectral increment shows a decreasing trend with each topdressing, and the biomass accumulation rate of the crop remains at the preset growth level, then it is determined that the top-level spectral response sequence does not produce an increase characteristic that conforms to the preset growth law. Pause the nutrient diagnostic pathway based on the top-level spectrum and activate the canopy penetration diagnostic mode based on the soil-plant hydrothermal conduction process.

3. The integrated monitoring of four conditions and coordinated water and fertilizer management method as described in claim 1, characterized in that, The comparison of the transduction coordination between the soil root water uptake rhythm signal and the canopy transpiration rhythm signal includes: Within a preset daily monitoring window, the peak water absorption time of the soil root water absorption rhythm signal and the peak transpiration time of the canopy transpiration rhythm signal are extracted. Calculate the time offset between the peak water absorption time point and the peak evaporation time point; If the time offset is greater than the preset time offset tolerance, and the peak evaporation time point lags behind the peak water absorption time point, then the conduction coordination degree is determined to be lower than the preset coordination threshold.

4. The integrated monitoring of four conditions and coordinated water and fertilizer management method as described in claim 1, characterized in that, The determination of the vertical depth of the water and fertilizer retention layer based on the deep distribution characteristics of the soil root water absorption rhythm signal includes: Obtain stratified root water absorption signals corresponding to different soil depth levels; Compare the response delays of the root water absorption signals of each layer with the transpiration rhythm signals of the upper canopy; The vertical spatial interval between the soil depth layer with the largest response delay and the upper part of the canopy is determined as the spatial range where the water and fertilizer blocking layer is located. The vertical depth of the water and fertilizer blocking layer is determined based on the spatial range.

5. The integrated monitoring of four conditions and coordinated water and fertilizer management method as described in claim 1, characterized in that, The step of cross-validating the distribution depth of the temperature heterogeneous region with the vertical depth of the water and fertilizer retardation layer to generate a deficiency confirmation result in the middle and lower layers includes: In the canopy thermal infrared image, the temperature difference sequence of adjacent pixel rows in the vertical direction is extracted, and the depth of the pixel row in the temperature difference sequence where the temperature change is greater than a preset temperature difference threshold is identified as the distribution depth of the temperature heterogeneous region. If the depth deviation between the distribution depth of the temperature heterogeneous region and the vertical depth of the water and fertilizer blocking layer is less than a preset deviation threshold, then the cross-validation is deemed successful, and a confirmation result indicating water and fertilizer deficiency in the lower middle layer is generated.

6. The integrated monitoring of four conditions and coordinated water and fertilizer management method as described in claim 5, characterized in that, Also includes: If the depth deviation between the distribution depth of the temperature heterogeneous region and the vertical depth position of the water and fertilizer blocking layer is greater than or equal to the preset deviation threshold, it is determined that the hydraulic blocking and thermal characterization are decoupled. In response to the decoupling of hydraulic hindrance and thermal characterization, a deficiency warning result for the middle and lower layers based solely on hydraulic hindrance is generated, and the preset replenishment level for subsequent deep-layer targeted water and fertilizer replenishment is reduced.

7. The integrated monitoring of four conditions and coordinated water and fertilizer management method as described in claim 1, characterized in that, The deep-depth directional water and fertilizer replenishment to the vertical depth location includes: Based on the vertical depth position of the water and fertilizer blocking layer, the fertilizer injection probe of the deep fertilizer injection actuator is controlled to descend to the vertical depth position. The soil texture and water-holding characteristics at the vertical depth position are obtained, and the duration and interval of the fertilization pulse are determined based on the soil texture and water-holding characteristics, so that the injected water and fertilizer form a lateral infiltration front at the vertical depth position and inhibit the longitudinal migration of water and fertilizer to the shallow root distribution area.

8. The integrated monitoring of four conditions and coordinated water and fertilizer management method as described in claim 1, characterized in that, The aforementioned transpiration regulation of the upper canopy to reduce the competitive advantage of upper water and fertilizer use includes: Before performing the deep-targeted water and fertilizer replenishment, spray a transpiration inhibitor onto the upper canopy; During a preset delay period after spraying the transpiration inhibitor, the magnitude of the decrease in the transpiration rhythm signal in the upper canopy was monitored. In response to the decrease reaching a preset inhibition ratio, the deep-layer targeted water and fertilizer replenishment is triggered to allow the deep-layer water and fertilizer replenishment to move to the middle and lower layers under the condition that the transpiration pull in the upper canopy is weakened.

9. The integrated monitoring of four conditions and coordinated water and fertilizer management method as described in claim 1, characterized in that, Prior to triggering the coronal penetration diagnostic mode, the following is also included: When the canopy closure of the crop is in the partial canopy closure transition range, the nutrient diagnosis channel based on the top spectral data and the auxiliary diagnosis channel based on soil-plant hydrothermal conduction are maintained simultaneously. Based on the increasing trend of canopy closure, the decision weight of the nutrient diagnosis channel based on the top-level spectrum is gradually reduced, while the decision weight of the auxiliary diagnosis channel is simultaneously increased, until it is determined that the canopy penetration diagnosis mode is entered.

10. The integrated monitoring of four conditions and coordinated water and fertilizer management method as described in claim 1, characterized in that, After generating and executing the water and fertilizer decoupling control command based on the middle and lower layer deficiency confirmation results, the method further includes: During the water and fertilizer transport recovery period after the implementation, the degree of coordination between the renewal and transport of the soil root water absorption rhythm signal and the canopy transpiration rhythm signal was continuously monitored. If the update transmission coordination degree rises above the preset coordination threshold, the current water and fertilizer replenishment status is maintained. If the update conduction synergy does not improve, the injection pulse frequency in the deep-seated directional water and fertilizer replenishment is dynamically adjusted based on the deterioration trend of the update conduction synergy.

Images