Daily-scale corn water stress diagnosis method based on UAV technology

Abstract

The present invention discloses a daily-scale corn water stress diagnosis method based on drone technology, which relates to the field of agricultural technology. Based on the definition of the water stress coefficient, the present invention introduces the Jarvis stomatal conductance model and combines canopy and plant height data obtained by drones to accurately estimate the aerodynamic impedance and canopy impedance of the canopy, obtains actual evapotranspiration and potential evapotranspiration through a multispectral leaf area index inversion model, and then calculates the daily-scale water stress index. According to the four important growth periods of corn, several groups of drone thermal images are acquired and a statistical water stress index is calculated. The characterization ability of the daily-scale water stress index and the water stress index at different time periods of the day for the measured water index is compared. The method is used to improve the problem that the instantaneous water stress coefficient is insufficiently representative in time and is easily affected by the environment, more accurately characterize crop water stress, and more easily obtain a monitoring effect close to the daily-scale water stress index, thereby improving the practicality of the index.

Description

Technical Field

[0001] The present invention relates to the field of agricultural technology, and in particular to a daily-scale corn water stress diagnosis method based on unmanned aerial vehicle (UAV) technology. Background Art

[0002] Crop water stress is an unavoidable condition in various agricultural production environments. Water shortages can affect crop growth and, in turn, reduce crop yields. Crop water stress is typically described using indicators such as soil moisture and crop stem / leaf water potential, stomatal conductance, and transpiration rate. However, these methods are time-consuming and labor-intensive, requiring expensive specialized equipment for point-scale measurements. Furthermore, they fail to represent the spatial variability of crop water status, making them unsuitable for large-scale farmland scenarios. With the development of drone technology, drones can now carry different types of sensors for crop monitoring. Combining data from these sensors can provide a more comprehensive understanding of the real-time crop situation. Furthermore, drone remote sensing offers advantages such as low cost, short measurement cycles, and flexible operation, facilitating canopy-scale crop water stress research.

[0003] Currently, the most commonly used method for crop water stress monitoring based on UAV platforms is to use the crop water stress index (CWSI) to reflect crop water shortage. The principle is the inverse relationship between canopy temperature (Tc) and leaf transpiration and gs. When plants are under water stress, stomata close, resulting in reduced transpiration cooling and thus higher Tc. CWSI is currently widely considered to be an appropriate indicator for characterizing plant water status.

[0004] The research on CWSI calculation methods based on UAV platforms is mainly divided into three types, which are different from the methods for determining the baseline without water stress and the baseline without transpiration.

[0005] The first is the theoretical CWSI, defined based on the ratio of actual evapotranspiration (ETa) to potential evapotranspiration (ETp) and derived from the crop canopy energy balance equation. The theoretical CWSI uses an energy balance method based on the crop canopy-air temperature difference (dT), net radiation (Rn), aerodynamic impedance (ra), and canopy impedance (rc) to determine upper and lower baselines. This method has a strong theoretical basis, but the parameters involved are extremely complex and are easily affected by the values of ra and rc, leading to incorrect diagnosis of crop water stress.

[0006] The second method is the empirical CWSI (CWSIe), which is derived by establishing upper and lower baselines based on the empirical relationship between dT and the saturation vapor pressure deficit (VPD). The CWSIe can also be calculated by directly measuring the canopy temperature of water-stressed crops (Tdry) and the canopy temperature of well-irrigated crops (Twet), or by using an artificial reference surface to measure temperature in these two situations. However, the upper and lower baselines of the CWSIe must be determined based on different climate environments, different crops, and even different observation equipment. These factors limit the application scenarios of empirical methods.

[0007] The third method is to calculate the statistical CWSI (CWSIs) by calculating the canopy temperature statistical histogram based on the thermal infrared image of the drone to obtain Tdry and Twet. In this method, Twet is estimated by the lowest part of the canopy temperature histogram, and Tdry is obtained from the highest part of the histogram. The calculation of CWSIs does not require meteorological parameters and ground references, which greatly reduces the calculation complexity of CWSI.

[0008] Although current research using unmanned aerial vehicle (UAV) remote sensing platforms to calculate CWSI for crop water stress diagnosis has shown good results in specific environments and at specific times, it is impossible to avoid the strong time variability of UAV observations under changing climatic conditions. Using daily-scale CWSI to reflect crop water stress is an effective method to improve the temporal representativeness of CWSI. Currently, this type of CWSI is observed using the definition method of CWSIt for daily-scale CWSI observations, but the ETa used to calculate this CWSI is often obtained by equipment such as the eddy covariance system, and ETp is mostly estimated by models or deployed with ground-fixed thermal infrared sensors for continuous observations at equal time intervals, and then combined with the calculation method of CWSIe or CWSIt to obtain the daily-scale CWSI.

[0009] These methods enable continuous, automatic, and accurate monitoring of plant water status, but expensive equipment, frequent maintenance, and limited monitoring range restrict their application to scientific research. Furthermore, drones are limited by battery life and cannot provide the same continuous, uninterrupted observations as ground-based equipment. To overcome this shortcoming, some studies have amplified daily "single-shot" images acquired by drone systems, using the instantaneous CWSI at a specific moment as a benchmark for assessing crop water stress. However, this can lead to variable results depending on the time of day and instantaneous meteorological conditions. Alternatively, the optimal daily CWSI observation time has been used instead of full-day observations. It has been found that noon (12:00-15:00) is the optimal time to obtain dT data for diagnosing plant water stress. However, these methods all suffer from the drawback of insufficient representativeness; using only a single instantaneous crop state to fully characterize crop water stress is incomplete.

[0010] In view of the above technical defects, a solution is now proposed. Summary of the Invention

[0011] The purpose of the present invention is to compare the characterization ability of the daily-scale water stress index and the water stress index at different times of the day for the measured water indicators, to improve the problem that the instantaneous water stress coefficient is insufficiently representative in time and is easily affected by the environment, to more accurately characterize crop water stress, to obtain a monitoring effect close to that of the daily-scale water stress index in a more relaxed way, and to improve the practicality of the index.

[0012] To achieve the above objectives, the present invention adopts the following technical solution: a daily-scale corn water stress diagnosis method based on drone technology, comprising the following steps:

[0013] Step 1: Set up several corn planting experimental plots and manage them in groups. Set up several ground control points in each corn planting experimental plot at the corners of the field ridges to facilitate on-site image registration and obtain field data, including soil moisture content, stomatal conductance, meteorological data, and leaf area index.

[0014] The specific process of setting up several corn planting experimental plots is as follows:

[0015] Twelve 8m×8m experimental plots were set up, corresponding to four irrigation treatments, including severe water shortage (W1), moderate water shortage (W2), mild water shortage (W3), and a local empirical irrigation quota of 25mm (W4). Two replicate plots were set up for each treatment. Each plot was equipped with a turbine flowmeter for independent water control, and a 1.5m separation zone was set between the plots.

[0016] To address the issue of water stress in corn caused by irrigation management, W1, W2, and W3 were set at 40%, 60%, and 80% of the W4 treatment's irrigation quota, respectively, and irrigated six times. Furthermore, each treatment was flooded once with a high quota before planting, with the irrigation volume being 225 mm, which was the amount used by local farmers.

[0017] The spring corn planted in the corn planting experimental area was sown on April 24 (DOY = 114), the variety was Simon 3358, and the growing period was 156 days. To avoid soil nutrient deficiency, base fertilizers including urea (CH4N2O, 46%, 200 kg / hm2) and diammonium phosphate fertilizer ((NH4)2HPO4, 14%, 90 kg / hm2) were applied in the field before sowing for all treatments. No additional fertilizer was applied during the growth of the corn.

[0018] Step 2: Based on the definition of the theoretical water stress coefficient, i.e., CWSIt = 1-ETa / ETp, where ETa is the actual evapotranspiration and ETp is the potential evapotranspiration, the soil moisture influence function was introduced and the Jarvis stomatal conductance model was established in combination with field data.

[0019] Step 3: Acquire image data of the corn planting experimental area using a drone, wherein the image data includes multispectral images, visible light images, and thermal infrared images; establish a leaf area index inversion model based on the multispectral images and the leaf area index; calculate air impedance and canopy distribution maps based on the visible light images; and calculate the instantaneous water stress coefficient by combining the canopy distribution map and the thermal infrared image;

[0020] The drone shown is the Phantom 4-M (P4M), a multispectral version of the integrated drone manufactured by Shenzhen DJI Innovations Technology Co., Ltd. This drone is equipped with six 2.9-inch CMOS image sensors, including one color sensor for conventional visible light imaging and five monochrome sensors for multispectral imaging: blue (B, wavelength 450nm±16nm), green (G, wavelength 560nm±16nm), red (R, 650nm±16nm), red edge (RE, 730nm±16nm), and near-infrared (NIR, 840nm±26nm). The P4M also features an integrated light intensity sensor that captures sunlight intensity for image capture and illumination compensation, improving the accuracy of multispectral data acquisition.

[0021] P4M acquisition is performed on the same day as thermal image acquisition, with clear, cloudless weather and wind speeds below level 3. P4M acquisition is generally performed immediately after thermal infrared data acquisition. To ensure image accuracy, the flight altitude is set at 20m, the flight speed is 5m / s, the image is captured while hovering, the pixel resolution is 1.06cm / pixel, and the overlap between heading and lateral directions is 80%.

[0022] Based on the Jarvis stomatal conductance model combined with the canopy leaf area index inversion results, leaf stomatal conductance was upscaled to canopy conductance, and canopy stomatal impedance was calculated for each irrigation treatment at different growth stages. Statistical plots of the diurnal changes in canopy stomatal impedance (Rn>0) for each treatment during the four growth stages showed that canopy stomatal impedance showed a trend of slow changes in the morning and noon, and a gradual increase in the afternoon. A sharp increase occurred after 3:00 PM, which was related to the increase in Ta and VPD in the afternoon, and the decrease in the water supply capacity of the surface soil due to evaporation.

[0023] At this time, near sunset, the solar altitude angle is low, incident radiation decreases, the canopy extinction coefficient increases, the light energy reaching the leaves decreases, and stomata contract, leading to increased canopy resistance. In addition, there are slight differences in canopy stomatal resistance between different growth stages. The rc during V9 and R3 is greater than during VT and R1. This is because the canopy LAI of the corn during V9 is low, so the canopy stomatal resistance is higher at this time. During VT and R1, corn leaves are growing vigorously and LAI is at its peak, so the canopy stomatal resistance is low. However, during R3, which is the late stage of corn growth, the canopy leaves are aging, and the stomatal conductance decreases, resulting in increased canopy stomatal resistance.

[0024] Looking at the different water control treatments, canopy stomatal impedance showed a trend of W1>W2>W3>W4 across the four growth stages. Corn canopy stomatal impedance continued to increase with reduced irrigation water. This made it difficult for the roots to obtain sufficient water from the dry soil to meet the plant's carbon metabolism and assimilation needs, leading to reduced leaf water potential and cell tension, which in turn reduced stomatal conductance. Consequently, canopy stomatal impedance increased with reduced irrigation water.

[0025] Step 4: Establish a canopy impedance model based on the Jarvis stomatal conductance model and the leaf area index inversion model. The canopy impedance model is combined with air impedance to calculate the daily actual evapotranspiration, and meteorological data is obtained to calculate the daily potential evapotranspiration. Then, the daily water stress index is calculated based on the actual evapotranspiration and the potential evapotranspiration.

[0026] Step 5. Acquire several sets of drone thermal images according to the four important growth periods of corn and calculate the statistical water stress index. Compare the characterization ability of the daily-scale water stress index and the statistical water stress index for the measured moisture indicators at different times of the day. Divide the training set and test set into a ratio of 70% and 30%, and put them into the RF regression model in order of importance for modeling, and output the results.

[0027] The temperature Ta, relative humidity RH, emissivity ε, and flight altitude measured by the weather station at the time of the drone flight were obtained. These data were then input into DJIThermalSDK1.4 software to obtain the actual temperature Ts. The images were then batch-converted from R-JPEG to RAW format. The Python Numpy module was used to convert the RAW format to TIF format. The piexif module was then used to export the shooting position and orientation system information. The extracted orientation system information and TIF files were then imported into Pix4D software for orthorectification and stitching. GCPs were then used for optimization to keep the ground error less than 3 cm.

[0028] Before each flight, a black and white plate on the ground was photographed and its temperature was measured using a handheld infrared thermometer to calibrate Tc.

[0029] Furthermore, based on the soil moisture influence function f θ (θ) The specific process of constructing the stomatal conductance model is as follows:

[0030] Preset f θ When the (θ) term is 1, the crop reaches a non-water stress state, and the canopy impedance at this time is rcp. The model form is: g s =f p (PAR)·f v (VPD)·f t (T a )·f c (C)·f θ (θ)#(13), where meteorological factors include photosynthetically active radiation f p (PAR), saturated vapor pressure difference f v (VPD), air temperature f t (T a ), CO2 concentration in the atmosphere f c (C) Soil moisture content of the cultivated layer f θ (θ);

[0031] The model formula for meteorological factors and stomatal conductance is as follows:

[0032] The light and effective radiation f p (PAR):

[0033]

[0034] The saturated vapor pressure difference f v (VPD):

[0035]

[0036] The air temperature f t (T a ):

[0037]

[0038] The soil moisture content f θ (θ):

[0039] Since the CO2 concentration in the environment changes very little, the effect on stomatal conductance can be ignored, where f c (C) takes the value of 1, θ is the average soil moisture content of the root water absorption layer, and the average soil moisture content of 0-60 cm is selected; θ w is the wilting moisture content; θ cm is the field water capacity;

[0040] f p (PAR), f v (VPD), f t (T a ),f θ (θ) can be combined to obtain N different Jarvis stomatal conductance models. The measured stomatal conductance data of several groups are divided into training set and validation set according to the ratio of 7:3. According to the validation set model R 2 The optimal stomatal conductance model was obtained by filtering with RMSE.

[0041] Furthermore, the specific process of obtaining visible light data and collecting multispectral images is as follows:

[0042] Planar images of the corn planting experimental area were acquired using a drone. DJI Terra 3.6.8 software was used to reconstruct and stitch these images, and black-and-white ground plate photos were added for radiation correction. This yielded MS orthophotos and digital surface images of the corn planting experimental area. Ground control points were added for optimization, achieving a ground error of less than 3 cm. A total of 1,081 to 1,092 images were acquired for each mission.

[0043] Furthermore, the specific process of calculating the canopy leaf area index and plant height data is as follows:

[0044] By direct measurement, three representative corn plants with uniform growth were selected from each corn planting experimental plot. The leaf length Lij and maximum leaf width Bij of each leaf were measured with a tape measure. The canopy leaf area index LAI was calculated according to the following:

[0045] Where n is the total number of leaves of the jth plant, m is the number of plants; ρ is the planting density; 0.75 is the correction factor;

[0046] In each corn planting experimental area, a corn plant was selected and its height was measured with a tape measure. The longitude and latitude of the measured plant were recorded to calibrate the plant height data obtained by the drone.

[0047] Furthermore, the specific process of calculating canopy impedance data is as follows:

[0048] A drone was used to obtain planar images of the corn planting experimental area, and the LCPro T fully automatic portable photosynthetic meter produced by AT Company in the UK was used to measure transpiration rate, stomatal conductance, and photosynthetic rate.

[0049] In each corn planting experimental area, three typical corn leaves were randomly selected to measure the stomatal conductance at the 1 / 3 of the leaf surface close to the leaf tip, and the average value was taken as the stomatal conductance value of the sampling point.

[0050] Soil sampling was carried out 5-7 days after irrigation in each corn planting experimental plot. Three sampling points were randomly selected near the center of the plot. Soil samples at three depths, 0-20 cm, 20-40 cm, and 40-60 cm, were obtained using a soil drill. The soil samples from the three sampling points at each depth were mixed and placed in an aluminum box. The soil moisture content at each depth was measured after drying in an oven at 105°C for 8 hours. This was used as a set of soil moisture sample values. A total of 48 sets of measured soil moisture data were obtained in the experiment.

[0051] The stomatal conductance g of the leaf is calculated according to the following formula s Converted to pore impedance r s :

[0052] The default Jarvis stomatal conductance model is f θ (θ)=1, that is, the average soil moisture content is equal to the field water holding capacity θ=θ c = 0.32, potential evaporation occurs, and then r sp ;

[0053] Canopy impedance data were calculated according to the following formula: Among them LAI c To remove the soil background value, r c is the canopy stomatal impedance, and r can be obtained by the same logic. cp .

[0054] Furthermore, the actual evapotranspiration and potential evapotranspiration are calculated using the PM method as follows:

[0055]

[0056] Where R n is the net radiation MJm / 2day, G is the soil heat flux MJm / 2day, r a is the aerodynamic drag sm-1, r c is the canopy resistance sm-1, r cp is the potential canopy resistance during potential evapotranspiration sm-1, γ is the dry-wet constant, ρ is the air density kgm-3, VPD is the saturated water vapor pressure difference kPa, T is the daily average temperature at 2m height, Δ is the slope of the water vapor pressure curve, C p is the heat capacity of air, specifically 1.013MJ / kg-1℃-1)

[0057] The air density is calculated according to the following formula: ρ = 1.2837 - 0.039T a #(6), where T a is the air temperature;

[0058] The dry and wet constant γ is calculated according to the following formula: Where Z is the altitude;

[0059] The slope of the water vapor pressure curve Δ is calculated according to the following formula:

[0060] The saturated vapor pressure difference VPD is calculated according to the following formula: Where RH represents the relative humidity of air;

[0061] The aerodynamic resistance r a Calculated according to the following formula: Where d represents the zero-plane displacement in m, which is 0.63h; z0 represents the roughness in m, which is 0.13h, where h represents the crop height in m; k represents the Kalman coefficient, which is 0.41; μ represents the wind speed at the reference height in m / s, and z represents the reference height of the weather station in m, i.e., the wind speed measurement height.

[0062] Furthermore, the specific process of calculating the daily water stress index is as follows:

[0063] S101. Calculate the water stress index according to the following formula: Where T c is the average canopy temperature of each treatment, T wet and T dry Corresponds to canopy temperatures when corn plants are transpiring at their maximum rate and when they are not transpiring at all;

[0064] The canopy temperature frequency histograms used to calculate each treatment were drawn using the ROIComputeStatistics function of ENVI5.6. The temperature distribution of the corn canopy and soil in the image is close to a bimodal form, and the temperature distribution of the pure corn canopy is a Gaussian distribution.

[0065] T wet and T dry The 0.5% and 99.5% mean values were taken from the pure corn canopy temperature histogram, respectively.

[0066] S102, calculate theoretical water stress index t according to the following formula: Where ET a is the actual evapotranspiration, ET p is the potential evapotranspiration;

[0067] The theoretical water stress index and the statistical water stress index are both dimensionless, and their values range from 0 to 1, where 0 and 1 represent no water stress and the most severe water stress, respectively.

[0068] S103. The meteorological data observation frequency is preset to 30 minutes / time, that is, the average meteorological data is obtained every 30 minutes. The unit of the calculated actual evapotranspiration and potential evapotranspiration is mm / 30 minutes. Where ET a_i is the ET of the i-th 30 minutes in a day a ,According to the analysis of daytime conditions, Rn≥0.

[0069] Furthermore, the calculation method of the plant height h of the crop at each growth stage is as follows:

[0070] The digital surface model DSM0 of the farmland before seedling emergence obtained by the drone is used as the base value, and ground landmarks are selected to calibrate the DSM of each growth period. i , and then get h i =DSM i -DSM0(i=1,2,3,...,i)≠(11);

[0071] After correcting hi with the actual plant height measured on the ground, the final plant height h of crops at each growth stage was obtained using ENVI5.6 software.

[0072] In summary, due to the adoption of the above technical solution, the beneficial effects of the present invention are:

[0073] This daily-scale corn water stress diagnosis method based on drone technology is based on the definition of the water stress coefficient. It introduces the Jarvis stomatal conductance model and combines the canopy and plant height data obtained by drones to accurately estimate the aerodynamic impedance and canopy impedance of the canopy. The actual evapotranspiration and potential evapotranspiration are obtained through the multispectral leaf area index inversion model, and then the daily-scale water stress index is calculated. According to the four important growth periods of corn, several groups of drone thermal images are collected and the statistical water stress index is calculated. The characterization ability of the daily-scale water stress index and the water stress index at different times of the day for the measured water indicators is compared. This method is used to improve the problem that the instantaneous water stress coefficient is insufficiently representative in time and is easily affected by the environment, more accurately characterize crop water stress, and more easily obtain monitoring effects close to the daily-scale water stress index, thereby improving the practicality of the index. BRIEF DESCRIPTION OF THE DRAWINGS

[0074] Figure 1 The figure shows a schematic diagram of the daily scale corn water stress calculation process of the present invention;

[0075] Figure 2 A schematic diagram of the daily scale corn water stress calculation process of the present invention is shown;

[0076] Figure 3 A schematic diagram of the classification of the Jarvis stomatal conductance model of the present invention is shown;

[0077] Figure 4 A schematic diagram of daily variation data of canopy impedance during the growth period of corn according to the present invention is shown;

[0078] Figure 5 A schematic diagram of the output results of the RF regression model of the present invention is shown. DETAILED DESCRIPTION

[0079] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0080] Example 1:

[0081] like Figure 1-5 As shown in FIG, the daily-scale corn water stress diagnosis method based on UAV technology includes the following steps:

[0082] Step 1: Set up several corn planting experimental plots and manage them in groups. Set up several ground control points in each corn planting experimental plot at the corners of the field ridges to facilitate on-site image registration and obtain field data, including soil moisture content, stomatal conductance, meteorological data, and leaf area index.

[0083] The specific process of setting up several corn planting experimental plots is as follows:

[0084] Twelve 8m×8m experimental plots were set up, corresponding to four irrigation treatments, including severe water shortage (W1), moderate water shortage (W2), mild water shortage (W3), and a local empirical irrigation quota of 25mm (W4). Two replicate plots were set up for each treatment. Each plot was equipped with a turbine flowmeter for independent water control, and a 1.5m separation zone was set between the plots.

[0085] To address the issue of water stress in corn caused by irrigation management, W1, W2, and W3 were set at 40%, 60%, and 80% of the W4 treatment's irrigation quota, respectively, and irrigated six times. Furthermore, each treatment was flooded once with a high quota before planting, with the irrigation volume being 225 mm, which was the amount used by local farmers.

[0086] The spring corn planted in the corn planting experimental area was sown on April 24 (DOY = 114), the variety was Simon 3358, and the growing period was 156 days. To avoid soil nutrient deficiency, base fertilizers including urea (CH4N2O, 46%, 200 kg / hm2) and diammonium phosphate fertilizer ((NH4)2HPO4, 14%, 90 kg / hm2) were applied in the field before sowing for all treatments. No additional fertilizer was applied during the growth of the corn.

[0087] Step 2: Based on the definition of the theoretical water stress coefficient, i.e., CWSIt = 1-ETa / ETp, where ETa is the actual evapotranspiration and ETp is the potential evapotranspiration, the soil moisture influence function was introduced and the Jarvis stomatal conductance model was established in combination with field data.

[0088] The temperature Ta, relative humidity RH, emissivity ε, and flight altitude measured by the weather station at the time of the drone flight were obtained. These data were then input into DJIThermalSDK1.4 software to obtain the actual temperature Ts. The images were then batch-converted from R-JPEG to RAW format. The Python Numpy module was used to convert the RAW format to TIF format. The piexif module was then used to export the shooting position and orientation system information. The extracted orientation system information and TIF files were then imported into Pix4D software for orthorectification and stitching. GCPs were then used for optimization to keep the ground error less than 3 cm.

[0089] Before each flight, a black and white plate on the ground was photographed and its temperature was measured using a handheld infrared thermometer to calibrate Tc.

[0090] Based on the soil moisture influence function f θ (θ) The specific process of constructing the stomatal conductance model is as follows:

[0091] Preset f θ When the (θ) term is 1, the crop reaches a non-water stress state, and the canopy impedance at this time is rcp. The model form is: g s =f p (PAR)·f v (VPD)·f t (T a )·f c (C)·f θ (θ)#(13), where meteorological factors include photosynthetically active radiation f p (PAR), saturated vapor pressure difference f v (VPD), air temperature f t (T a ), CO2 concentration in the atmosphere f c (C) Soil moisture content of the cultivated layer f θ (θ);

[0092] The model formula for meteorological factors and stomatal conductance is as follows:

[0093] Light and effective radiation f p (PAR):

[0094]

[0095] Saturated vapor pressure difference fv (VPD):

[0096]

[0097] Temperature t (T a ):

[0098]

[0099] Soil moisture content f θ (θ):

[0100] Since the CO2 concentration in the environment changes very little, the effect on stomatal conductance can be ignored, where f c (C) takes the value of 1, θ is the average soil moisture content of the root water absorption layer, and the average soil moisture content of 0-60 cm is selected; θ w is the wilting moisture content; θ cm is the field water capacity;

[0101] f p (PAR), f v (VPD), f t (T a ),f θ (θ) combination can obtain N different Jarvis stomatal conductance models, which are 1-1-1-3-3-2, such as Figure 3 As shown in the figure, the measured stomatal conductance data of several groups were divided into training set and validation set in the ratio of 7:3. 2 The optimal stomatal conductance model was obtained by filtering with RMSE.

[0102] Step 3: Acquire image data of the corn planting experimental area using a drone, wherein the image data includes multispectral images, visible light images, and thermal infrared images; establish a leaf area index inversion model based on the multispectral images and the leaf area index; calculate air impedance and canopy distribution maps based on the visible light images; and calculate the instantaneous water stress coefficient by combining the canopy distribution map and the thermal infrared image;

[0103] The drone shown is the Phantom 4-M (P4M), a multispectral version of the integrated drone manufactured by Shenzhen DJI Innovations Technology Co., Ltd. This drone is equipped with six 2.9-inch CMOS image sensors, including one color sensor for conventional visible light imaging and five monochrome sensors for multispectral imaging: blue (B, wavelength 450nm±16nm), green (G, wavelength 560nm±16nm), red (R, 650nm±16nm), red edge (RE, 730nm±16nm), and near-infrared (NIR, 840nm±26nm). The P4M also features an integrated light intensity sensor that captures sunlight intensity for image capture and illumination compensation, improving the accuracy of multispectral data acquisition.

[0104] P4M acquisition is performed on the same day as thermal image acquisition, with clear, cloudless weather and wind speeds below level 3. P4M acquisition is generally performed immediately after thermal infrared data acquisition. To ensure image accuracy, the flight altitude is set at 20m, the flight speed is 5m / s, the image is captured while hovering, the pixel resolution is 1.06cm / pixel, and the overlap between heading and lateral directions is 80%.

[0105] Based on the Jarvis stomatal conductance model and the canopy leaf area index inversion results, the leaf stomatal conductance was upscaled to the canopy conductance, and the canopy stomatal impedance of each irrigation treatment at different growth stages was calculated. The diurnal changes of canopy stomatal impedance (Rn>0) of each treatment at four growth stages were statistically plotted. Figure 4 As shown, the canopy stomatal impedance showed a trend of slow change in the morning and noon, and gradually increased in the afternoon for all four days, and a sharp increase after 15:00. This is related to the increase in Ta and VPD in the afternoon, and the decrease in the water supply capacity of the surface soil due to evaporation.

[0106] At this time, near sunset, the solar altitude angle is low, incident radiation decreases, the canopy extinction coefficient increases, the light energy reaching the leaves decreases, and stomata contract, leading to increased canopy resistance. In addition, there are slight differences in canopy stomatal resistance between different growth stages. The rc during V9 and R3 is greater than during VT and R1. This is because the canopy LAI of the corn during V9 is low, so the canopy stomatal resistance is higher at this time. During VT and R1, corn leaves are growing vigorously and LAI is at its peak, so the canopy stomatal resistance is low. However, during R3, which is the late stage of corn growth, the canopy leaves are aging, and the stomatal conductance decreases, resulting in increased canopy stomatal resistance.

[0107] Looking at the different water control treatments, canopy stomatal impedance showed a trend of W1>W2>W3>W4 across the four growth stages. Corn canopy stomatal impedance continued to increase with reduced irrigation water. This made it difficult for the roots to obtain sufficient water from the dry soil to meet the plant's carbon metabolism and assimilation needs, leading to reduced leaf water potential and cell tension, which in turn reduced stomatal conductance. Consequently, canopy stomatal impedance increased with reduced irrigation water.

[0108] The specific process of acquiring visible light data and collecting multispectral images is as follows:

[0109] Planar images of the corn planting experimental area were acquired using a drone. DJI Terra 3.6.8 software was used to reconstruct and stitch these images, and black-and-white ground plate photos were added for radiation correction. This yielded MS orthophotos and digital surface images of the corn planting experimental area. Ground control points were added for optimization, achieving a ground error of less than 3 cm. A total of 1,081 to 1,092 images were acquired for each mission.

[0110] The specific process of calculating the canopy leaf area index and plant height data is as follows:

[0111] By direct measurement, three representative corn plants with uniform growth were selected from each corn planting experimental plot. The leaf length Lij and maximum leaf width Bij of each leaf were measured with a tape measure. The canopy leaf area index LAI was calculated according to the following:

[0112] Where n is the total number of leaves of the jth plant, m is the number of plants; ρ is the planting density; 0.75 is the correction factor;

[0113] In each corn planting experimental area, a corn plant was selected and its height was measured with a tape measure. The longitude and latitude of the measured plant were recorded to calibrate the plant height data obtained by the drone.

[0114] The specific process of calculating canopy impedance data is as follows:

[0115] A drone was used to obtain planar images of the corn planting experimental area, and the LCPro T fully automatic portable photosynthetic meter produced by AT Company in the UK was used to measure transpiration rate, stomatal conductance, and photosynthetic rate.

[0116] In each corn planting experimental area, three typical corn leaves were randomly selected to measure the stomatal conductance at the 1 / 3 of the leaf surface close to the leaf tip, and the average value was taken as the stomatal conductance value of the sampling point.

[0117] Soil sampling was carried out 5-7 days after irrigation in each corn planting experimental plot. Three sampling points were randomly selected near the center of the plot. Soil samples at three depths, 0-20 cm, 20-40 cm, and 40-60 cm, were obtained using a soil drill. The soil samples from the three sampling points at each depth were mixed and placed in an aluminum box. The soil moisture content at each depth was measured after drying in an oven at 105°C for 8 hours. This was used as a set of soil moisture sample values. A total of 48 sets of measured soil moisture data were obtained in the experiment.

[0118] The stomatal conductance g of the leaf is calculated according to the following formula s Converted to pore impedance r s :

[0119] The default Jarvis stomatal conductance model is f θ (θ)=1, that is, the average soil moisture content is equal to the field water holding capacity θ=θ c = 0.32, potential evaporation occurs, and then r sp ;

[0120] Canopy impedance data were calculated according to the following formula: Among them LAI c To remove the soil background value, the canopy LAI distribution, r c is the canopy stomatal impedance, and r can be obtained by the same logic. cp .

[0121] The calculation process of actual evapotranspiration and potential evapotranspiration using the PM method is as follows:

[0122]

[0123] Where R n is the net radiation MJm / 2day, G is the soil heat flux MJm / 2day, r a is the aerodynamic drag sm-1, r c is the canopy resistance sm-1, r cp is the potential canopy resistance during potential evapotranspiration sm-1, γ is the dry-wet constant, ρ is the air density kgm-3, VPD is the saturated water vapor pressure difference kPa, T is the daily average temperature at 2m height, Δ is the slope of the water vapor pressure curve, C p is the heat capacity of air, specifically 1.013MJ / kg-1℃-1)

[0124] The air density is calculated according to the following formula: ρ = 1.2837 - 0.039T a #(6), where T a is the air temperature;

[0125] The dry and wet constant γ is calculated according to the following formula: Where Z is the altitude;

[0126] The slope of the water vapor pressure curve Δ is calculated according to the following formula:

[0127] The saturated vapor pressure difference VPD is calculated according to the following formula: Where RH represents the relative humidity of air;

[0128] The aerodynamic resistance r a Calculated according to the following formula: Where d represents the zero-plane displacement in m, which is 0.63h; z0 represents the roughness in m, which is 0.13h, where h represents the crop height in m; k represents the Kalman coefficient, which is 0.41; μ represents the wind speed at the reference height in m / s, and z represents the reference height of the weather station in m, i.e., the wind speed measurement height.

[0129] Step 4: Establish a canopy impedance model based on the Jarvis stomatal conductance model and the leaf area index inversion model. The canopy impedance model is combined with air impedance to calculate the daily actual evapotranspiration, and meteorological data is obtained to calculate the daily potential evapotranspiration. Then, the daily water stress index is calculated based on the actual evapotranspiration and the potential evapotranspiration.

[0130] The specific process of calculating the daily water stress index is as follows:

[0131] S101. Calculate the water stress index according to the following formula: Where T c is the average canopy temperature of each treatment, T wet and T dry Corresponds to canopy temperatures when corn plants are transpiring at their maximum rate and when they are not transpiring at all;

[0132] The canopy temperature frequency histograms used to calculate each treatment were drawn using the ROIComputeStatistics function of ENVI5.6. The temperature distribution of the corn canopy and soil in the image is close to a bimodal form, and the temperature distribution of the pure corn canopy is a Gaussian distribution.

[0133] T wet and T dry The 0.5% and 99.5% mean values were taken from the pure corn canopy temperature histogram, respectively.

[0134] S102, calculate theoretical water stress index t according to the following formula: Where ET a is the actual evapotranspiration, ET p is the potential evapotranspiration;

[0135] The theoretical water stress index and the statistical water stress index are dimensionless, and their values range from 0 to 1, where 0 and 1 represent no water stress and the most severe water stress, respectively.

[0136] S103. The meteorological data observation frequency is preset to 30 minutes / time, that is, the average meteorological data is obtained every 30 minutes. The unit of the calculated actual evapotranspiration and potential evapotranspiration is mm / 30 minutes. Where ET a_i is the ET of the i-th 30 minutes in a day a ,According to the analysis of daytime conditions, Rn≥0.

[0137] The calculation method of plant height h in each growth period of crops is as follows:

[0138] The digital surface model DSM0 of the farmland before seedling emergence obtained by the drone is used as the base value, and ground landmarks are selected to calibrate the DSM of each growth period. i , and then get h i =DSM i -DSM0(i=1,2,3,...,i)#(11);

[0139] After correcting hi with the actual plant height measured on the ground, the final plant height h of crops at each growth stage was obtained using ENVI5.6 software.

[0140] Step 5. Acquire several sets of drone thermal images according to the four important growth periods of corn and calculate the statistical water stress index. Compare the characterization ability of the daily-scale water stress index and the statistical water stress index for the measured moisture indicators at different times of the day. Divide the training set and test set into a ratio of 70% and 30%, and put them into the RF regression model in order of importance for modeling, and output the results.

[0141] The water stress index values at different growth stages were used as independent variables at different time periods of the day, and the daily water stress index was used as the dependent variable to evaluate the importance of the RF model. The training set and the test set were divided into 70% and 30% respectively, and the values were ranked according to importance and put into the RF regression model for modeling. The results showed that Figure 5 As shown;

[0142] The model prediction accuracy of the data set with all-day observations in 6 periods was the highest, with an R2 of 0.81 and an RMSE of 0.039. The R2 reached 0.79 and the RMSE reached 0.04, which is very close to the accuracy of all-day observations in 6 periods. This shows that observations in three periods from noon to early afternoon can replace all-day observations and achieve similar results, thereby improving the cost-effectiveness of daily water stress index observations.

[0143] Based on the definition of the water stress coefficient, the present invention introduces the Jarvis stomatal conductance model and combines it with the canopy and plant height data obtained by drones to accurately estimate the aerodynamic impedance and canopy impedance of the canopy. The actual evapotranspiration and potential evapotranspiration are obtained through the multispectral leaf area index inversion model, and then the daily water stress index is calculated. According to the four important growth periods of corn, several groups of drone thermal images are acquired and the statistical water stress index is calculated. The characterization ability of the daily water stress index and the water stress index for the measured water indicators at different times of the day are compared. This method is used to improve the problems of insufficient temporal representativeness of the instantaneous water stress coefficient and susceptibility to environmental influences, more accurately characterize crop water stress, and more easily obtain a monitoring effect close to the daily water stress index, thereby improving the practicality of the index.

[0144] The above formulas are all dimensionless and numerical calculations. The formulas are obtained by collecting a large amount of data and performing software simulation to obtain the most recent real situation. The preset parameters in the formulas are set by those skilled in the art according to actual conditions.

[0145] The above description is only a preferred specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any technician familiar with the technical field, within the technical scope disclosed by the present invention, who makes equivalent replacements or changes based on the technical solution and inventive concept of the present invention, should be covered by the scope of protection of the present invention.

Claims

1. A daily-scale corn water stress diagnosis method based on drone technology, characterized in that: The following steps are included: Step 1: Set up several corn planting experimental plots and manage them in groups. Set up several ground control points in each corn planting experimental plot, set them at the corners of the field ridges, and obtain field data. The field data includes soil moisture content, stomatal conductance, meteorological data, and leaf area index. Step 2: Based on the definition of the theoretical water stress coefficient, i.e., CWSIt = 1-ETa / ETp, where ETa is the actual evapotranspiration and ETp is the potential evapotranspiration, the soil moisture influence function was introduced and the Jarvis stomatal conductance model was established in combination with field data. Step 3: Acquire image data of the corn planting experimental area using a drone, wherein the image data includes multispectral images, visible light images, and thermal infrared images; establish a leaf area index inversion model based on the multispectral images and the leaf area index; calculate air impedance and canopy distribution maps based on the visible light images; and calculate a statistical water stress index at an instantaneous scale by combining the canopy distribution maps and thermal infrared images; Step 4: Establish a canopy impedance model based on the Jarvis stomatal conductance model and the leaf area index inversion model. The canopy impedance model is combined with air impedance to calculate the daily actual evapotranspiration, and meteorological data is obtained to calculate the daily potential evapotranspiration. Then, the daily water stress index is calculated based on the actual evapotranspiration and the potential evapotranspiration. The specific process of calculating the daily water stress index is as follows: S101. Calculate the water stress index according to the following formula: Where T c is the average canopy temperature of each treatment, T wet and T dry Corresponds to canopy temperatures when corn plants are transpiring at their maximum rate and when they are not transpiring at all; T wet and T dry The 0.5% and 99.5% averages of the pure corn canopy temperature histogram were taken, respectively; S102, calculate theoretical water stress index t according to the following formula: Where ET a is the actual evapotranspiration, ET p is the potential evapotranspiration; S103. The observation frequency of meteorological data is preset to 30 minutes / time, that is, the average meteorological data is obtained every 30 minutes. The unit of the calculated actual evapotranspiration and potential evapotranspiration is mm / 30 minutes. Then ET a_day =∑ET a_i #(12), where ET a_i is the ET of the i-th 30 minutes in a day a ,According to the analysis of daytime conditions, Rn≥0; Step 5. Acquire several sets of drone thermal images according to the four important growth periods of corn and calculate the instantaneous statistical water stress index. Compare the characterization ability of the daily water stress index and the statistical water stress index for the measured moisture index at different times of the day. Divide the training set and test set into a ratio of 70% and 30%, and put them into the RF regression model in order of importance for modeling, and output the results.

2. The daily scale corn water stress diagnosis method based on drone technology according to claim 1 is characterized in that: Based on the soil moisture influence function f θ (θ) The specific process of constructing the stomatal conductance model is as follows: Preset f θ When the (θ) term is 1, the crop reaches a non-water stress state, and the canopy impedance at this time is rcp. The model form is: g s =f p (PAR)·f v (VPD)·f t (T a )·f c (C)·f θ (θ)#(13), where meteorological factors include photosynthetically active radiation influence function f p (PAR), saturated vapor pressure difference influence function f v (VPD), temperature impact function f t (T a ), the influence function of CO2 concentration in the atmosphere f c (C) Influence function of soil moisture content in the cultivated layer f θ (θ); The model formula for meteorological factors and stomatal conductance is as follows: The photosynthetically active radiation influence function f p (PAR): The saturated vapor pressure difference influence function f v (VPD): The temperature influence function f t (T a ): The soil moisture influence function f θ (θ): where f c (C) is taken as 1, θ is the average soil moisture content in the root water absorption layer, θ w is the wilting moisture content; θ c is the field water capacity; f p (PAR), f v (VPD), f t (Ta), f θ (θ) can be combined to obtain N different Jarvis stomatal conductance models. The measured stomatal conductance data of several groups are divided into training set and validation set according to the ratio of 7:

3. According to the validation set model R 2 The optimal stomatal conductance model was obtained by filtering with RMSE.

3. The daily scale corn water stress diagnosis method based on drone technology according to claim 1 is characterized in that: The specific process of acquiring visible light data and collecting multispectral images is as follows: Planar images of the corn planting experimental area were acquired using a drone. DJI Terra 3.6.8 software was used to reconstruct and stitch these images, and black-and-white ground plate photos were added for radiation correction. This yielded MS orthophotos and digital surface images of the corn planting experimental area. Ground control points were added for optimization, achieving a ground error of less than 3 cm. A total of 1,081 to 1,092 images were acquired for each mission.

4. The daily scale corn water stress diagnosis method based on drone technology according to claim 1 is characterized in that: The specific process of calculating the canopy leaf area index and plant height data is as follows: By direct measurement, three representative corn plants with uniform growth were selected from each corn planting experimental plot. The leaf length Lij and maximum leaf width Bij of each leaf were measured with a tape measure. The canopy leaf area index LAI was calculated according to the following: Where n is the total number of leaves of the jth plant, m is the number of plants; ρ is the planting density; 0.75 is the correction factor; In each corn planting experimental area, a corn plant was selected and its height was measured with a tape measure. The longitude and latitude of the measured plant were recorded to calibrate the plant height data obtained by the drone.

5. The daily scale corn water stress diagnosis method based on drone technology according to claim 1 is characterized in that: The specific process of calculating canopy impedance data is as follows: A drone was used to obtain planar images of the corn planting experimental area, and the LCPro T fully automatic portable photosynthetic meter produced by AT Company in the UK was used to measure transpiration rate, stomatal conductance, and photosynthetic rate. The stomatal conductance g of the leaf is calculated according to the following formula s Converted to pore impedance r s : The default Jarvis stomatal conductance model is f θ (θ)=1, that is, the average soil moisture content is equal to the field water holding capacity θ=θ c = 0.32, potential evaporation occurs, and then r sp ; Canopy impedance data were calculated according to the following formula: Among them LAI c To remove the soil background value, r c is the canopy stomatal impedance, and r can be obtained by the same logic. cp .

6. The daily scale corn water stress diagnosis method based on drone technology according to claim 1 is characterized in that: The calculation process of actual evapotranspiration and potential evapotranspiration using the PM method is as follows: Where R n is the net radiation MJm / 2day, G is the soil heat flux MJm / 2day, r a is the aerodynamic drag sm -1 , r c is the canopy resistance sm -1 , r cp is the potential canopy resistance sm when potential evapotranspiration is -1 , γ is the dry and wet constant, ρ is the air density kgm -3 , VPD is the saturated water vapor pressure difference kPa, T is the daily average temperature at a height of 2m, Δ is the slope of the water vapor pressure curve, C p is the heat capacity of air, specifically 1.013MJ / kg -1 ℃ -1 ; The air density is calculated according to the following formula: ρ = 1.2837-0.039T a #(6), where T a is the air temperature; The dry and wet constant γ is calculated according to the following formula: Where Z is the altitude; The slope of the water vapor pressure curve Δ is calculated according to the following formula: The saturated vapor pressure difference VPD is calculated according to the following formula: Where RH represents the relative humidity of air; The aerodynamic resistance r a Calculated according to the following formula: Where d represents the zero-plane displacement in m, which is 0.63h; z0 represents the roughness in m, which is 0.13h, where h represents the crop height in m; k represents the Kalman coefficient, which is 0.41; μ represents the wind speed at the reference height in m / s, and z represents the reference height of the weather station in m, i.e., the wind speed measurement height.

7. The daily scale corn water stress diagnosis method based on drone technology according to claim 1 is characterized in that: The calculation method of plant height h in each growth period of crops is as follows: The digital surface model DSM0 of the farmland before seedling emergence obtained by the drone is used as the base value, and ground landmarks are selected to calibrate the DSM of each growth period. i , and then get h i =DSM i -DSM0(i=1,2,3,...,i)#(11); After correcting hi with the actual plant height measured on the ground, the final plant height h of crops at each growth stage was obtained using ENVI5.6 software.

Images