Crop growth assessment method, water requirement inversion method and system based on growth stage identification
Patent Information
- Application Number
- CN202611082331.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-07-21
- Publication Date
- 2026-09-29
Smart Images
Figure CN122841963A_ABST
Abstract
Description
Technical Field
[0001] This specification relates to one or more embodiments in the fields of agricultural remote sensing, crop growth monitoring, farmland water management, evapotranspiration estimation and geographic information processing, and in particular to a method for constructing crop growth status based on growth stage identification, a method for inverting water demand and a system thereof. Background Technology
[0002] Crop growth monitoring and water requirement estimation are crucial foundational components of digital agriculture, smart irrigation districts, and agricultural water-saving management. In recent years, the development of high spatial resolution multispectral remote sensing imagery, meteorological reanalysis data, and cloud-based geocomputing platforms has provided new technological conditions for crop identification, phenological monitoring, growth evaluation, and water requirement inversion at both the regional and plot scales. Existing research indicates that extracting crop vegetation indices and biophysical parameters from remote sensing imagery and combining this with meteorological data to estimate crop evapotranspiration is an important technical approach for achieving refined crop monitoring and water management.
[0003] Related technologies include: first, judging crop growth status based on a single vegetation index (such as the Normalized Difference Vegetation Index, NDVI) combined with empirical thresholds; second, identifying crop phenology based on multi-temporal optical remote sensing or radar imagery; and third, calculating theoretical water requirements based on the FAO-56 framework using the crop coefficient (Kc) and reference evapotranspiration (ET0). However, these technologies still have significant limitations under conditions of high-resolution imagery, small plots, complex cropping systems, inconsistent sowing and planting dates, and significant differences between fields. Application content
[0004] This application describes a method for constructing crop growth status based on growth stage identification, a method for retrieving water demand, and a system that can solve the aforementioned technical problems.
[0005] According to the first aspect, a method for crop water requirement inversion based on growth stage identification is provided, comprising the following steps: extracting remote sensing features from each pixel in the target crop distribution area to obtain target crop pixel-level remote sensing feature data; the remote sensing feature data includes one or more of spectral indices and biophysical parameters; constructing a pixel-level decision tree based on the remote sensing feature data to identify the current growth stage of the crop corresponding to each pixel in the target crop distribution area; and assigning a basic crop coefficient based on the current growth stage; performing stage-based grouping processing on the target crop distribution area according to the target crop growth stage identification results, and determining the mean of the remote sensing feature data of the respective growth stages; obtaining the stage-based growth classification results based on the mean of the remote sensing feature data and constructing a growth correction factor; and obtaining the reference crop evapotranspiration of the remote sensing image of the target crop distribution area on the current day or during the target time period based on the basic crop coefficient and the growth correction factor, to obtain the water requirement depth of each pixel in the target crop.
[0006] In some embodiments, obtaining the phased growth classification result based on the mean of the remote sensing feature data includes: calculating the relative deviation of the growth stage of any pixel to be evaluated based on the mean of the remote sensing feature data; obtaining the phased growth classification result based on the relative deviation and constructing a growth correction factor.
[0007] In some embodiments, remote sensing feature extraction is performed on each pixel of the target crop distribution area; including:
[0008] High spatial resolution multispectral remote sensing images are acquired and preprocessed; the multispectral remote sensing images include at least four bands: blue, green, red, and near-infrared; based on the preprocessed multispectral remote sensing images, spectral parameters and biophysical parameters are constructed.
[0009] In some embodiments, based on the remote sensing feature data, a pixel-level decision tree is constructed to identify the current growth stage of the crop corresponding to each pixel in the target crop distribution area, including: constructing a decision tree based on normalized water index, normalized vegetation index, etc.; comparing the pixels in the study area with the pre-established spectral thresholds, and determining the growth stage of each pixel.
[0010] In some embodiments, the determination of the growth correction factor includes:
[0011] If a pixel is classified as having poor growth and a low NDWI, it is determined to be a drought-prone weak seedling, and the growth correction factor F = 1.20.
[0012] If a pixel is classified as having poor growth and a high NDWI, it is determined to be a weak seedling due to excessive moisture or waterlogging, and the growth correction factor F = 0.80.
[0013] If a pixel belongs to the category of good growth, then the growth correction factor F = 1.05;
[0014] If a pixel belongs to the growth range, then the growth correction factor F = 1.00.
[0015] In some embodiments, the method further includes: converting the water depth required by a pixel into the theoretical water requirement per mu (unit of land area), the conversion formula is: Q_mu(p)=ET_theory(p)×0.6667, the unit is m3 / (mu·day), 1 mu = 666.67m2, and 1mm water layer acting on 1 mu area corresponds to 0.6667m3 of water.
[0016] According to the second aspect, a method for constructing a growth index based on fertility stage identification is provided, comprising the following steps: extracting remote sensing features from each pixel in the distribution area of the target crop to obtain pixel-level remote sensing feature data of the target crop; the remote sensing feature data includes one or more of spectral indices and biophysical parameters; constructing a pixel-level decision tree based on the remote sensing feature data to identify the current fertility stage of the crop corresponding to each pixel in the distribution area of the target crop; grouping the distribution area of the target crop into stages according to the fertility stage identification results of the target crop, and determining the mean of the remote sensing feature data of the respective fertility stage; obtaining the staged growth classification results based on the mean of the remote sensing feature data and constructing a growth correction factor.
[0017] According to the third aspect, a crop water requirement retrieval system based on growth stage identification is provided, comprising: an extraction module for extracting spectral parameters corresponding to each pixel in the distribution area of the target crop to obtain pixel-level spectral feature data of the target crop; the spectral feature data of the target crop includes one or more of vegetation index, water index, leaf area index, and green light normalized vegetation index; a determination module for constructing a pixel-level decision tree based on the spectral feature data to identify the current growth stage of the crop corresponding to each pixel in the distribution area of the target crop; and assigning a basic crop coefficient based on the current growth stage; a construction module for performing stage-based grouping processing on the distribution area of the target crop according to the growth stage identification results of the target crop, determining the mean values of spectral parameters and biophysical parameters of the respective growth stages; constructing a growth correction factor based on the mean values of the spectral parameters and biophysical parameters, and based on the stage-based growth classification results; and an acquisition module for acquiring the reference crop evapotranspiration of the remote sensing image of the distribution area of the target crop on the current day or during the target time period, based on the basic crop coefficient and the growth correction factor, to obtain the water requirement depth of each pixel in the target crop.
[0018] According to a fourth aspect, a computer storage medium is provided, on which a computer program is stored, which, when executed by one or more processors, implements the crop water requirement inversion method based on fertility stage identification as described in any of the above embodiments.
[0019] According to a fifth aspect, an electronic device is provided, including a memory and one or more processors, wherein the memory stores a computer program that, when executed by the one or more processors, implements the crop water requirement inversion method based on fertility stage identification as described in any of the above embodiments. Attached Figure Description
[0020] To more clearly illustrate the technical solutions of the embodiments of this application, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0021] Figure 1 This is a schematic flowchart of a crop water requirement inversion method based on growth stage identification provided in an embodiment of this application;
[0022] Figure 2 A schematic diagram of a crop staged classification decision tree (rice) provided for an embodiment of this application;
[0023] Figure 3 A schematic diagram of the GNDVI growth grading standard and the LAI growth grading standard based on relative deviation provided for embodiments of this application;
[0024] Figure 4 A schematic diagram of a CGI comprehensive index grading standard provided in this application embodiment;
[0025] Figure 5 A schematic diagram illustrating the recommended irrigation amounts provided in the embodiments of this application;
[0026] Figure 6 A schematic diagram showing the comparison of original images, growth grading, and theoretical water requirements provided for embodiments of this application;
[0027] Figure 7 A flowchart illustrating another crop water requirement inversion method based on growth stage identification provided in this application embodiment;
[0028] Figure 8 This is a flowchart illustrating a method for constructing a growth index based on reproductive stage identification, as provided in an embodiment of this application.
[0029] Figure 9This is a schematic diagram of a crop water requirement inversion system based on growth stage identification provided in an embodiment of this application. Detailed Implementation
[0030] The solution provided in this specification will now be described with reference to the accompanying drawings.
[0031] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions in the embodiments of this application will be described below with reference to the accompanying drawings.
[0032] In the description of the embodiments of this application, the words "exemplary," "for example," or "for instance" are used to indicate examples, illustrations, or explanations. Any embodiment or design described as "exemplary," "for example," or "for instance" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of the words "exemplary," "for example," or "for instance" is intended to present the relevant concepts in a specific manner.
[0033] In the description of the embodiments of this application, the term "and / or" is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, B existing alone, and A and B existing simultaneously. Furthermore, unless otherwise stated, the term "multiple" means two or more.
[0034] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. The terms "comprising," "including," "having," and their variations all mean "including but not limited to," unless otherwise specifically emphasized.
[0035] Extracting crop vegetation indices and biophysical parameters from remote sensing imagery and combining them with meteorological data to estimate crop evapotranspiration is an important technical approach for achieving refined crop monitoring and water management. However, the performance of existing methods still has significant limitations. Common approaches in the current technology include: first, judging crop growth status based on a single vegetation index (such as the Normalized Difference Vegetation Index, NDVI) combined with empirical thresholds; second, identifying crop phenology based on multi-temporal optical remote sensing or radar imagery; and third, calculating theoretical water requirements based on the FAO-56 framework using the crop coefficient (Kc) and reference evapotranspiration (ET0). These approaches have been widely applied in regional agricultural monitoring, but they still have significant limitations under conditions of high-resolution imagery, small plots, complex cropping systems, inconsistent planting dates, and significant differences between fields.
[0036] First, traditional methods for identifying growth stages based on fixed dates or uniform time-series curves typically assume that target crops in the same region are at similar phenological stages. However, in actual agricultural production, due to differences in sowing or transplanting times, irrigation systems, varieties, and topographical conditions, multiple growth stages often coexist in the same image, making reliable classification difficult using only the imaging date or a single time label. This problem is particularly pronounced for transplanted crops such as rice.
[0037] Second, traditional methods for evaluating crop growth often use an absolute NDVI threshold to classify crops as "excellent," "medium," or "poor." However, the spectral responses of crops at different growth stages vary significantly. For example, NDVI is higher during the vigorous growth stage when the canopy is closed and the leaves are dark green, while NDVI decreases during the senescence or maturity stage when chlorophyll naturally declines. If a uniform threshold is still used, physiological chlorosis can easily be misjudged as poor growth. Furthermore, NDVI exhibits a saturation effect under high coverage conditions and makes it difficult to distinguish from the false lushness phenomenon of "green leaves but sparse canopy."
[0038] Third, existing theoretical water demand estimates are usually based on statistical yearbooks, irrigation district empirical coefficient tables, or single-point observations from meteorological stations, which makes it difficult to reflect the spatial heterogeneity within irrigation districts and between fields. Some methods directly use a uniform crop coefficient Kc or estimate theoretical water demand based on regional averages, failing to reveal the subtle differences in theoretical water requirements of crops at different plots and stages. For high spatial resolution commercial satellite imagery, thermal infrared bands are often lacking, making it difficult to directly apply energy balance models that rely on surface temperature inversion.
[0039] Fourth, while existing research has demonstrated the potential of remote sensing vegetation indices in spatially estimating crop coefficients and characterizing chlorophyll and canopy structure, most methods still rely on single indices, single growth stages, or low-to-medium resolution data, lacking an integrated processing framework that continuously couples "growth period—growth vigor—water requirement." Especially for crops exhibiting early-stage surface water evaporation and later-stage drainage and water control or natural senescence and defoliation, relying solely on a single greenness index is insufficient to balance agronomic interpretability and spatial expressive power.
[0040] Therefore, this application proposes a crop water requirement inversion method based on growth stage identification. It involves a three-stage coupling of target crop growth stage identification, comprehensive growth refinement evaluation, and theoretical water requirement inversion. A pixel-level spectral parameter set of the target crop is constructed based on high-resolution multispectral imagery. Then, the crop parameter library is combined to complete the stage identification. A dynamic growth benchmark is then constructed within each stage. Finally, it is coupled with meteorological evapotranspiration data to output a pixel-level theoretical water requirement map per acre.
[0041] Figure 1 This is a schematic flowchart illustrating a crop water requirement inversion method based on growth stage identification, provided in an embodiment of this application. Figure 1 As shown in the figure, this application proposes a method for crop water requirement inversion based on growth stage identification. Specifically, the method for crop water requirement inversion based on growth stage identification includes the following steps:
[0042] 110: Extract remote sensing features from each pixel in the distribution area of the target crop to obtain pixel-level remote sensing feature data of the target crop; the remote sensing feature data includes one or more of spectral indices and biophysical parameters.
[0043] In one possible implementation, embodiments of this application first preprocess the input remote sensing data. This includes:
[0044] 1. Input Data: High spatial resolution multispectral remote sensing imagery containing four bands—blue, green, red, and near-infrared—is preferred. In a preferred embodiment, Jilin-1 PSH / MSS imagery may be used, but it is not limited to this; any high-resolution imagery capable of providing at least four bands (blue, green, red, and near-infrared) and supporting crop index calculation can be used to implement this invention.
[0045] 2. Study Area: Import the original vector image of the irrigation area or farmland boundary and perform radiometric normalization processing to scale the pixel values to the 0-1 reflectance range; if the study area is covered by multiple images, perform mosaic processing first, and then crop; at the same time, remove black edges, invalid values and non-study areas through masking.
[0046] 4. In a preferred embodiment, the preprocessed image is denoted as img, and the blue, green, red, and near-infrared bands are denoted as B, G, R, and N, respectively.
[0047] Based on the preprocessed multispectral imagery, the following remote sensing feature data were constructed, including spectral indices and biophysical parameters:
[0048] 1. Normalized Difference Vegetation Index (NDVI): NDVI = (NR) / (N+R), used to characterize the overall greenness of vegetation and as an important basis for dividing growth stages. Where N is NIR (near-infrared band), B is Blue (blue light band), and R is Red (red light band).
[0049] 2. Normalized Difference Water Index (NDWI): NDWI = (GN) / (G + N), used to characterize the degree of surface moisture, is a key constraint for distinguishing between "early water presence" and "later yellowing and drainage".
[0050] 3. Green normalized vegetation index (GNDVI): GNDVI = (NG) / (N+G), used to improve the sensitivity to changes in chlorophyll concentration and nitrogen, serving as a "quality" dimension indicator in growth assessment.
[0051] 4. Enhance vegetation index (EVI): EVI = 2.5 × (NR) / (N + 6R - 7.5B + 1), used to reduce the influence of soil background and atmosphere.
[0052] 5. Leaf Area Index (LAI): In a preferred embodiment, the empirical relationship LAI = 3.618 × EVI - 0.118 can be used, and the results can be limited to a reasonable range to characterize population density and canopy biomass.
[0053] 120: Based on the remote sensing feature data, construct a pixel-level decision tree to identify the current growth stage of the crop corresponding to each pixel in the target crop distribution area; and allocate basic crop coefficients based on the current growth stage.
[0054] This application adopts a staged discrimination strategy based on a crop parameter library. For the target crop to be monitored, a corresponding spectral threshold or rule library can be pre-established according to its physiological development law. Then, a pixel-level decision tree is constructed based on indicators such as NDVI and NDWI to distinguish the growth stage of each pixel in the study area. In the case of coexistence of different phenological stages, this application avoids the problem of "cross-stage incomparability" caused by directly using the absolute threshold of NDVI to evaluate growth. The method of combining the crop parameter library with the pixel-level dual-indicator or multi-indicator decision tree is adopted to realize the automatic discrimination of the key growth stages of the target crop. In the preferred embodiment, NDVI and NDWI are used to jointly distinguish the seedling stage, tillering stage, jointing stage, heading stage and grain filling and ripening stage of rice. It should be noted that the initial threshold is determined by statistical analysis of typical samples and can be adaptively corrected according to the image phase, sensor radiation characteristics and sample distribution of the study area. In the preferred embodiment, rice is taken as an example and it is divided into five stages. The preferred threshold settings are shown in Table 1:
[0055] Table 1: Division of Rice Growth Stages
[0056] 1 Seedling stage 0.10 ≤ NDVI < 0.30 and NDWI > -0.15 Newly transplanted / emerged rice seedlings have a significant amount of water surface exposed. 2 Tillering stage 0.30 ≤ NDVI < 0.55 and NDWI > -0.15 The plant begins to expand, still with a distinct water layer background. 3 Propagation period 0.55≤NDVI<0.75 Rapid growth, canopy gradually closes 4 Heading stage NDVI ≥ 0.75 The canopy is at its most lush, with leaf area and biomass nearing their peak. 5 Grouting maturity period 0.25 ≤ NDVI < 0.55 and NDWI ≤ -0.15 The leaves have faded in color and the field has been drained and dried, indicating that the plant is entering the ripening stage.
[0057] One key feature of this invention is the use of water-related indices (NDVIs) to separate stages with similar NDVI values but different agronomical significance for crops exhibiting significant surface moisture changes. Taking rice as an example, by introducing NDVI constraints, the early stage with a "moist background" can be effectively distinguished from the later stage under "drainage or drainage conditions," avoiding confusion caused by relying solely on greenness values. Figure 2 This is a schematic diagram of a crop stage classification decision tree (rice) provided in an embodiment of this application.
[0058] In one possible implementation, the classification result is output as a non-zero stage code, and the background region is masked, retaining only the regions determined to be the growth stage of the target crop. For other crops, the method of this invention can be transferred and applied by replacing or recalibrating the stage division rules.
[0059] This application's embodiments, without relying on complete long-term sequences, can achieve accurate stage-by-stage identification of target crops under single-scene or limited-image high-resolution multispectral remote sensing conditions, thus solving the identification difficulties caused by the coexistence of multiple growth stages in the same image under single-scene or limited-image conditions. Compared with methods that classify by date, by uniform time-series curves, or by a single threshold, this invention achieves more agronomically interpretable stage identification through joint discrimination using a crop parameter library and multiple indicators; for the rice example, it can also effectively distinguish between the early flooding stage and the later receding and ripening stage. By replacing the threshold combinations in the crop parameter library, this invention can also be transferred to other crop scenarios such as corn, wheat, soybeans, and cotton.
[0060] 130: Based on the identification results of the growth stage of the target crop, the distribution area of the target crop is divided into stages and grouped to determine the mean value of the remote sensing feature data of the respective growth stage.
[0061] 140: Based on the mean of the remote sensing feature data, the phased growth classification results are obtained and a growth correction factor is constructed.
[0062] To address the issue of not being able to directly compare the absolute values of NDVI across different phenological stages, this invention further constructs a Comprehensive Growth Index (CGI). This index uses GNDVI to represent leaf quality and LAI to represent population size, and performs dynamic normalization comparisons within each growth stage to generate comparable growth evaluation results across phenological stages.
[0063] 1. Based on the five-stage classification results, determine the mean GNDVI and mean LAI values of all pixels in each stage, namely Mean_GNDVI(s) and Mean_LAI(s), where s represents the stage code.
[0064] 2. For any pixel p to be evaluated, if its stage is s, then its relative deviation is defined as:
[0065] ΔGNDVI(p)=[GNDVI(p)-Mean_GNDVI(s)] / [Mean_GNDVI(s)+ε];
[0066] ΔLAI(p)=[LAI(p)-Mean_LAI(s)] / [Mean_LAI(s)+ε];
[0067] Where ε is a small constant to prevent the denominator from being zero, preferably 0.001.
[0068] 3. The comprehensive growth index is defined as: CGI(p) = 0.5 × ΔGNDVI(p) + 0.5 × ΔLAI(p). Figure 4 This diagram illustrates a CGI comprehensive index grading standard provided in an embodiment of this application. In a preferred embodiment, both the quality and quantity dimensions have a weight of 0.5, but this invention does not limit this and the weights can be adjusted based on regional experience or sample calibration. A comprehensive growth evaluation index capable of simultaneously characterizing leaf quality and population size is constructed.
[0069] 4. CGI classification rules are shown in Table 2.
[0070] Table 2: Grading Rules for Growth Index
[0071] Difference CGI<-0.10 More than 10% lower than the average level of the same period Key inspections and analysis of moisture levels to determine the cause. middle -0.10≤CGI≤0.10 Approximately the average level of the same period Routine monitoring excellent CGI>0.10 More than 10% higher than the average level of the same period Prioritize ensuring high-yield potential
[0072] Figure 3 This diagram illustrates the GNDVI growth grading standard and the LAI growth grading standard based on relative deviation, provided for embodiments of this application. Compared to existing schemes that only use the absolute value of NDVI, this invention eliminates misjudgments caused by maturity fading or physiological differences at different stages through "intra-stage comparison," while introducing LAI to suppress the problem of false lushness caused by simply high greenness but sparse canopy. This logic is applicable not only to rice but also to other target crops with obvious stage-specific spectral variation characteristics.
[0073] 150: Based on the aforementioned basic crop coefficient and growth correction factor, and by obtaining the reference crop evapotranspiration of the remote sensing image of the target crop distribution area on the same day or during the target time period, the water requirement depth of each pixel in the target crop is obtained.
[0074] In this application, GNDVI and LAI are used to characterize leaf quality and population size, respectively. The stage average is calculated according to the growth stage, and then a comprehensive growth index (CGI) is constructed using a relative deviation method, thereby achieving fair evaluation across phenology.
[0075] After obtaining the target crop stage identification results and CGI growth grading results, this invention further performs pixel-level water demand inversion. To avoid confusion with engineering irrigation quotas, canal distribution losses, or actual dispatch volume, this invention defines the final result as "theoretical water demand," that is, the theoretical water requirement required for the crop to maintain normal physiological activities under a given meteorological background, growth stage, and growth status.
[0076] 1. Reference Evapotranspiration (ET0) Acquisition: The reference evapotranspiration of the study area on the day the image was captured is extracted from the meteorological reanalysis dataset. In the preferred embodiment, meteorological variables such as ERA5-Land temperature, wind speed, humidity, and radiation are used, and the reference crop evapotranspiration (ET0) is calculated according to the FAO-56 Penman-Monteith method to obtain the ET0 of the study area on that day. For cases where the study area is small, regional mean statistics can be performed first, and then used as the meteorological background input for the entire region on the same day.
[0077] 2. Assignment of basic crop coefficient Kc for each growth stage: Assign a basic crop coefficient Kc_base to each pixel according to the stage identification results. Different crops can be configured with different stage Kc parameter tables. In the preferred embodiment, the preferred assignment is given for rice with five stages, as shown in Table 3.
[0078] Table 3: Kc parameter configuration for different stages of rice cultivation
[0079] 1 Seedling stage 1.05 Although the vegetation is sparse, the evaporation from the flooded fields is significant. 2 Tillering stage 1.10 As the plant expands, transpiration gradually increases. 3 Propagation period 1.15 Rapid growth increases water demand 4 Heading stage 1.20 The canopy is closed, and transpiration is near its peak. 5 Grouting maturity period 0.90 Leaf senescence and decreased water requirement after drainage and drying of the field
[0080] 3. Construction of Theoretical Water Requirement Correction Factor: Considering that the theoretical water requirement level of pixels at the same stage still varies under different growth and water conditions, this invention further constructs a theoretical water requirement correction factor F(p). This correction factor can be determined jointly based on CGI classification and water-related indices, thus enabling water requirement inversion to take into account meteorological background, crop development status, and current growth status. The preferred rules are as follows:
[0081] (1) If a pixel belongs to poor growth and has a low NDWI, it is judged to be a drought-type weak seedling, and F=1.20 is preferred;
[0082] (2) If a pixel is of poor growth and has a high NDWI, it is determined to be a weak seedling of the wet or waterlogged type, and it is preferable to set F=0.80;
[0083] (3) If the pixel belongs to the category of superior growth, then it is preferable to set F=1.05;
[0084] (4) If the pixel belongs to the growth state, it is preferable to let F=1.00.
[0085] 4. Calculation of Theoretical Water Depth: For each pixel p, the theoretical water depth ET_theory(p) is calculated using the following formula:
[0086] ET_theory(p) = Kc_base(p) × ET0 × F(p), in mm / day.
[0087] 5. Theoretical water requirement per acre conversion: Figure 5 This is a schematic diagram illustrating the recommended irrigation amounts provided in the embodiments of this application. Since local governments and irrigation district management departments are more accustomed to understanding this in terms of area units, this invention further converts the pixel-level theoretical water requirement depth into a pixel-level theoretical water requirement per acre, Q_mu(p):
[0088] Q_mu(p) = ET_theory(p) × 0.6667, where the unit is m3 / (mu·day).
[0089] The conversion factor mentioned above is derived from the following: 1 mu = 666.67 m2, and 1 mm of water layer acting on 1 mu corresponds to 0.6667 m3 of water. This conversion preserves the original pixel-level spatial differences while expressing the results in a unit format more suitable for management departments to understand and apply. For other management needs, it can also be further converted to standardized area expressions such as m3 / (hectare·day) or m3 / (plot·day). By replacing the stage discrimination threshold, basic crop coefficient, and correction factor rules in the crop parameter library, this invention can also be extended to the growth monitoring and theoretical water requirement inversion scenarios of other crops such as corn, wheat, soybeans, and cotton, without changing the basic technical idea.
[0090] 6. Output results: Figure 6This is a schematic diagram showing the original image, growth grading, and theoretical water requirement comparison provided in the embodiments of this application. Based on the stage-allocated basic crop coefficient Kc_base, combined with the reference evapotranspiration ET0 and the theoretical water requirement correction factor F, the pixel-level theoretical water requirement depth and the pixel-level theoretical water requirement per acre are output.
[0091] In this embodiment of the application, under the condition of lacking high-resolution multispectral imagery in the thermal infrared band, remote sensing spectral information, crop growth period information, growth status information and meteorological evapotranspiration information are coupled to construct a pixel-level theoretical water demand inversion model, which is further expressed as theoretical water demand per acre.
[0092] For example, embodiments of this application can run on cloud-based remote sensing computing platforms, such as Google Earth Engine, or on local remote sensing processing platforms or other platforms with multispectral image processing and spatial computing capabilities. It should be noted that embodiments of this application are not limited to a specific platform, a specific satellite brand, or a specific software environment.
[0093] In summary, unlike related technologies that rely solely on dates, NDVI, or empirical calculations based solely on Kc, this embodiment uses a stage-based crop coefficient Kc_base, combined with reference evapotranspiration ET0 and theoretical water requirement correction factor F, to output pixel-level theoretical water requirement depth and pixel-level theoretical water requirement per acre. This solves the problem of difficulty in identifying multiple growth stages of a target crop under single-scene or limited image conditions. Compared to methods that classify by date, uniform time-series curves, or single thresholds, this invention achieves more agronomically interpretable stage identification through a crop parameter library and multi-indicator joint discrimination; for the rice example, it can also effectively distinguish between the early flooding stage and the later receding and ripening stage. By replacing the threshold combinations in the crop parameter library, this invention can also be applied to other crop scenarios such as corn, wheat, soybeans, and cotton. This invention establishes a continuous chain of results—"growth stage—theoretical water requirement"—providing a unified data foundation for subsequent irrigation area water allocation, crop condition evaluation, variable management, and agricultural water-saving assessment. Furthermore, through crop parameter database replacement and stage rule recalibration, this invention can be extended to other target crops while maintaining the basic technical approach, demonstrating good methodological versatility and application promotion value. Even in the absence of high-resolution multispectral remote sensing in the thermal infrared band, pixel-level theoretical water requirement inversion can still be achieved through the coupling of ET0, Kc_base, and correction factor F, overcoming the limitation of relying solely on medium- and low-resolution thermal infrared inversion.
[0094] Furthermore, based on the relationship that 1 mu (unit of land area) corresponds to 0.6667 m3 / mm, the theoretical water requirement depth is converted into m3 / (mu·day) and the result is finally expressed as a pixel-level theoretical water requirement per mu. This not only preserves high-resolution spatial differences, but also more closely matches the language used by local governments, irrigation district management departments, and grassroots farmland managers, expressing "theoretical water requirement per mu", which is convenient for interpretation, display, and application.
[0095] Figure 7 This is a schematic flowchart illustrating another crop water requirement inversion method based on growth stage identification, provided as an embodiment of this application. Figure 7 As shown, a pixel-level set of key parameters for the target crop is constructed based on high-resolution multispectral images. Then, the crop parameter library is combined to complete the stage identification. Within each stage, a growth benchmark is constructed. Finally, it is coupled with meteorological evapotranspiration data to output the pixel-level theoretical water requirement per acre.
[0096] Figure 8 This is a flowchart illustrating a method for constructing a growth index based on reproductive stage identification, as provided in an embodiment of this application. Figure 8 As shown, it includes the following steps:
[0097] Step 810: Extract remote sensing features from each pixel in the distribution area of the target crop to obtain pixel-level remote sensing feature data of the target crop; the remote sensing feature data includes one or more of spectral indices and biophysical parameters.
[0098] Step 820: Based on the remote sensing feature data, construct a pixel-level decision tree to identify the current growth stage of the crop corresponding to each pixel within the target crop distribution area.
[0099] This application's embodiments can distinguish between growth stages with similar NDVI values but different agronomical meanings, especially for crops with significant surface moisture in the early stages but changing moisture conditions later. This avoids misjudgments such as "false green" or "high cover saturation" caused by relying solely on a single greenness index.
[0100] Step 830: Based on the identification results of the growth stage of the target crop, the distribution area of the target crop is divided into stages and groups to determine the mean value of the remote sensing feature data of the respective growth stage.
[0101] Step 840: Based on the mean of the remote sensing feature data, obtain the phased growth classification results and construct the growth correction factor.
[0102] This application's embodiments address the problems of "incomparability across phenological stages" and "high coverage saturation" in traditional NDVI growth evaluation. By using a dual-core approach of GNDVI and LAI, combined with intra-stage relative deviations, this invention can more accurately reflect the overall state of leaf quality and population size, resulting in evaluation results that better align with actual production practices.
[0103] In summary, the embodiments of this application form a comprehensive growth index (CGI) based on the coupling of the relative deviation between GNDVI and LAI, which enables fair evaluation across phenology and plots.
[0104] Figure 9 This is a schematic diagram of a crop water requirement inversion system based on growth stage identification provided in an embodiment of this application. Figure 9 As shown, a crop water requirement inversion system based on growth stage identification includes:
[0105] The extraction module is used to extract the spectral parameters corresponding to each pixel in the distribution area of the target crop, and obtain the pixel-level spectral feature data of the target crop; the spectral feature data of the target crop includes one or more of the following: vegetation index, water index, leaf area index, and green light normalized vegetation index.
[0106] The determination module is used to construct a pixel-level decision tree based on the spectral feature data, identify the current growth stage of the crop corresponding to each pixel in the target crop distribution area, and allocate basic crop coefficients based on the current growth stage.
[0107] The module is used to group the distribution area of the target crop into stages based on the identification results of the growth stage of the target crop, and determine the mean values of the spectral parameters and biophysical parameters of the respective growth stages; based on the mean values of the spectral parameters and biophysical parameters, and based on the staged growth grading results, a growth correction factor is constructed.
[0108] The acquisition module is used to obtain the water requirement depth of each pixel in the target crop based on the basic crop coefficient and the growth correction factor.
[0109] In summary, the embodiments of this application can achieve automatic pixel-level identification of key growth stages of target crops; in the preferred embodiments, automatic pixel-level identification of rice seedling stage, tillering stage, jointing stage, heading stage, and grain-filling maturity stage can be achieved. Finally, pixel-level theoretical water requirement depth map (mm / day) and pixel-level theoretical water requirement per acre map (m3 / (acre·day)) can be generated, providing intuitive spatial water use references for governments and irrigation district management departments. A continuous result chain of "growth stage - growth status - theoretical water requirement" is formed, providing a unified data foundation for subsequent irrigation district water allocation, crop condition evaluation, variable management, and agricultural water-saving assessment; at the same time, through crop parameter library replacement and stage rule recalibration, this invention can be extended to other target crops without changing the basic technical idea, and has good method versatility and application promotion value.
[0110] According to another embodiment, an electronic device is provided, including a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the program, it implements the crop water requirement inversion method based on the identification of the growth stage as described in the above technical solution.
[0111] According to another embodiment, a computer-readable storage medium is also provided, on which a computer program is stored, which, when executed in a computer, causes the computer to perform a crop water requirement inversion method based on the identification of the growth stage.
[0112] Those skilled in the art will recognize that, in one or more of the examples above, the functions described in this invention can be implemented using hardware, software, firmware, or any combination thereof. When implemented in software, these functions can be stored in a computer-readable medium or transmitted as one or more instructions or code on a computer-readable medium.
[0113] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made on the basis of the technical solution of the present invention should be included within the scope of protection of the present invention.
Claims
1. A method for crop water requirement inversion based on growth stage identification, characterized in that, Includes the following steps: Remote sensing features are extracted from each pixel in the distribution area of the target crop to obtain pixel-level remote sensing feature data of the target crop; the remote sensing feature data includes one or more of spectral indices and biophysical parameters. Based on the remote sensing feature data, a pixel-level decision tree is constructed to identify the current growth stage of the crop corresponding to each pixel in the target crop distribution area; and a basic crop coefficient is assigned based on the current growth stage. Based on the identification results of the growth stage of the target crop, the distribution area of the target crop is divided into stages and grouped to determine the mean value of the remote sensing feature data of the respective growth stage. Based on the mean of the remote sensing feature data, the phased growth classification results are obtained and a growth correction factor is constructed. Based on the basic crop coefficient and growth correction factor, and by obtaining the reference crop evapotranspiration of the remote sensing image of the target crop distribution area on the same day or during the target time period, the water requirement depth of each pixel in the target crop is obtained.
2. The method according to claim 1, characterized in that, Based on the mean of the remote sensing feature data, the phased growth classification results are obtained, including: Based on the mean of the remote sensing feature data, for any pixel to be evaluated, the relative deviation of the reproductive stage of the pixel is calculated; Based on the relative deviation, the phased growth classification results are obtained and a growth correction factor is constructed.
3. The method according to any one of claims 1-2, characterized in that, The step of extracting remote sensing features from each pixel in the target crop distribution area includes: Acquire high spatial resolution multispectral remote sensing images and perform preprocessing; the multispectral remote sensing images include at least four bands: blue, green, red, and near-infrared. Based on the preprocessed multispectral remote sensing images, spectral parameters and biophysical parameters are constructed.
4. The method according to any one of claims 1-3, characterized in that, Based on the remote sensing feature data, a pixel-level decision tree is constructed to identify the current growth stage of the crop corresponding to each pixel within the target crop distribution area, including: Decision trees are constructed based on normalized water index and normalized vegetation index; The pixels within the study area are compared with pre-established spectral thresholds to determine the reproductive stage for each pixel.
5. The method according to any one of claims 1-4, characterized in that, The determination of the growth correction factor includes: If a pixel is classified as having poor growth and a low NDWI, it is determined to be a drought-prone weak seedling, and the growth correction factor F = 1.
20. If a pixel is classified as having poor growth and a high NDWI, it is determined to be a weak seedling due to excessive moisture or waterlogging, and the growth correction factor F = 0.
80. If a pixel belongs to the category of good growth, then the growth correction factor F = 1.05; If a pixel belongs to the growth range, then the growth correction factor F = 1.
00.
6. The method according to claim 1, characterized in that, The method further includes: The water depth required by a pixel is converted into the theoretical water requirement per acre. The conversion formula is: Q_mu(p)=ET_theory(p)×0.6667, with the unit being m3 / (acre·day). 1 acre = 666.67m2, and 1mm of water layer corresponds to 0.6667m3 of water volume applied to an area of 1 acre.
7. A method for constructing a growth index based on reproductive stage identification, characterized in that, Includes the following steps: Remote sensing features are extracted from each pixel in the distribution area of the target crop to obtain pixel-level remote sensing feature data of the target crop; the remote sensing feature data includes one or more of spectral indices and biophysical parameters. Based on the remote sensing feature data, a pixel-level decision tree is constructed to identify the current growth stage of the crop corresponding to each pixel in the target crop distribution area. Based on the identification results of the growth stage of the target crop, the distribution area of the target crop is divided into stages and grouped to determine the mean value of the remote sensing feature data of the respective growth stage. Based on the mean of the remote sensing feature data, the phased growth classification results are obtained and a growth correction factor is constructed.
8. A crop water requirement inversion system based on growth stage identification, characterized in that, include: The extraction module is used to extract the spectral parameters corresponding to each pixel in the distribution area of the target crop, and obtain the pixel-level spectral feature data of the target crop. The spectral characteristic data of the target crop include one or more of the following: vegetation index, water index, leaf area index, and green light normalized vegetation index. The determination module is used to construct a pixel-level decision tree based on the spectral feature data, identify the current growth stage of the crop corresponding to each pixel in the target crop distribution area, and allocate basic crop coefficients based on the current growth stage. The module is used to group the distribution area of the target crop into stages based on the identification results of the growth stage of the target crop, and determine the mean values of the spectral parameters and biophysical parameters of the respective growth stages. Based on the mean values of the spectral and biophysical parameters, and based on the phased growth grading results, a growth correction factor is constructed. The acquisition module is used to obtain the reference crop evapotranspiration of the remote sensing image of the target crop distribution area on the same day or during the target time period based on the basic crop coefficient and growth correction factor, and to obtain the water requirement depth of each pixel in the target crop.
9. An electronic device, characterized in that, It includes a memory and one or more processors, wherein the memory stores a computer program that, when executed by the one or more processors, implements a crop water requirement inversion method based on the identification of growth stages as described in any one of claims 1 to 6.
10. A computer storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by one or more processors, implements a crop water requirement inversion method based on fertility stage identification as described in any one of claims 1 to 6.