Statistical data downscaling method for crop yield based on integrated agricultural condition index deviation from mean

Abstract

The application discloses a crop yield statistical data downscaling method based on integrated agricultural condition index standardization deviation from mean, which comprises the following steps: A1: data preparation; A2: agricultural condition parameter inversion; A3: integrated agricultural condition index construction; integrated agricultural condition index of key phenophase is constructed, and the correlation between crop agricultural condition parameters and statistical yield in the key phenophase is analyzed, the integrated agricultural condition index of the key phenophase based on the normalized regression coefficient weight determination method is constructed, and the integrated agricultural condition index of the key phenophase based on the yield correlation is verified in precision; A4: yield statistical data downscaling and precision analysis. In addition to comprehensively considering the correlation between crop agricultural condition parameters in the key growth period and the yield change of crops, the application also considers the influence of spatial heterogeneity between integrated agricultural condition parameters pixels on the yield change of the region at the regional level. The crop growth state spatial heterogeneity expression method of the integrated agricultural condition parameter space-time dynamic weight proposed by the application has important significance for improving the simulation capability of the spatial change of the yield of the crop pixel scale.

Description

Technical Field

[0001] This invention relates to the fields of software and information technology, and in particular to a method for downscaling statistical data on crop yield based on the standardized deviation from the mean of comprehensive agricultural indicators. Background Technology

[0002] Crop statistics (including crop planting area and crop yield per unit area) are a crucial component of core agricultural statistics indicators and two key data elements related to a nation's grain production capacity. As the contribution of crop yield per unit area to grain production increases in my country continues to rise, improving crop yield per unit area has become a significant driving force for enhancing my country's overall grain production capacity. Therefore, accurate crop yield statistics play a vital role in national food security and decision-making by agricultural management departments. Furthermore, with the deepening quantitative research on the impacts of climate change, resource environment, and natural disasters on the grain production system, crop yield statistics have become indispensable basic data in these areas of research.

[0003] However, traditional agricultural statistics on crop yield per unit area are mostly tabular data based on administrative units at various levels. These statistics can only reflect differences between administrative regions, while the spatial differences of elements within administrative regions are masked. They fail to truly reveal the real distribution and spatial variation of crop yield per unit area in geographic space. This is highly detrimental to the comprehensive spatial analysis and data fusion of crop yield statistics with other natural, geographical, and ecological spatial data on a spatial grid, thus greatly restricting the widespread application of crop yield statistics in geoscientific spatial analysis (Khan et al., 2010; Hu Yunfeng et al., 2011). Although the traditional method of directly assigning crop yield statistics to administrative units can obtain spatialized information on crop yield statistics in the form of map patches, the problem of homogeneity of crop yield statistics within map patches still exists. It still cannot resolve the contradictions between the spatial heterogeneity of agricultural production factors and the intra-regional homogeneity of statistical data, and between the landscape scale of geo-ecological processes and the administrative unit scale of statistical data (Liu Zhonghe and Li Baoguo, 2012; Fan Yida et al., 2004). Therefore, with the increasing demands for quantification and precision in modern scientific and technological research, the spatial information of crop yield statistics obtained by traditional methods can no longer meet the precision requirements of research on the impact of climate change, resource environment, and natural disasters on the food production system. There is an urgent need to carry out research on new technologies and methods for spatializing crop yield statistics, so as to solve the current bottleneck problem of obtaining high-precision spatial information of agricultural production statistics.

[0004] Furthermore, with the continuous application of artificial intelligence technology in agricultural remote sensing both domestically and internationally in recent years, many research results have emerged in crop yield estimation techniques based on administrative unit crop yield statistics and machine / deep learning algorithms. However, with the development of science and technology, the yield estimation results at the administrative unit scale can no longer meet the increasingly higher spatial resolution requirements of agricultural management departments for crop yield estimation results, making it urgent to conduct research on pixel-scale large-scale crop yield prediction technology. However, in pixel-scale crop yield prediction and estimation based on machine / deep learning, to ensure the accuracy of the crop yield prediction model, sufficient and complete crop yield sample data are needed as model training data to achieve large-scale, high-precision crop yield prediction at the pixel scale. Since the number of yield samples collected from the ground is generally limited, it is difficult to meet the large sample requirements of machine / deep learning models. Therefore, sample generation has become one of the key bottlenecks restricting the development of yield prediction technology. However, generating gridded spatial distribution information of crop yield from crop yield statistics has become one of the effective and important ways to obtain crop yield samples. Therefore, it is urgent to carry out research on crop yield statistics downscaling technology to meet the development needs of yield sample expansion based on pixel-scale crop yield prediction and estimation technology.

[0005] Compared with existing research on the spatialization of socio-economic statistical data (such as population and GDP), the spatialization of crop yield (yield per unit area and total output) statistical data, especially yield per unit area statistical data, is relatively rare in agricultural statistical data spatialization research. Common methods for spatializing crop yield per unit area statistical data often employ simple direct assignment, i.e., using GIS to link crop yield per unit area statistical data with administrative division data to form vector data, and then converting it into raster data of a certain scale. However, this method still cannot completely solve the problem of homogeneity of crop yield statistical data within statistical units. With the development of computer technology, information technology, and spatial technology, scholars at home and abroad have made some progress in the spatialization of crop yield per unit area statistical data in recent years. For example, abroad, You et al. (2006), supported by multi-source data such as land use, agricultural irrigation, crop multiple cropping, arable land suitability, population density, and crop distribution maps, used cross-entropy theory to establish a crop spatial distribution model in Brazil, obtaining spatial distribution information of total crop output statistical data for nine crops including wheat and corn at a 10km * 10km grid scale. Among these, the spatialization of crop yield statistics involves using the proportion of actual cultivated land yield within a grid unit to the regional average potential crop yield to allocate crop yield per unit area across an administrative unit on a spatial grid. However, this method of spatializing crop yield statistics essentially relies primarily on experience or statistical models. In China, Shi Shuqin et al. (2011) used a combination of zoning and statistical modeling to study the spatialization of crop yield statistics. First, they established a multivariate statistical model relationship between crop yield statistics and natural environmental factors such as topography, climate, and soil. Then, by statistically analyzing the values ​​of relevant factors within a standard grid unit (e.g., a 1km * 1km grid), they achieved the quantitative distribution of crop yield statistics within the spatial unit, thus providing a data foundation for integration with geospatial data. However, this method is limited by the simulation accuracy of multivariate regression analysis for the spatial distribution of crop yield, affecting not only the accuracy of the grid-scale crop yield simulation results but also the accuracy of the spatialized crop yield area. Therefore, it is necessary to implement forced weight correction on the crop yield regression simulation results to meet the accuracy requirements of the spatialized crop yield area. Liu et al. (2012) used population density as the dependent variable and land use data as an auxiliary factor to distribute provincial grain output onto a spatial grid, obtaining a 1 km × 1 km grid map of China's grain output in 2000. The spatialization method for provincial grain yield statistics mainly utilized a linear model between provincial population density and grain output per unit cultivated land area to achieve the spatialization of provincial grain yield statistics. Zhao et al. (2023) used multiple linear regression analysis and error correction methods to achieve a large-scale spatial simulation of spring maize yield in Northeast China, based on 13 influencing factors including soil, topography, and climate.

[0006] In summary, current research on the spatialization of crop yield statistics for downscaling is relatively limited, particularly in the area of ​​yield statistics spatialization, which requires further strengthening. Current methods for downscaling yield statistics primarily consider natural environmental conditions such as land use, arable land suitability, topography, climate, soil, and population density, while neglecting crop agronomic parameters that significantly impact yield. Alternatively, they often rely on establishing correlation models between a single agronomic parameter (such as vegetation indices) and statistical yield to achieve the spatialization of regional crop yield statistics for downscaling. However, single agronomic parameters have limited expressive power regarding the formation patterns of crop yield and cannot comprehensively characterize the complex processes of crop growth and development, thus limiting the accuracy of crop yield statistics downscaling methods. Therefore, to overcome these shortcomings, research is needed on downscaling crop yield statistics by constructing Comprehensive Agricultural Indicators (CAIs) using multiple growth stages and agronomic parameters, thereby improving the accuracy and reliability of the spatial expression of yield statistics for downscaling. In addition, existing studies on the spatialization of crop yield per unit area have not considered the impact of spatial heterogeneity in crop growth within administrative units on the formation and yield level of regional crop yield per unit area, which to some extent hinders the further improvement of the accuracy of crop yield spatialization. Summary of the Invention

[0007] The technical problem to be solved by this invention is to provide a method for downscaling crop yield statistics based on standardized deviations from the mean using comprehensive agricultural indicators, which addresses the shortcomings of existing technologies. Considering the temporal changes and spatial heterogeneity of crop growth, this invention conducts research on a method for downscaling crop yield statistics based on standardized deviations from the mean using comprehensive agricultural indicators, in order to further improve the accuracy and level of spatialization of crop yield statistics.

[0008] The technical solution of the present invention is as follows:

[0009] A method for downscaling crop yield statistics based on deviations from the mean of comprehensive agricultural indicators includes the following steps:

[0010] A1: Data Preparation; Prepare ground-measured data, remote sensing data, and auxiliary data; The ground-measured data includes: ground-measured data of crop growth parameters and yield per unit area, crop leaf area index, crop cover, crop aboveground biomass, crop canopy water content, crop canopy chlorophyll, and ground-measured crop yield per unit area; The remote sensing data is Sentinel-2 multispectral satellite data; The auxiliary data includes crop distribution data and vector data of the study area;

[0011] A2: Crop Parameter Inversion; First, crop parameters are inverted based on the SNAP model; then, NPP is obtained based on MODIS and Sentinel-2 remote sensing data; finally, the accuracy of remote sensing inverted crop parameters is verified using ground-measured crop parameters.

[0012] A3: Construction of comprehensive agricultural indicators; Construct comprehensive agricultural indicators for key phenological periods, analyze the correlation between crop agricultural parameters and statistical yields during key phenological periods, construct comprehensive agricultural indicators for key phenological periods based on the normalized regression coefficient weighting method, and verify the accuracy of comprehensive agricultural indicators for key phenological periods based on yield correlation.

[0013] A4: Downscaling and precision analysis of yield statistics;

[0014] A4.1 Calculate the spatiotemporal comprehensive weights of comprehensive agricultural indicators for key phenological periods;

[0015] A 4.2 Calculate the spatial difference index of comprehensive agricultural conditions during the growing season;

[0016] A 4.3 Downscaling of regional crop yield statistics;

[0017] A 4.4 Accuracy Verification.

[0018] In the method described, step A3, specifically analyzing the correlation between crop agronomic parameters during key phenological stages and statistical yield, involves:

[0019] Establish a univariate regression equation between various agricultural parameters during key phenological periods and the statistical yield of crops, and determine the regression coefficients as weighting indicators for the agricultural parameters.

[0020] (5)

[0021] In the formula, Indicates the actual measured yield per unit area on the ground (kg / ha); This represents the relationship between the i-th agricultural condition parameter and the measured yield per unit area during a certain phenological period. The univariate regression coefficient; ... The larger the value, the stronger the impact of agricultural parameters on the measured yield per unit area, and the higher the corresponding weighting coefficient.

[0022] In the method described, step A3, specifically the construction of a comprehensive agricultural condition index for key phenological periods based on the normalized regression coefficient weighting method, involves:

[0023] The method of determining the weight of normalized regression coefficients is used to construct the comprehensive agricultural condition index for key phenological periods. That is, based on obtaining the regression coefficients of the univariate regression equation between each agricultural condition parameter and the statistical yield for each key phenological period, the regression coefficients and the comprehensive agricultural condition parameters are normalized respectively to obtain the corresponding weight coefficients of the normalized agricultural condition parameters for each key phenological period. Then, the above-mentioned normalized regression coefficients are used as weights to construct the comprehensive agricultural condition index for that phenological period.

[0024] (6)

[0025] (7)

[0026] In the formula, Let be the pixel value of the kth pixel of the i-th agricultural condition parameter; The normalized pixel value of the kth pixel of the i-th agricultural condition parameter; Let be the minimum value of the i-th agricultural condition parameter in a pixel. The maximum value of the i-th agricultural condition parameter in a pixel; This represents the i-th agricultural condition parameter after normalization. This represents a comprehensive agricultural condition index for a specific phenological period; m is the total number of agricultural condition parameters.

[0027] In the method described, step A4, specifically step 4.1, includes the following steps:

[0028] A4.1.1 Determine the regression coefficients of comprehensive agricultural indicators during key phenological periods;

[0029] A4.1.2 Calculate the coefficient of variation of comprehensive agricultural indicators;

[0030] A4.1.3 Calculate the spatiotemporal comprehensive weight of key phenological periods.

[0031] In the method described above, step 4.1.1 specifically includes the following steps: establishing a univariate regression equation between the comprehensive agricultural condition index of the key phenological period of crops and the statistical yield per unit area, determining the regression coefficient, and using it as a time weight measurement index for the comprehensive crop index of the phenological period.

[0032] (8)

[0033] In the formula, For the j-th key phenological period, the comprehensive agricultural condition indicator is used. This represents the comprehensive agricultural condition index and the measured yield per unit area during the j-th phenological period. The regression coefficients, The larger the value of , the more significant the impact of the comprehensive agricultural indicators on the measured yield per unit area, and the higher the corresponding time weighting coefficient; c is a constant.

[0034] In the method described, step 4.1.2 specifically includes the following steps: The formula for calculating the coefficient of difference in comprehensive agricultural indicators is as follows:

[0035] (9)

[0036] (10)

[0037] (11)

[0038] In the formula, The k-th pixel value of the comprehensive agricultural condition index for the j-th key phenological period; denoted as the proportion of the k-th pixel in the comprehensive agricultural condition index during the j-th key phenological period in the study area; n represents the total number of crop pixels in the study area. Let be the entropy value of the comprehensive agricultural condition index for the j-th phenological period; The coefficient of variation of comprehensive agricultural indicators during a certain phenological period in the study area.

[0039] In the method described above, step 4.2 in step A4 specifically includes the following steps: First, calculate the standardized deviation from the mean of the comprehensive agricultural condition indicators for each key phenological period at the pixel scale; then, multiply the normalized spatiotemporal dynamic weights of the comprehensive agricultural condition indicators for each key phenological period with the standardized deviation from the mean to obtain the spatial difference index of comprehensive agricultural condition during the growing season.

[0040] The method described in step A4, specifically step 4.2, includes the following steps:

[0041] a) Calculate the regional average of comprehensive agricultural indicators for key phenological periods.

[0042] (15)

[0043] In the formula, Let represent the regional mean of the comprehensive agricultural condition index for the j-th key phenological period (j∈1,2,3,4). represents the value of the k-th pixel of the comprehensive agricultural condition index for the j-th key phenological period; n represents the total number of crop pixels in the area to be spatialized;

[0044] b) Standardized deviation calculation of comprehensive agricultural indicators

[0045] (16)

[0046] In the formula, Let be the pixel value of the k-th pixel of the comprehensive agricultural condition index for the j-th key phenological period; Let $\frac{k}{j}$ be the standardized mean deviation of the $k$-th pixel of the comprehensive agricultural condition index for the $j$-th key phenological period.

[0047] c) Calculation based on the spatial difference index of comprehensive agricultural conditions during the growing season

[0048] (17)

[0049] In the formula, , which is the spatial difference index of comprehensive agricultural conditions during the growing season for the k-th pixel.

[0050] In the method described above, step 4.3 in step A4 specifically includes the following steps: First, the present invention assigns a statistical yield value to each pixel; then, the regional average yield is multiplied by the comprehensive agricultural spatial difference index of the growing season to obtain the yield change corresponding to the comprehensive agricultural spatial difference index of the growing season; then, the regional average yield is added to the corresponding yield change to obtain the spatialization result of yield statistics downscaling, thereby realizing the conversion of corn yield statistics from administrative unit scale to pixel scale.

[0051] The beneficial effects of this invention are as follows:

[0052] (1) This invention proposes a spatial representation method for downscaling maize yield statistics based on spatiotemporally dynamic weighted comprehensive agronomic parameters. This method, based on multi-source remote sensing data, ground-measured agronomic parameters, ground-measured maize yield, and regional yield statistics for the main growth stages, constructs a comprehensive agronomic index using the correlation between agronomic parameters and yield during key phenological periods. Furthermore, considering the spatial heterogeneity of crop growth states, the spatiotemporal dynamic weights of the agronomic index are calculated. The standardized deviation from the mean of the comprehensive agronomic index for each key phenological period is used to characterize the change in pixel-scale yield relative to regional statistical yield, thereby establishing a comprehensive spatial difference index for agronomic conditions during the crop growing season. This completes the research on downscaling methods for regional yield statistics, and this method is of great significance for promoting the spatial development of crop statistical data downscaling.

[0053] (2) This invention utilizes the correlation between key phenological period agricultural parameters and yield per unit area to complete the construction of a comprehensive agricultural index for key phenological periods based on the weight of normalized regression coefficients. The obtained comprehensive agricultural index for key phenological periods provides important basic data for the study of downscaling spatial expression of yield statistics. It also constructs a comprehensive agricultural spatial difference index based on the growing season. By determining the spatiotemporal dynamic weights of comprehensive agricultural parameters, it completes the expression of spatial heterogeneity of crop growth status and realizes the acquisition of the comprehensive agricultural spatial difference index for the growing season. This index is of great significance for the study of downscaling spatial expression methods for maize yield statistics. Finally, based on the comprehensive agricultural spatial difference index, it realizes the acquisition of downscaling spatial information of maize yield statistics and realizes the transformation of crop yield statistics from the regional administrative unit scale to the pixel scale.

[0054] (3) This invention was applied to the spatial representation of maize yield statistics based on spatiotemporal dynamic weighted comprehensive agricultural parameters in Hailun City, Heilongjiang Province, using downscaling. Verification using ground survey data and crop yield statistics showed that the downscaling method for crop yield statistics achieved nearly 100% regional accuracy. In terms of pixel-level accuracy, the difference between the measured yield on the ground and the spatialized pixel yield was significant. 2 The accuracy is 0.82, RMSE is 516.08 kg / ha, NRMSE is 13.15%, and MRE is 4.75%. The above verification accuracy can meet the accuracy requirements of downscaling of crop yield statistics data over a large range, proving that the spatialization method for downscaling crop yield statistics data proposed in this invention is reasonable and feasible, and also indicating that the spatialization method for downscaling yield data proposed in this study has the potential to be applied on a larger scale. Attached Figure Description

[0055] Figure 1 This is a technical roadmap for the present invention;

[0056] Figure 2 Verification results of remote sensing inversion of key agricultural parameters in the study area (2023);

[0057] Figure 3 Correlation between key agronomic parameters during critical phenological stages of maize and crop yield;

[0058] Figure 4 To verify the accuracy of comprehensive agricultural indicators during key phenological stages of maize;

[0059] Figure 5 To verify the downscaling accuracy of maize yield statistics based on ground-measured yield data. Detailed Implementation

[0060] The present invention will be described in detail below with reference to specific embodiments.

[0061] refer to Figure 1Supported by integrated air-ground observation technology, and based on multi-source remote sensing data of the main growth stages, measured ground-based crop condition parameters, measured maize yield per unit area, and regional statistical yield data, this invention constructs a comprehensive crop condition index that fully characterizes crop growth and development by comprehensively considering the correlation between multiple crop condition parameters and yield through a normalized regression coefficient weighting method. Then, considering the impact of the dynamic changes in crop growth status in time and space on the correlation between the comprehensive crop condition index and yield, the spatiotemporal dynamic weights of the comprehensive crop condition index for each key phenological period are calculated. Subsequently, to more accurately describe the changes in pixel yield values ​​relative to the regional mean, this invention uses the pixel-standardized deviation from the mean of the comprehensive crop condition index for key phenological periods as the research basis and the spatiotemporal dynamic weights of the comprehensive crop condition index for key phenological periods as the main reference to construct a comprehensive spatial difference index of crop growth conditions (CSDICGC). Based on this, the crop yield value per pixel within each administrative unit is obtained by combining it with the regional yield mean. Finally, a downscaled spatial representation of regional yield statistics is achieved.

[0062] The following details each step of the method of the present invention:

[0063] Step 1: Data Preparation;

[0064] 1.1 Ground Measured Data

[0065] 1.1.1 Crop growth parameters and measured ground yield data

[0066] To verify the accuracy of remote sensing inversion of crop parameters and spatialization of crop yield, this invention conducted field observations and surveys of crops. The main observation indicators included crop leaf area index (LAI), canopy chlorophyll content (CCC), canopy water content (CWC), fraternal vegetation cover (FVC), aboveground biomass (AGB), crop phenological information, and measured crop yield data. A total of five ground surveys were conducted, and the main observation times and measured parameters of the main crops obtained are shown in Table 1.

[0067] Table 1 Ground observation time and main observation parameters

[0068]

[0069] 1.1.2 Crop Leaf Area Index (LAI)

[0070] This invention utilizes the LAI-2200 canopy analyzer to determine the leaf area index (LAI), a commonly used optical device for measuring LAI. This portable and easy-to-operate device is suitable for field environments. For setting up measurement points in the lower canopy, displacement observation points are established along the vertical row direction at 1 / 4, 1 / 2, and 3 / 4 row spacing, using the row center as a reference. Three independent measurements are performed at each point to eliminate random errors. The instrument calculates the canopy transmittance by measuring the radiation values ​​at the top and bottom of the canopy. Based on the relationship between the radiation attenuation rate and LAI and canopy structure, the LAI is inferred. Finally, the arithmetic mean of the three LAI measurements at each point is taken as the final measured value, ensuring data accuracy and reliability.

[0071] 1.1.3 Crop Cover (FVC)

[0072] This invention utilizes unmanned aerial vehicle (UAV) remote sensing technology to estimate corn vegetation cover (FVC). A quadcopter UAV platform equipped with a visible light sensor was selected, and surface image data was acquired using a vertical downward-looking photography mode at a relative flight altitude of 10 meters. Image acquisition was conducted between 10:00 AM and 2:00 PM local time to minimize terrain shadow interference and ensure the uniformity of lighting conditions in time and space. The UAV is capable of collecting spectral information of the corn canopy, clearly presenting the corn vegetation cover status. The acquired UAV image data is then combined with algorithms such as a pixel-based binary model to segment pixels, distinguishing between vegetated and non-vegetated areas, thereby achieving an accurate estimation of corn vegetation cover. Extensive field testing has verified that this technology can quickly acquire cover information for large areas of cornfields, with accuracy highly meeting the requirements of agricultural production monitoring, providing solid data support for corn growth assessment and field management decisions.

[0073] 1.1.4 Aboveground biomass of crops (AGB)

[0074] In this invention, a destructive sampling method is used to measure aboveground biomass (AGB). In each sampling area, samples are taken from the base of the maize stalk and above, quickly sealed in resealable bags, and then transported to the laboratory for further analysis. In the laboratory, the maize plant stems, leaves, and ears are separated, placed in paper bags, and their fresh weight is accurately measured. After fresh weight measurement, each sample is placed in an oven, first blanched at 105°C for 2 hours, then dried at 75°C until constant weight (mass change less than 0.5%), and the dry weight data is recorded in detail. Finally, combined with the planting density of the sampling area, the biomass per unit area is calculated to obtain accurate maize aboveground biomass data, providing support for the accuracy verification of the key agronomic parameter, Net Primary Productivity (NPP).

[0075] 1.1.5 Crop canopy water content (CWC)

[0076] This invention employs a stratified sampling method to obtain canopy moisture data. The specific steps are as follows: Sampling points (≥3 replicates) are randomly distributed within the target plot, and three representative leaves from the canopy are collected from each point. After collection, the samples are immediately sealed and stored in aluminum foil bags, and the sampling point number and time are recorded. The fresh weight (FW) is determined using an electronic balance (accuracy 0.01 g). The samples are then blanched at 105℃ for 2 hours, and then dried at 75℃ until constant weight to obtain the dry weight (DW). The canopy moisture content (CWC = (FW−DW) / DW) is calculated.

[0077] 1.1.6 Crop Canopy Chlorophyll (CCC)

[0078] In field data collection, this invention selected representative maize plants at each sampling point and collected fresh leaf samples from the upper and middle parts of the plants, with each group repeated twice. Chlorophyll content was determined using a spectrophotometer method. The leaf samples were stored in the dark, flash-frozen in liquid nitrogen, and then extracted with an ethanol-acetone mixture (1:1 volume ratio). The pigment extract was separated by centrifugation, and the absorbance values ​​at wavelengths of 663 nm and 645 nm were measured using a UV-Vis spectrophotometer. Chlorophyll a, b, and total chlorophyll (Chl(a+b)) were calculated based on the Arnon formula. This method was validated using extinction coefficients and standard curves to ensure data accuracy. However, attention must be paid to the degree of sample homogenization and the light-protection treatment of the extract to avoid photodegradation errors. The above CCC measurement results provide ground-based true-value data for subsequent remote sensing inversion of key agricultural parameters (CCC).

[0079] 1.1.7 Actual measured crop yield data

[0080] This invention was implemented on October 2, 2023, in Hailun City, where observation plots were set up to collect data on maize yield at maturity. For the actual yield measurement, 75 sampling points were first selected in the field (53 for verifying the accuracy of comprehensive agricultural indicators, and 22 for verifying the downscaling accuracy of crop yield statistics). At each sampling point, 3-5 maize plants were selected, and the maize ears were naturally air-dried and threshed. The number of ears, the number of kernels per ear, and the weight of 100 kernels were recorded. Additionally, a 3m × 5m area was selected using a measuring tape, and the number of maize plants in each plot was measured to calculate the planting density. The actual yield measurement results were calculated using the planting density, number of ears, and weight of 100 kernels.

[0081] 1.2 Remote Sensing Data Acquisition and Preprocessing

[0082] This invention utilizes Sentinel-2 multispectral satellite data provided by the European Copernicus Data Center (https: / / scihub.copernicus.eu / ), including band information from 13 visible to shortwave infrared bands, with a maximum spatial resolution of 10 m. This constellation consists of two Multispectral Imager (MSI) satellites, which, working together, can achieve a 5-day global revisit cycle. The red-edge bands onboard the Sentinel-2 satellite provide reliable multispectral data support for the dynamic monitoring and simulation of crop growth during key phenological stages. Its high spatiotemporal resolution and the presence of red-edge bands provide an ideal data source for simulating crop growth during key phenological stages (Clerici et al., 2017). This invention processes 2023 Sentinel-2 time-series data using the Google Earth Engine (GEE) cloud platform. In this invention, images of key phenological periods with cloud cover below 30% are selected. For each key phenological period, one high-quality remote sensing image is acquired. The acquired Sentinel-2 images are then processed using the QA band provided by the GEE platform (Hemmerling et al., 2021). Simultaneously, linear interpolation is used to fill in the masked pixels (Kandasamy et al., 2013). The Sentinel-2 remote sensing image bands and resolutions used in this invention are shown in Table 2.

[0083] Table 2. Main Sentinel-2 bands used in this invention

[0084]

[0085] 1.3 Auxiliary Data

[0086] In this invention, auxiliary data used for the spatialization study of crop yield statistics downscaling include crop distribution data and vector data of the study area. The Sentinel-2 cloud mask data for the study area is QA band data provided by the GEE platform. The 30-meter maize crop distribution data for the study area comes from the National Ecological Science Data Center (https: / / doi.org / 10.57760 / sciencedb.08490. https: / / cstr.cn / 31253.11.sciencedb.08490.). This crop distribution dataset is mainly generated based on the Landsat / Sentinel-2 NDVI fusion dataset and the Time-Weighted Dynamic Time Planning (TWDTW) algorithm, using a total of 54,281 samples. After verification, the overall accuracy for 22 provincial-level administrative regions averaged 80.06%; at the county level, the correlation coefficient (R²) between the identified area and the statistical area was [not specified].2 The value is between 0.657 and 0.903, which meets the accuracy requirements of crop masking data for the spatialization of crop yield statistics.

[0087] Step 2: Inversion of agricultural parameters;

[0088] 2.1 Inversion of Agricultural Parameters

[0089] 2.1.1 Inversion of agricultural parameters based on the SNAP model

[0090] This invention utilizes the BiophysicalO module of SNAP (Sentinel Application Platform) to conduct research on the inversion of crop agronomic parameters during key phenological stages. The model effectively combines the advantages of the PROSAIL radiative transfer model and neural network algorithms, achieving high-precision inversion of key agronomic parameters such as canopy chlorophyll, leaf area index, canopy water content, and canopy coverage during multiple phenological stages. Accuracy is verified using ground-measured data, providing foundational data for the construction of comprehensive agronomic parameters and methods for downscaling yield statistics. The main agronomic parameter inversions and image dates are shown in Table 3.

[0091] Table 3. Remote sensing acquisition of agricultural parameters based on Sentinel-2

[0092]

[0093] 2.1.2 NPP Acquisition Based on MODIS and Sentinel-2 Remote Sensing Data

[0094] This invention, based on existing biomass estimation methods, utilizes a combination of MODIS and Sentinel-2 remote sensing data to acquire 10m resolution NPP (Natural Biomass Per Scale). Specifically, by fusing 8-day PSNnet data from MODIS and 8-day maximum composite data of LAI (Local Area Intake) retrieved from Sentinel-2 remote sensing data, a refined estimation of 8-day composite NPP with a spatial resolution of 10m is achieved during the crop growing season. The accuracy is verified using ground-measured aboveground biomass data. The main calculation formulas are as follows:

[0095] (1)

[0096] (2)

[0097] base (3)

[0098] ratio (4)

[0099] In the formula: The MOD17A2 product contains data products every 8 days, which is discussed in this invention. and These are leaf growth respiration and root growth respiration (kg C / 8 day). -1 ); The highest leaf area index over 8 days. This is the conversion factor. According to the MOD17 product user guide (https: / / lpdaac.usgs.gov / products / mod17a2hv006 / ), this factor has a value of 1. base is the leaf growth respiration baseline value, which is 0.30; The ratio is the ratio of root respiration to leaf respiration, and its value is 2.0.

[0100] 2.2 Verification of the accuracy of agricultural parameter inversion

[0101] In this invention, ground-measured agricultural parameters are used to verify the accuracy of remote sensing-derived agricultural parameters. In this invention, the determination coefficient (R²) is selected. 2 The root mean square error (RMSE), normalized root mean square error (NRMSE), and mean relative error (MRE) are used as accuracy verification indicators for agricultural parameters. The specific calculations are shown in formulas (19) to (22) in 4.4.1.

[0102] In this invention, 53 ground survey sampling points were used for the accuracy verification of agricultural parameter results; see section 5.2.4, "Accuracy Verification of Comprehensive Agricultural Indicators During Key Phenological Periods," in the reference results and analysis.

[0103] Step 3: Construction of comprehensive agricultural indicators;

[0104] 3.1 Construction of Comprehensive Agricultural Indicators for Key Phenological Periods

[0105] This invention screened four key phenological stages within the maize growing season: jointing, tasseling, milk stage, and waxy stage. Using canopy chlorophyll content (CCC), leaf area index (LAI), canopy water content (CWC), phenological vegetation cover (FVC), and net primary productivity (NPP) as the main agronomic parameters, different weighting coefficients were assigned to the agronomic parameters at each phenological stage based on their correlation with yield, thus constructing a comprehensive agronomic index for the key phenological stages.

[0106] 3.2 Correlation Analysis between Crop Agronomic Parameters and Statistical Yields during Key Phenological Stages

[0107] A univariate regression equation was established between various agronomic parameters during key phenological stages and the statistical yield of maize, and the regression coefficients were determined as weighting indicators for these agronomic parameters. Specifically, 53 ground survey points were used in this invention for the correlation analysis between crop agronomic parameters during key phenological stages and the statistical yield.

[0108] (5)

[0109] In the formula, Indicates the actual measured yield per unit area on the ground (kg / ha); This represents the relationship between the i-th agricultural condition parameter and the measured yield per unit area during a certain phenological period. The univariate regression coefficient; ... The larger the value, the stronger the impact of agricultural parameters on the measured yield per unit area, and the higher the corresponding weighting coefficient.

[0110] 3.3 Construction of Comprehensive Agricultural Indicators for Key Phenological Stages Based on Normalized Regression Coefficient Weighting Method

[0111] This invention proposes a Normalized Regression Coefficient Weight Determination Method (NRCWDM) for constructing comprehensive agricultural indicators for key phenological periods. Specifically, based on the regression coefficients of the univariate regression equations between each agricultural parameter and the statistical yield for each key phenological period, the regression coefficients and the comprehensive agricultural parameters are normalized to obtain the corresponding weight coefficients for each normalized agricultural parameter. Then, these normalized regression coefficients are used as weights to construct the comprehensive agricultural indicators for that phenological period.

[0112] (6)

[0113] (7)

[0114] In the formula, Let be the pixel value of the kth pixel of the i-th agricultural condition parameter; The normalized pixel value of the kth pixel of the i-th agricultural condition parameter; Let be the minimum value of the i-th agricultural condition parameter in a pixel. Let be the maximum value of the i-th agricultural condition parameter in a pixel. This represents the i-th agricultural condition parameter after normalization. This represents the comprehensive agricultural condition index for a certain phenological period; m is the total number of agricultural condition parameters, and in this invention, m = 5.

[0115] 3.4 Accuracy Verification of Integrated Agricultural Indicators Based on Key Phenological Periods Correlation with Yield

[0116] Since the comprehensive agronomic index in this invention is a comprehensive parameter characterizing crop growth and development constructed using multiple agronomic parameters during the main growth stages, the results of this comprehensive parameter construction cannot be obtained from actual ground measurements. Considering that individual agronomic parameters (such as leaf area index, canopy water content, canopy chlorophyll, canopy coverage, and NPP) all have a certain correlation with crop yield (Yao et al., 2022), the comprehensive agronomic index obtained from the above agronomic parameters should also have a certain correlation with crop yield. Therefore, this invention uses the correlation between the comprehensive agronomic index and the measured crop yield data as the basis for indirectly verifying the accuracy of the comprehensive agronomic index. It is generally believed that the greater the correlation between the comprehensive agronomic index during the key phenological period and the measured crop yield, the stronger the ability of the comprehensive agronomic index to describe changes in crop yield.

[0117] In this invention, the determination coefficient (R²) is selected. 2 The root mean square error (RMSE), normalized root mean square error (NRMSE), and mean relative error (MRE) are used as accuracy verification indicators for the comprehensive agricultural situation indicators. The specific calculations are shown in formulas (19) to (22) in 4.4.1.

[0118] The invention uses 53 ground-level yield survey points for the verification of comprehensive agricultural indicators; see section 5.2.1, "Correlation Analysis between Key Phenological Period Agricultural Parameters and Crop Yield," in the reference results and analysis.

[0119] Step 4: Downscaling and precision analysis of yield statistics;

[0120] 4.1 Calculation of the Spatiotemporal Weights of Comprehensive Agricultural Indicators During Key Phenological Periods

[0121] This invention calculates the time weight by analyzing the correlation between comprehensive agricultural indicators during key phenological periods and yield per unit area. Then, it calculates the spatial weight using the spatial information entropy of the comprehensive agricultural indicators during key phenological periods. Finally, it normalizes the time and spatial weights to obtain the spatiotemporal comprehensive weight of the comprehensive agricultural indicators during key phenological periods.

[0122] 4.1.1 Determination of Regression Coefficients for Comprehensive Agricultural Indicators During Key Phenological Periods

[0123] A univariate regression equation was established between comprehensive agronomic indicators and statistical yields for key phenological stages of crops, and the regression coefficients were determined as time-weighted indicators for the comprehensive crop indicators for that phenological stage. Specifically, 53 ground survey points were used in this invention for the correlation analysis between comprehensive agronomic indicators and statistical yields for single phenological stages.

[0124] (8)

[0125] In the formula, For the j-th key phenological period, the comprehensive agricultural condition indicator is used. This represents the comprehensive agricultural condition index and the measured yield per unit area during the j-th phenological period. The regression coefficients, The larger the value of , the more significant the impact of the comprehensive agricultural indicators on the measured yield per unit area, and the higher the corresponding time weighting coefficient; c is a constant.

[0126] 4.1.2 Calculation of the coefficient of difference for comprehensive agricultural indicators

[0127] To accurately reflect the spatial distribution characteristics of yield per unit area in the study region, the spatial weight coefficient of the comprehensive agricultural condition index is quantified based on the magnitude of the difference coefficient in the Entropy Weight Method (EWM) to accurately characterize the spatial heterogeneity of yield distribution. The difference coefficient represents the spatial variation in crop growth conditions within the region. A larger difference coefficient indicates a greater degree of unevenness in crop growth conditions within the region, and a stronger ability to express the spatial distribution of crop yield per unit area; therefore, it represents a larger spatial weight. Conversely, a smaller difference coefficient represents a smaller spatial weight. The formula for calculating the difference coefficient of the comprehensive agricultural condition index is as follows:

[0128] (9)

[0129] (10)

[0130] (11)

[0131] In the formula, The k-th pixel value of the comprehensive agricultural condition index for the j-th key phenological period; denoted as the proportion of the k-th pixel in the comprehensive agricultural condition index during the j-th key phenological period in the study area; n represents the total number of crop pixels in the study area. Let be the entropy value of the comprehensive agricultural condition index for the j-th phenological period; The coefficient of variation of comprehensive agricultural indicators during a certain phenological period in the study area.

[0132] 4.1.3 Calculation of the Spatiotemporal Comprehensive Weight of Key Phenological Periods

[0133] a) Calculation of time weights for key phenological periods

[0134] (12)

[0135] In the formula, t represents the time weight of the j-th key phenological period; t represents the total number of key phenological periods, and in this invention, t=4.

[0136] b) Calculation of spatial weights for key phenological periods

[0137] Spatial weights are calculated based on the spatial difference coefficient of comprehensive agricultural indicators during key phenological periods. A larger difference coefficient indicates greater unevenness in crop growth within a region, a stronger ability to express the spatial distribution of crop yield, and consequently, a larger spatial weight. The formula is as follows:

[0138] (13)

[0139] In the formula, Let be the spatial weight of the j-th key phenological period.

[0140] c) Calculation of the spatiotemporal comprehensive weight of key phenological periods

[0141] (14)

[0142] In the formula, is the spatiotemporal normalized weight of the j-th key phenological period.

[0143] 4.2 Calculation of Spatial Difference Index of Comprehensive Agricultural Conditions during the Growing Season

[0144] This invention calculates the standardized deviation from the mean of the comprehensive agricultural condition index by utilizing the difference between the comprehensive agricultural condition index during key phenological periods and the regional mean. Based on this, the spatiotemporal comprehensive weights obtained in section 4.1.3 are multiplied together. Then, the results for the four key phenological periods are summed to finally obtain the spatial difference index of comprehensive agricultural condition during the growing season.

[0145] Due to the influence of the crop growing environment, crop growth and development exhibit spatial heterogeneity within a region. To accurately describe the differences in crop growth and development at the pixel scale and more precisely depict the spatial heterogeneity of crop yield distribution, this invention proposes the concept of a comprehensive agricultural condition spatial difference index for the growing season. First, the standardized deviations from the mean of the comprehensive agricultural condition index for each key phenological stage are calculated at the pixel scale. Then, the normalized spatiotemporal dynamic weights of the comprehensive agricultural condition index for each key phenological stage are multiplied by the standardized deviations from the mean to obtain the comprehensive agricultural condition spatial difference index for the growing season.

[0146] a) Calculate the regional average of comprehensive agricultural indicators for key phenological periods.

[0147] (15)

[0148] In the formula, Let represent the regional mean of the comprehensive agricultural condition index for the j-th key phenological period (j∈1,2,3,4). represents the value of the k-th pixel of the comprehensive agricultural condition index for the j-th key phenological period; n represents the total number of crop pixels in the area to be spatialized.

[0149] b) Standardized deviation calculation of comprehensive agricultural indicators

[0150] (16)

[0151] In the formula, Let be the pixel value of the k-th pixel of the comprehensive agricultural condition index for the j-th key phenological period; Let be the standardized mean deviation of the k-th pixel of the comprehensive agricultural condition index for the j-th key phenological period.

[0152] c) Calculation based on the spatial difference index of comprehensive agricultural conditions during the growing season

[0153] (17)

[0154] In the formula, , which is the spatial difference index of comprehensive agricultural conditions during the growing season for the k-th pixel.

[0155] 4.3 Downscaling of regional crop yield statistics

[0156] Since the regional yield statistics are the average yield of the region, i.e., the average yield of all pixels, this invention first assigns a statistical yield value to each pixel. Then, the regional average yield is multiplied by the comprehensive agronomic spatial variation index for the growing season to obtain the yield change corresponding to the index. The regional average yield is then added to the corresponding yield change to obtain the spatialized downscaling result of the yield statistics, thus realizing the conversion of maize yield statistics from the administrative unit scale to the pixel scale. Finally, the accuracy of the crop yield statistics downscaling result is verified using ground yield survey data.

[0157] (18)

[0158] In the formula, The yield per unit area (kg / ha) was statistically calculated for the administrative units of the study area. The pixel yield (kg / ha) of the k-th pixel.

[0159] 4.4 Accuracy Verification

[0160] 4.4.1 Accuracy Verification Based on Ground Measured Data

[0161] In this invention, measured yield data from the ground are used to verify the accuracy of yield statistics downscaling results. In this invention, the determination coefficient (R²) is selected. 2 The root mean square error (RMSE), normalized root mean square error (NRMSE), and mean relative error (MRE) are used as accuracy verification indicators for the downscaling and spatialization results of crop yield statistics. The specific calculations are shown in formulas (19) to (22).

[0162] (19)

[0163] (20)

[0164] (twenty one)

[0165] (twenty two)

[0166] in, Simulated values ​​for verification points of crop yield statistics downscaling results; The ground-measured values ​​corresponding to the verification points of the crop yield statistics downscaled; The average of the simulation results for all verification points; denoted as the mean of all corresponding ground-based measured values; s represents the total number of measured data. In this study, the smaller the RMSE in the accuracy verification index, the higher the accuracy. When NRMSE and MRE are less than 10%, the simulation result is considered excellent; when NRMSE and MRE are greater than 10% but less than 20%, the simulation result is good; when NRMSE and MRE are greater than 20% but less than 30%, the simulation result is moderate; and when NRMSE and MRE are greater than 30%, the simulation result is poor (Michele et al., 2003). When evaluating accuracy, NRMSE is given priority, followed by MRE. Furthermore, the consistency between simulated and observed values ​​can be measured by the coefficient of determination (R²). The closer the R² value is to 1, the better the consistency between simulated and observed values, and vice versa.

[0167] In this invention, 22 ground-based yield survey points are used for verifying the accuracy of crop yield spatialization results.

[0168] 4.4.2 Regional Validation Based on Crop Yield Statistics

[0169] In addition to verifying the pixel-scale accuracy of crop yield statistics by using ground-measured yield data, this invention also verifies the regional-scale accuracy of crop yield statistics by using crop yield statistics. Based on the downscaled results of maize yield statistics obtained by constructing comprehensive agricultural parameters using spatiotemporal dynamic weighting, regional accuracy analysis and evaluation of the spatial distribution of crop yield statistics are conducted using administrative unit crop yield statistics.

[0170] This invention uses statistical data on crop yield per unit area of ​​administrative units as the "true mean" for verifying the spatialization results of crop yield. The regional mean of crop yield per unit area is obtained by downscaling and spatializing the statistical data of maize yield per unit area based on the spatiotemporally dynamically weighted comprehensive agricultural parameters. )and By comparison, the regional precision (w) within the administrative units of the study area is obtained:

[0171] (twenty three)

[0172] (twenty four)

[0173] in, This represents the spatialized regional mean of crop yield per unit area within the administrative units of the study region. The yield value (kg / ha) of the k-th pixel in the downscaling spatialization result of crop yield statistics; n is the total number of crop pixels in the study area; To improve the regional accuracy of spatialization results of crop yield statistics within a given area; The data represents the crop yield per unit area (kg / ha) for administrative units in the study area.

[0174] 5. Results and Analysis

[0175] Supported by integrated air-ground observation technology, this invention proposes a method for downscaling yield statistics based on comprehensive agricultural indicators for key phenological periods. This method utilizes multi-source remote sensing data, measured ground-based crop parameters, measured maize yield per unit area, and regional statistical yield data. First, comprehensive agricultural indicators for key phenological periods are constructed, and their spatiotemporal weights are calculated using a spatiotemporal dynamic weighting method. Then, the standardized deviations from the mean of these indicators are calculated at the pixel scale, and a spatial difference value index for comprehensive agricultural conditions during the growing season is constructed through weighted summation. Finally, combined with regional statistical yield data, the spatial difference value of comprehensive agricultural conditions during the growing season is used to extrapolate crop yield information at the pixel scale, yielding the downscaling results of regional maize yield statistics.

[0176] 5.1 Inversion Results and Accuracy Verification of Key Fertility Period Parameters

[0177] In maize yield prediction studies, agronomic parameters such as chlorophyll content (CCC), canopy cover (FVC), canopy water content (CWC), net primary productivity (NPP), and canopy chlorophyll content are important variables characterizing crop growth. They reflect the physiological state and biomass accumulation of crops from different perspectives and play a crucial role in the spatialization of yield statistics. Simultaneously, the accuracy of remote sensing inversion of these parameters directly affects the reliability and accuracy of yield statistics downscaling results. This invention utilizes the BiophysicalO module of SNAP and Sentinel-2 remote sensing imagery to invert agronomic parameters for crops during key phenological stages, obtaining agronomic parameters such as maize canopy chlorophyll, leaf area index, canopy water content, canopy cover, and NPP during key phenological stages in the study area. This provides important agronomic parameter data for the construction of comprehensive agronomic parameters and research on yield statistics downscaling methods.

[0178] This invention uses key agricultural parameters measured on the ground to verify the accuracy of the above remote sensing inversion results. The verification results are as follows: Figure 2 As shown in Table 4, the leaf area index retrieval accuracy (R²) was 0.66–0.80, and the RMSE was 0.30–0.69; the canopy water content retrieval accuracy (R²) was 0.66–0.78, and the RMSE was 0.01–0.02 g / cm³. 2 The accuracy of maize canopy chlorophyll retrieval was R² 0.68–0.89, and the RMSE was 23.46–78.62 μg / cm³. 2The accuracy of canopy cover retrieval was 0.62-0.79 (R²) and 0.03-0.12 (RMSE); the accuracy of net primary productivity retrieval was 0.66-0.78 (R²) and 14.87-41.24 g C / m². 2 From the perspective of NRMSE index, the NRMSE of the five key agricultural parameters retrieved by this invention ranges from 3.81% to 26.60%, with an average NRMSE of 15.48%. This indicates that the overall accuracy of the agricultural parameter retrieval has reached a high level, and the accuracy of the retrieval results meets the requirements for subsequent applications.

[0179] Table 4. Verification results of remote sensing inversion accuracy of key agricultural parameters

[0180]

[0181] Note: Leaf area index is in m². 2 / m 2 Canopy water content is expressed in g / cm³. 2 Canopy coverage is dimensionless, and canopy chlorophyll is measured in μg / cm³. 2 The unit of NPP is g C / m 2 Additionally, ** indicates that the result is highly significant at the P<0.01 level.

[0182] 5.2 Construction of Comprehensive Agricultural Indicators for Key Phenological Periods

[0183] Crop growth and yield formation is a complex physiological and ecological process, influenced by multiple factors and exhibiting distinct temporal characteristics. Although agronomic parameters retrieved from remote sensing data show some correlation with crop yield, the correlation varies across different phenological stages, making it difficult to comprehensively characterize the spatial distribution of yield using a single agronomic parameter. Compared to a single agronomic parameter, establishing a comprehensive agronomic index by integrating multiple parameters from key phenological stages can more accurately express the spatial distribution patterns of yield, thereby further improving the downscaling accuracy of crop yield statistics.

[0184] 5.2.1 Correlation analysis between key phenological parameters and crop yield

[0185] To determine the correlation between agronomic parameters and yield at different phenological stages, this invention selects key phenological stages such as jointing, tasseling, milk stage, and waxy maturity stage, and establishes a univariate linear regression model between major agronomic parameters (such as crop leaf area index, canopy coverage, canopy chlorophyll, canopy water content, and NPP) and maize yield. The regression equation and accuracy are as follows: Figure 3 As shown. In this invention, the coefficient of determination (R²) is used. 2The root mean square error (RMSE), normalized root mean square error (NRMSE), and mean relative error (MRE) were used as accuracy verification indicators for the univariate regression results. The correlation between comprehensive agronomic parameters during key phenological periods and crop yield is shown in Table 5.

[0186] Table 5. Correlation analysis results between key agronomic parameters during critical phenological stages of maize and crop yield.

[0187]

[0188] Note: Leaf area index is in m². 2 / m 2 Canopy water content is expressed in g / cm³. 2 Canopy coverage is dimensionless, and canopy chlorophyll is measured in μg / cm³. 2 The unit of net primary productivity is g C / m 2 Additionally, ** indicates that the result is highly significant at the P<0.01 level.

[0189] As shown in Table 5, during the jointing stage, the correlation between comprehensive agronomic indicators and yield was NPP > LAI > FVC > CCC > CWC; during the tasseling stage, the correlation was LAI > NPP > CCC > FVC > CWC; during the milk stage, the correlation was LAI > NPP > CCC > FVC > CWC; and during the waxy ripening stage, the correlation was NPP > LAI > FVC > CCC > CWC. Among these, the key phenological parameters NPP and LAI showed the best correlation with maize yield, outperforming the other three parameters, while CWC showed the weakest correlation. Furthermore, throughout the maize growing season, the overall correlation between agronomic parameters such as LAI, NPP, FVC, CCC, and CWC and crop yield showed the pattern of tasseling stage > milky ripening stage > waxy ripening stage > jointing stage, meaning the highest correlation between agronomic parameters and crop yield occurred from the tasseling to the milky ripening stage.

[0190] 5.2.2 Determination of Weights for Crop Condition Parameters During Key Phenological Periods

[0191] To comprehensively compare the correlation between various agronomic parameters and measured yield data at the same level and to construct a comprehensive agronomic index, this invention first normalizes the agronomic parameters within the key phenological stages. Then, based on the coefficients of the univariate regression equation between the normalized agronomic parameters and the measured yield data, weight values ​​are assigned to obtain the weight coefficients of agronomic parameters for each phenological stage. The weight coefficients of agronomic parameters for key phenological stages are shown in Table 6. As can be seen from the table, for different phenological stages (such as jointing stage, tasseling stage, milk stage, and waxy ripening stage), the higher the weight coefficient, the stronger the correlation between the agronomic parameter and the crop yield level at the corresponding phenological stage. Specifically, LAI has the lowest weight of 0.1976 at the jointing stage and reaches the highest weight of 0.2183 at the tasseling stage; as the crop grows, CWC reaches its highest weight of 0.1894 at the milk stage and then decreases at the waxy ripening stage. The highest concentration of Fiber Vitality (FVC) was 0.2317 at the jointing stage, gradually decreasing to 0.1741 at the waxy maturity stage. This reflects the importance of FVC in the early growth stages; its contribution to yield gradually decreases as leaves age in the later stages of growth. The tasseling stage is the most active stage of photosynthesis; therefore, the concentration of CCC (Chemical Oxygen Capacity) reached its highest at tasseling (0.2110) and its lowest at waxy maturity (0.1479). The highest concentration of Non-Polymer Protease (NPP) was 0.2882 at waxy maturity, with lower concentrations at tasseling and milk maturity (0.2108 and 0.2126 respectively). This indicates that NPP's contribution to crop yield formation is more significant in the later stages, especially at waxy maturity.

[0192] Table 6. Results of determining the weights of key agricultural parameters during critical phenological stages of maize.

[0193] Phenological period Leaf Area Index (LAI) Canopy water content (CWC) Canopy coverage (FVC) Canopy chlorophyll (CCC) NPP Propagation period 0.1976 0.1412 0.2317 0.1681 0.2614 Ecchymosis period 0.2183 0.1757 0.1842 0.2110 0.2108 milk maturity period 0.2045 0.1894 0.2079 0.1856 0.2126 wax curing period 0.2175 0.1723 0.1741 0.1479 0.2882

[0194] 5.2.3 Construction of Comprehensive Agricultural Indicators for Key Phenological Periods

[0195] The agricultural parameters for each key phenological period are weighted and summed according to the corresponding weight coefficients in Table 6 to obtain the comprehensive agricultural index for the key phenological period.

[0196] 5.2.4 Accuracy Verification of Comprehensive Agricultural Indicators During Key Phenological Periods

[0197] This invention establishes a univariate linear regression model by utilizing the correlation between comprehensive agricultural indicators during key phenological periods and measured yield data on the ground, and employs the coefficient of determination (R²). 2 The root mean square error (RMSE), normalized root mean square error (NRMSE), and mean relative error (MRE) are used as accuracy verification indicators for the univariate regression results, thereby characterizing the construction accuracy of the comprehensive agricultural indicators for key phenological periods. The correlation analysis results between the comprehensive agricultural indicators for key phenological periods and the measured yield data are as follows: Figure 4 As shown in Table 7.

[0198] As shown in Table 7, during the maize growing season, the correlation between the comprehensive agronomic indicators at key phenological stages and crop yield reached a highly significant level, with the overall correlation showing the order of tasseling stage > milk stage > waxy maturity stage > jointing stage. Among them, the comprehensive agronomic indicators at the tasseling and milk stages showed the highest correlation with the measured yield data, with a coefficient of determination R0. 2 The correlations were 0.68 and 0.63, respectively; RMSEs were 587.39 kg / ha and 626.38 kg / ha, respectively; NRMSEs were 7.29% and 7.77%; and MREs were 6.33% and 6.48%, respectively. The correlation between the comprehensive agronomic indicators at the jointing stage was the weakest. Compared to the correlation between single agronomic parameters (such as leaf area index, canopy water content, canopy chlorophyll, canopy coverage, NPP, etc.) and crop yield at key phenological stages, such as... Figure 3 As shown in Table 5, the coefficient of determination between a single agronomic parameter at the jointing stage and crop yield ranges from 0.13 to 0.33, while the coefficient of determination between the comprehensive agronomic index for key phenological stages constructed in this invention and the measured yield data is 0.37; the coefficient of determination between a single agronomic parameter at the tasseling stage and crop yield ranges from 0.41 to 0.66, while the coefficient of determination between the comprehensive agronomic index for key phenological stages constructed in this invention and the measured yield data is 0.68; the coefficient of determination between a single agronomic parameter at the milk-ripe stage and crop yield ranges from 0.40 to 0.59, while the coefficient of determination between the comprehensive agronomic index for key phenological stages constructed in this invention and the measured yield data is 0.63; the coefficient of determination between a single agronomic parameter at the waxy ripe stage and crop yield ranges from 0.16 to 0.44, while the coefficient of determination between the comprehensive agronomic index for key phenological stages constructed in this invention and the measured yield data is 0.55. It is evident that, compared to using a single agronomic parameter, adopting a comprehensive agronomic index can significantly improve the ability to express the spatial distribution of crop yield, better explain the dynamic changes in yield during key phenological periods, and make it more reasonable to use the established model to characterize yield changes.

[0199] Table 7. Correlation Analysis between Comprehensive Agricultural Indicators and Crop Yield during Key Phenological Stages of Maize

[0200]

[0201] 5.3 Calculation based on the spatial difference index of comprehensive agricultural conditions during the growing season

[0202] Due to changes in the crop growing environment, the spatial distribution of crop yield exhibits significant spatial heterogeneity. To accurately describe the temporal dynamics of crop yield formation and consider the spatial heterogeneity of crop yield distribution in the study area, this invention uses the correlation between comprehensive agronomic indicators and measured yield data as a time-weighted indicator, and the difference coefficient of comprehensive agronomic indicators in the study area as a spatial weighted indicator of yield distribution. Based on this, the standardized deviations from the mean of comprehensive agronomic indicators at each key phenological stage are calculated. Finally, the spatial difference value index of comprehensive agronomic conditions during the growing season is constructed.

[0203] 5.3.1 Calculation of Time Weights for Comprehensive Agricultural Indicators During Key Phenological Periods

[0204] Based on the normalized regression coefficient weighting method, the regression coefficients of the univariate regression equations between the comprehensive agricultural condition indicators of key phenological stages and the measured yield data in Table 7 were normalized and used as the time weights of the comprehensive agricultural condition indicators for each phenological stage. The specific time weights of the comprehensive agricultural condition indicators are shown in Table 8. It can be seen that the time weight is the highest at the milk stage, at 0.3334, indicating that the comprehensive agricultural condition indicators at this growth stage have the strongest correlation with yield, further verifying their key role in the yield formation process.

[0205] Table 8. Time Weights of Comprehensive Agricultural Indicators During Key Phenological Stages of Maize

[0206]

[0207] 5.3.2 Spatial Weight Calculation of Comprehensive Agricultural Indicators During Key Phenological Periods

[0208] To fully consider the spatial heterogeneity of crop yield per unit area in a region, this invention calculates the coefficient of variation based on the establishment of comprehensive agronomic indicators for key phenological stages, and uses the coefficient of variation as a measure of the spatial weight of different key phenological stages. The larger the coefficient of variation, the greater the unevenness of crop growth within the region, the stronger its ability to express the spatial distribution of crop yield per unit area, and the greater its spatial weight. Conversely, the smaller the coefficient of variation, the smaller its spatial weight. Finally, the spatial weights of the comprehensive agronomic indicators for key phenological stages are shown in Table 9. It can be seen that the coefficient of variation for the comprehensive agronomic indicators during the tasseling stage is the largest, reaching 0.0770, indicating that the comprehensive agronomic indicators during the tasseling stage have greater spatial heterogeneity; therefore, the spatial weight of the comprehensive agronomic indicators during the tasseling stage is relatively large, reaching 0.4200. Conversely, the coefficient of variation for the comprehensive agronomic indicators during the jointing stage is the smallest, only 0.0257, indicating that the comprehensive agronomic indicators during the jointing stage have a relatively uniform spatial distribution; therefore, the spatial weight of the comprehensive agronomic indicators during the jointing stage is the smallest, only 0.1401.

[0209] Table 9 Spatial Weights of Comprehensive Agricultural Indicators During Key Phenological Stages of Maize

[0210]

[0211] 5.3.3 Calculation of Spatiotemporal Dynamic Weighting Coefficients for Comprehensive Agricultural Indicators During Key Phenological Periods

[0212] In the invention of downscaling yield statistics, accurately describing the correlation between comprehensive agronomic indicators and yield, as well as the spatial heterogeneity of yield distribution in the study area, is key to improving the spatial accuracy of yield statistics downscaling. This invention comprehensively considers the temporal and spatial weights of comprehensive agronomic indicators during key phenological periods in the study area, ultimately obtaining the spatiotemporal dynamic weight coefficients of comprehensive agronomic indicators during key phenological periods. The specific spatiotemporal dynamic weight coefficients of comprehensive agronomic indicators obtained based on the temporal weights in Table 8 and the spatial weights in Table 9 are shown in Table 10. Calculation of the spatiotemporal dynamic weights of comprehensive agronomic indicators during key phenological periods reveals that the expressive power of comprehensive agronomic indicators for the spatial distribution of yield varies spatiotemporally. Specifically, the spatiotemporal dynamic weights show that the tasseling stage contributes the most to the expression of yield spatial distribution, with a spatiotemporal weight value of 0.3356; the milk-ripe and wax-ripe stages contribute at a moderate level, with spatiotemporal weight values ​​of 0.2871 and 0.2000, respectively; and the jointing stage contributes the least, with a spatiotemporal weight value of 0.1773.

[0213] Table 10 Spatiotemporal dynamic weighting coefficients of comprehensive agricultural indicators during key phenological stages of maize

[0214]

[0215] 5.3.4 Calculation based on the spatial difference index of comprehensive agricultural conditions during the growing season

[0216] Due to the influence of the crop growing environment, crop growth exhibits spatial heterogeneity within a region. To effectively capture the differences in crop growth at the pixel scale and more reasonably characterize the spatial distribution differences in crop yield, this invention constructs a spatiotemporal dynamic weighting system to assign reasonable weights to comprehensive agricultural indicators at different phenological stages. Then, the standardized deviation from the mean of comprehensive agricultural indicators at each key phenological stage is used to obtain the change in pixel-scale yield relative to the regional average statistical yield, finally yielding a spatial difference index of comprehensive agricultural conditions throughout the growing season.

[0217] 5.4 Downscaling and Accuracy Analysis of Regional Maize Yield Statistics

[0218] To accurately describe the spatial distribution of regional yield, this invention utilizes the regional average yield and the comprehensive agricultural spatial difference index for the growing season to obtain the corresponding yield variation. Then, the regional average yield and the yield variation are added together, thereby converting maize yield statistics from the administrative statistical unit scale to the pixel-level spatial scale. Finally, based on ground-level yield survey data from the study area, an accuracy verification analysis is conducted on the downscaled spatial distribution results of crop yield. Specific verification results are as follows: Figure 5 As shown.

[0219] from Figure 5 It can be seen that, by comparing the statistical data of maize yield per unit area in the study area with the regional mean of the downscaled crop yield results in the study area, the regional mean of the spatialized maize yield per unit area in the study area in 2023 was 7791 kg / ha, while the statistical data of maize yield per unit area in Hailun City in 2023 was 7761 kg / ha. This shows that the regional accuracy is 99.6%, close to 100%. Furthermore, by verifying the downscaled spatialized yield statistical data of this invention using ground-measured maize yield survey data, it can be seen that the coefficient of determination R between the ground-measured yield verification points and the correlated pixel values ​​of the downscaled spatial yield results of this invention is [value missing]. 2 The yield per unit area (RMSE) reached 0.82, with an RMSE of 516.08 kg / ha, an NRMSE of 13.15%, and a MRE of 4.75%. This demonstrates the effectiveness and feasibility of the proposed method for downscaling and spatially representing maize yield statistics based on spatiotemporally dynamic weighted comprehensive agricultural parameters. It can obtain high-precision downscaling spatial distribution results of maize yield statistics over a large area. These results not only address the need for spatialized information on crop yield statistics in studies of the impacts of climate change, resource environment, and natural disasters on food production systems, but also provide effective basic spatialized yield information support for the establishment and accuracy verification of large-scale crop yield estimation and prediction models. Furthermore, it has significant application value in the field of precision agriculture.

[0220] It should be understood that those skilled in the art can make improvements or modifications based on the above description, and all such improvements and modifications should fall within the protection scope of the appended claims.

Claims

1. A method for downscaling crop yield statistics based on deviations from the mean of comprehensive agricultural indicators, characterized in that, Includes the following steps: A1: Data preparation; Prepare ground-based measured data, remote sensing data, and auxiliary data; A2: Crop Parameter Inversion; First, crop parameters are inverted based on the SNAP model; then, NPP is obtained based on MODIS and Sentinel-2 remote sensing data; finally, the accuracy of remote sensing inverted crop parameters is verified using ground-measured crop parameters. A3: Construction of comprehensive agricultural indicators; Construct comprehensive agricultural indicators for key phenological periods, analyze the correlation between crop agricultural parameters and statistical yields during key phenological periods, construct comprehensive agricultural indicators for key phenological periods based on the normalized regression coefficient weighting method, and verify the accuracy of comprehensive agricultural indicators for key phenological periods based on yield correlation. A4: Downscaling and accuracy verification of yield statistics; A4.1 Calculate the spatiotemporal comprehensive weights of comprehensive agricultural indicators for key phenological periods; A 4.2 Calculate the spatial difference index of comprehensive agricultural conditions during the growing season; A 4.3 Downscaling of regional crop yield statistics; A 4.4 Accuracy verification; Step A4.2 specifically includes: a) Calculate the regional average of comprehensive agricultural indicators for key phenological periods; （15）； In the formula, Let represent the regional mean of the comprehensive agricultural condition index for the j-th key phenological period, where j∈1,2,3,4; represents the value of the k-th pixel of the comprehensive agricultural condition index for the j-th key phenological period; n represents the total number of crop pixels in the area to be spatialized; b) Calculation of standardized deviations from the mean of comprehensive agricultural indicators; （16）； In the formula, is the pixel value of the kth pixel of the comprehensive agricultural condition index for the jth key phenological period; Let $\frac{k}{j}$ be the standardized mean deviation of the $k$-th pixel of the comprehensive agricultural condition index for the $j$-th key phenological period. c) Calculation based on the spatial difference index of comprehensive agricultural conditions during the growing season; （17） ； In the formula, The growing season comprehensive agricultural spatial difference index for the k-th pixel; Step A4.3 specifically includes: First, assigning a statistical yield value to each pixel; then, multiplying the regional average yield by the comprehensive agricultural spatial difference index during the growing season to obtain the yield change corresponding to the comprehensive agricultural spatial difference index during the growing season; then adding the regional average yield to the corresponding yield change to obtain the spatialized downscaling result of the yield statistics, thereby realizing the conversion of maize yield statistics from the administrative unit scale to the pixel scale; finally, using ground yield survey data to verify the accuracy of the crop yield statistics downscaling result. （18） ； In the formula, To calculate the unit output of administrative units in the study area; This represents the pixel yield of the k-th pixel.

2. The method according to claim 1, characterized in that, In step A3, the analysis of the correlation between crop agronomic parameters and statistical yield during key phenological periods specifically involves: establishing a univariate regression equation between each agronomic parameter during the key phenological period and the statistical yield of the crop, and determining the regression coefficient as a weighting indicator for that agronomic parameter. （5）； In the formula, This represents the actual measured yield per unit area on the ground. This represents the relationship between the i-th agricultural condition parameter and the measured yield per unit area during a certain phenological period. The univariate regression coefficient; ... The larger the value, the stronger the impact of agricultural parameters on the measured yield per unit area, and the higher the corresponding weighting coefficient.

3. The method according to claim 1, characterized in that, In step A3, the construction of the comprehensive agricultural condition index for key phenological periods based on the normalized regression coefficient weight determination method is specifically as follows: the comprehensive agricultural condition index for key phenological periods is constructed using the normalized regression coefficient weight determination method. That is, based on obtaining the regression coefficients of the univariate regression equation between each agricultural condition parameter of the key phenological period and the statistical yield, the regression coefficients and the comprehensive agricultural condition parameters are normalized respectively to obtain the corresponding weight coefficients of the normalized agricultural condition parameters for each key phenological period. Then, the above-mentioned normalized regression coefficients are used as weights to construct the comprehensive agricultural condition index for that phenological period. （6）； （7）； In the formula, Let be the pixel value of the kth pixel of the i-th agricultural condition parameter; The normalized pixel value of the kth pixel of the i-th agricultural condition parameter; Let be the minimum value of the i-th agricultural condition parameter in a pixel. The maximum value of the i-th agricultural condition parameter in a pixel; This represents the i-th agricultural condition parameter after normalization. This represents a comprehensive agricultural condition index for a specific phenological period; m is the total number of agricultural condition parameters.

4. The method according to claim 1, characterized in that, In step A4, step A4.1 specifically includes: A4.1.1 determining the regression coefficient of the comprehensive agricultural condition index for key phenological periods; A4.1.2 calculating the difference coefficient of the comprehensive agricultural condition index; and A4.1.3 calculating the spatiotemporal comprehensive weight of the key phenological periods.

5. The method according to claim 4, characterized in that, Step A4.1.1 specifically includes: establishing a univariate regression equation between the comprehensive agricultural condition index of the key phenological period of crops and the statistical yield per unit area, determining the regression coefficient, and using it as a time weight measurement index for the comprehensive crop index of the phenological period; （8） ； In the formula, For the j-th key phenological period, the comprehensive agricultural condition indicator is used. This represents the comprehensive agricultural condition index and the measured yield per unit area during the j-th phenological period. The regression coefficients, The larger the value of , the more significant the impact of the comprehensive agricultural indicators on the measured yield per unit area, and the higher the corresponding time weighting coefficient; c is a constant.

6. The method according to claim 4, characterized in that, Step A4.1.2 specifically includes: The formula for calculating the coefficient of difference of comprehensive agricultural indicators is as follows: （9）； （10）； （11）； In the formula, The k-th pixel value of the comprehensive agricultural condition index for the j-th key phenological period; denoted as the proportion of the k-th pixel in the comprehensive agricultural condition index during the j-th key phenological period in the study area; n represents the total number of crop pixels in the study area. Let be the entropy value of the comprehensive agricultural condition index for the j-th phenological period; The coefficient of variation of comprehensive agricultural indicators during a certain phenological period in the study area.

Images