A wheat seedling emotion classification method and system based on a fitting model

Abstract

The application provides a wheat seedling condition classification method and system based on a fitting model, receives MODIS data of a winter wheat before overwintering in a preset time period through a satellite remote sensing receiving system; carries out scaling, geometric correction, projection and coordinate system conversion processing on the MODIS data; collects and analyzes ground data; respectively calculates the area ratio of each type of seedling in the seedling condition investigation data to a preset sowing area to obtain the percentage of each type of seedling in each planting area in the preset time period; selects a normalized vegetation index NDVI as a remote sensing evaluation index to construct a remote sensing monitoring model; analyzes the MODIS data through a multiple regression analysis mode; and outputs a fitting result after analysis. The application realizes dynamic and quantitative determination of a satellite remote sensing monitoring winter wheat seedling condition classification threshold. The wheat seedling condition classification based on the fitting model is more accurate, and the accuracy of the satellite remote sensing monitoring of the seedling condition classification is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of wheat seedling condition analysis, and in particular to a wheat seedling condition classification method and system based on a fitting model. Background Art

[0002] As a crop with a wide distribution, large cultivated area, and large trade volume worldwide, wheat has a significant impact on food supply. For agricultural production and management departments, timely, accurate, and objective information on winter wheat growth is crucial for agricultural decision-making and management. Satellite remote sensing data, with its wide coverage, large amount of information, fast update cycle, and strong currency, has become a key tool for agricultural informatization. Classification of winter wheat seedling condition during the wintering and greening stages is a key parameter used by agricultural production departments to measure the growth of winter wheat that year. It serves as a crucial basis for guiding wheat production activities and is crucial for strengthening crop production management and formulating scientific and rational agricultural management practices.

[0003] At present, both in scientific research and in agricultural management and decision-making, the horizontal and vertical quantitative comparison of wheat seedling conditions has always been a difficult problem to be solved. Regarding the quantitative calculation problem of wheat seedling diagnosis and analysis, due to insufficient in-depth research, there is still a lack of a unified and effective calculation method. Moreover, in the process of wheat seedling classification and evaluation, some qualitative analysis or subjective judgment elements are mixed in, and the results obtained have a certain deviation from the actual situation. In the existing technology, when managing agriculture, experts are organized to conduct wheat seedling classification and evaluation through field sampling surveys at the beginning of the wintering period each year. Although the survey results can give the percentages of vigorous seedlings, first-class seedlings, second-class seedlings, and third-class seedlings in various cities across the province, they cannot provide spatial distribution information of each type of wheat seedling, making it difficult for agricultural production units to carry out targeted farmland management measures. Summary of the Invention

[0004] In order to overcome the deficiencies in the above-mentioned prior art, the present invention provides a wheat seedling classification method based on a fitting model. The method utilizes long-term satellite data and seedling survey data of the agricultural department in the corresponding period, and uses vegetation inversion from satellite remote sensing to match the agricultural survey data, thereby obtaining a seedling classification statistical inversion model with a strong correlation with the agricultural survey data, thereby improving the accuracy of satellite remote sensing monitoring of seedling classification.

[0005] Wheat seedling condition assessment methods based on fitting models include:

[0006] Step 1: Receive MODIS data of winter wheat before wintering within a preset time period through a satellite remote sensing receiving system;

[0007] Step 2: calibrate, geometrically correct, project and convert the MODIS data into a coordinate system;

[0008] Step 3: Collect and analyze ground data;

[0009] Calculate the ratio of the area of ​​vigorous seedlings, first-class seedlings, second-class seedlings, and third-class seedlings in the seedling condition inspection data to the preset sowing area to obtain the percentage of vigorous seedlings, first-class seedlings, second-class seedlings, and third-class seedlings in each planting area within the preset time period;

[0010] Step 4: Select the Normalized Difference Vegetation Index (NDVI) as the remote sensing evaluation indicator and construct a remote sensing monitoring model;

[0011] Step 5: Analyze MODIS data through multiple regression analysis;

[0012] Step 6: Output the fitting results after analysis.

[0013] It should be further explained that in step 1, the MODIS data received are updated at least twice a day during the day and twice at night.

[0014] It should be further explained that step 2 also includes:

[0015] (1) Unpack MODIS data, interpret the CCSDS format of the data, extract time, scan line, various field of view data frames, lost packets, telemetry data, engineering data, satellite attitude information, ephemeris information, and detection data of various resolutions, and store them in layers and categories in HDF format;

[0016] (2) Calculate the geographic latitude and longitude of each detection data based on MODIS data and store it in HDF format;

[0017] (3) Calculate the reflectivity or radiation value of each channel detection data based on the satellite calibration coefficient sent with the data and store it in HDF format.

[0018] It should be further explained that step 4 also includes: calculating the Normalized Difference Vegetation Index (NDVI) based on single-time satellite data,

[0019]

[0020] Where:

[0021] I NDVI Monitor the NDVI of wheat pixels for a satellite at a single time;

[0022] R NIR is the reflectivity of the pixel in the near-infrared band;

[0023] R RED is the reflectivity of the pixel in the red light band;

[0024] Based on the normalized vegetation index of each time period within a certain time period, the maximum value of the vegetation index of the same pixel is selected as the value of the pixel after multiple time synthesis for calculation. The calculation method is as follows:

[0025] I NDVI (i)=max(I NDVI (i,1),I NDVI (i,2),…,I NDVI (i,t)) (2)

[0026] Where:

[0027] I NDVI (i) is the NDVI after synthesis of the i-th winter wheat pixel;

[0028] i is the pixel number of winter wheat in the region;

[0029] I NDVI (i, t) is the NDVI of the i-th winter wheat pixel at time t;

[0030] is the total number of observation times of the pixel in a given observation period;

[0031] Calculate the mean NDVI, which is the average value of the maximum NDVI values ​​of all winter wheat pixels in the region. The calculation method is as shown in formula (3);

[0032]

[0033] Where:

[0034] ---Regional NDVI mean;

[0035] m---the total number of winter wheat pixels in the area;

[0036] i---sequence number of winter wheat pixel in the region;

[0037] I i ---The NDVI after synthesis of the i-th winter wheat pixel in the region;

[0038] r---region code.

[0039] It should be further explained that the multiple regression analysis methods in step 5 include:

[0040] In the linear regression model, the standard deviation of the NDVI data of all pixels in a certain area is set as a variable to perform multivariate fitting;

[0041] The multiple regression relationship model is formula (5);

[0042]

[0043] Where:

[0044] T is the NDVI critical value for wheat seedling classification in a certain area. The critical values ​​for classification between vigorous seedlings and first-class seedlings, first-class seedlings and second-class seedlings, and second-class seedlings and third-class seedlings are calculated separately and recorded as T 0 / 1 ,T 1 / 2 and T 2 / 3 ;

[0045] is the average NDVI value of a certain area;

[0046] σ is the standard deviation of NDVI in a certain area;

[0047] a, b, c are regression coefficients.

[0048] It should be further explained that in the method, the critical values ​​of vigorous seedlings and first-class seedlings, the critical values ​​of first-class seedlings and second-class seedlings, and the critical values ​​of second-class seedlings and third-class seedlings are set respectively;

[0049] The critical value between vigorous seedlings and first-class seedlings is: R 2 : 0.7839;

[0050] Critical values ​​for Class I and Class II seedlings: R 2 : 0.8543;

[0051] Critical values ​​for Class II and Class III seedlings: R 2 :0.7976.

[0052] The present invention also provides a wheat seedling condition assessment system based on a fitting model, the system comprising: a winter wheat pre-wintering data acquisition module, a MODIS data preprocessing module, a ground data analysis module, a remote sensing monitoring module, a data analysis module, and a result output module;

[0053] The winter wheat pre-wintering data acquisition module is used to receive MODIS data of the winter wheat pre-wintering period within a preset time period through a satellite remote sensing receiving system;

[0054] MODIS data preprocessing module is used to perform calibration, geometric correction, projection and coordinate system conversion on MODIS data;

[0055] The ground data analysis module is used to collect and analyze ground data; calculate the ratio of the area of ​​vigorous seedlings, first-class seedlings, second-class seedlings, and third-class seedlings in the seedling condition inspection data to the preset sowing area, and calculate the percentage of vigorous seedlings, first-class seedlings, second-class seedlings, and third-class seedlings in each planting area within a preset time period;

[0056] Remote sensing monitoring module, used to select the Normalized Difference Vegetation Index (NDVI) as a remote sensing evaluation indicator and build a remote sensing monitoring model;

[0057] Data analysis module, used to analyze MODIS data through multiple regression analysis;

[0058] The result output module is used to output the fitting results after analysis.

[0059] It can be seen from the above technical solutions that the present invention has the following advantages:

[0060] The system of the present invention utilizes a long-term MODIS satellite data series set over many years and the agricultural department's winter wheat seedling field survey data during the wintering period in the corresponding years. It uses the vegetation index mean value, standard deviation and other statistical quantities of the winter wheat pixel to perform linear or nonlinear fitting with the vegetation index corresponding to the seedling classification field survey data, establishes a threshold algorithm model for satellite remote sensing seedling classification monitoring, calculates and analyzes the seedling classification thresholds in multiple locations, provides an objective and quantitative method for remote sensing monitoring of winter wheat seedling classification, and provides an objective basis and technical support for the refined management of winter wheat production.

[0061] Moreover, the present invention utilizes long-term satellite data and seedling survey data of the corresponding period, and based on the correspondence between the vegetation index inverted by satellite remote sensing and the survey data, obtains a seedling classification statistical inversion model with a strong correlation with the data, thereby improving the accuracy of satellite remote sensing monitoring of seedling classification. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] In order to more clearly illustrate the technical solution of the present invention, the following is a brief introduction to the drawings required for the description. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0063] Figure 1 This is a flow chart of the wheat seedling classification method based on the fitting model;

[0064] Figure 2 This is a composite map of the maximum value of the winter wheat pixel vegetation index (NDVI) in Shandong Province before wintering;

[0065] Figure 3 is the normal distribution curve;

[0066] Figure 4 Schematic diagram of the wheat seedling classification system based on the fitting model. DETAILED DESCRIPTION

[0067] The wheat seedling classification system based on the fitting model provided by the present invention is used in the actual production of winter wheat. In order to facilitate agricultural production management departments to understand the wheat production status of various regions, the system divides the local winter wheat into Class I seedlings, Class II seedlings, Class III seedlings and vigorous seedlings according to the growth and seedling classification indicators of the local winter wheat, and estimates the area and proportion of each seedling class. In the prior art, wheat seedling classification information is mainly obtained by ground sampling surveys. However, under normal circumstances, this method requires a lot of manpower and material support, the information is slow to be timely, and only the seedling classification data of each region can be obtained, but the specific spatial distribution of Class I seedlings, Class II seedlings, Class III seedlings and vigorous seedlings cannot be understood. The present invention uses satellite remote sensing data with the advantages of fast timeliness, objectivity, and a large monitoring range. Satellite remote sensing technology can be used to invert quantitative information on wheat growth in the province. Combined with satellite remote sensing seedling classification index parameters, the wheat seedling classification results and spatial distribution of each region can be obtained. Combined with sampling ground survey verification, it can provide more objective remote sensing monitoring results reflecting the actual situation of wheat growth, providing technical reference for production management departments and decision-making departments. Moreover, remote sensing monitoring indicators such as vegetation index can better reflect the growth status of winter wheat and can be used to evaluate the growth of winter wheat.

[0068] The present invention provides a wheat seedling condition assessment system based on a fitting model, which comprises: a winter wheat pre-wintering data acquisition module, a MODIS data preprocessing module, a ground data analysis module, a remote sensing monitoring module, a data analysis module, and a result output module;

[0069] The winter wheat pre-wintering data acquisition module is used to receive MODIS data of the winter wheat pre-wintering period within a preset time period through a satellite remote sensing receiving system;

[0070] MODIS data preprocessing module is used to perform calibration, geometric correction, projection and coordinate system conversion on MODIS data;

[0071] The ground data analysis module is used to collect and analyze ground data; calculate the ratio of the area of ​​vigorous seedlings, first-class seedlings, second-class seedlings, and third-class seedlings in the seedling condition inspection data to the preset sowing area, and calculate the percentage of vigorous seedlings, first-class seedlings, second-class seedlings, and third-class seedlings in each planting area within a preset time period;

[0072] Remote sensing monitoring module, used to select the Normalized Difference Vegetation Index (NDVI) as a remote sensing evaluation indicator and build a remote sensing monitoring model;

[0073] The data analysis module is used to analyze MODIS data through multiple regression analysis; the result output module is used to output the fitting results after analysis.

[0074] Thus, the system of the present application uses long time series set of MODIS satellite data for many years and winter wheat seedling condition field survey data in the corresponding year of the agricultural sector overwintering period, uses statistical quantities such as the average value and the standard deviation of the vegetation index of the winter wheat pixel, and the linear or nonlinear fitting of the corresponding vegetation index of the seedling condition classification field survey data to establish a threshold algorithm model for satellite remote sensing seedling condition classification monitoring, calculate and analyze the seedling condition classification threshold of multiple places, provide an objective and quantitative method for winter wheat seedling condition classification remote sensing monitoring, and provide an objective basis and technical support for fine management of winter wheat production.

[0075] Moreover, the present application uses long time series satellite data and seedling condition survey data in the corresponding period, and the corresponding relationship between the satellite remote sensing inversion vegetation index and the survey data, so as to obtain a seedling condition classification statistical inversion model with strong data correlation, and improve the accuracy of satellite remote sensing monitoring of seedling condition classification.

[0076] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0077] The wheat seedling condition evaluation method based on the fitting model provided by the present application comprises: Figure 1 As shown in the figure, S101, receiving MODIS data of a winter wheat pre-overwintering period in a preset time period through a satellite remote sensing receiving system;

[0078] Specifically, the satellite data used by the present application is based on the MODIS data received by the satellite remote sensing receiving system. The MODIS data has a wide band range, including 36 bands, and the data spatial resolution includes three scales of 250 meters, 500 meters and 1000 meters. (Table 1: MODIS band distribution characteristics). These data have high practical value for comprehensive research of earth science and research of land, atmosphere and ocean in different categories; in addition, TERRA and AQUA satellites are sun-synchronous polar orbit satellites, TERRA passes over in the morning local time, and AQUA passes over in the afternoon local time.

[0079] The MODIS data on TERRA and AQUA are matched in time update frequency, plus the night overpass data, so that at least 2 times of daytime and 2 times of night update data can be obtained every day for receiving MODIS data. Such data update frequency has very important practical value for real-time earth observation and research of earth system in daily frequency.

[0080] Table 1: MODIS band distribution characteristics

[0081]

[0082]

[0083]

[0084] S102, performing calibration, geometric correction, projection and coordinate system conversion processing on MODIS data;

[0085] In an embodiment of the present invention, MODIS data of winter wheat before wintering for several preset years are collected and organized, and the satellite data are pre-processed by calibration, geometric correction, projection, coordinate system conversion, cloud detection, etc. All raster data and vector data are unified in geographic coordinate system and projection method.

[0086] For EOS satellite MODIS data, IMAPP (International MODIS / AIRS Preprocessing Package) is used, which is available to any user with MODIS direct broadcast data. The basic calculation process of the IMAPP package is:

[0087] PDS (i.e. MODIS Level 0) data => (Unpack program)

[0088] =>MOD01 (i.e. MODIS 1A) data =>(Geolocate positioning program)

[0089] =>MOD03 data =>(Calibrate program)

[0090] =>MOD02 (i.e. MODIS 1B) data

[0091] The corresponding components include:

[0092] (1) Unpack program: interprets the CCSDS format of the PDS file, extracts time, scan lines, various field of view data frames, lost packets, telemetry data, engineering data, satellite attitude information, ephemeris information, and detection data of various resolutions, etc., and stores this information in layers and categories in HDF format.

[0093] (2) Geolocate positioning program: Calculates the geographic latitude and longitude of each detection data based on MOD01 data (such as time, telemetry data, attitude, ephemeris and other information and some auxiliary input data), and stores this information in HDF format.

[0094] (3) Calibrate calibration program: Calculates the reflectivity or radiation value of each channel detection data based on the satellite-borne calibration coefficients sent with the data and some auxiliary input data, and stores it in HDF format.

[0095] S103, collecting and analyzing ground data;

[0096] In the embodiments of the present invention, the criteria for classifying winter wheat seedling conditions before overwintering are often referenced to the various indicators in Table 2. Taking prefecture-level cities as units, the ratio of the area of ​​vigorous seedlings, first-class seedlings, second-class seedlings, and third-class seedlings in the seedling condition investigation results to the total sown area of ​​the city is calculated to obtain the percentages of vigorous seedlings, first-class seedlings, second-class seedlings, and third-class seedlings in each prefecture-level city from 2008 to 2019.

[0097] Table 2 Classification criteria for winter wheat seedling condition before wintering

[0098]

[0099] S104, selecting the Normalized Difference Vegetation Index (NDVI) as a remote sensing evaluation indicator and constructing a remote sensing monitoring model;

[0100] The selection of remote sensing indicators includes the following: The Normalized Difference Vegetation Index (NDVI) is the most commonly used indicator for evaluating vegetation growth. Compared with other vegetation indices, NDVI can better reflect crop growth. NDVI is sensitive to changes in soil background. When vegetation cover is less than 15%, the NDVI of vegetation is slightly greater than that of bare soil. When vegetation cover is between 25% and 80%, NDVI increases nearly linearly with increasing vegetation cover. Furthermore, NDVI eliminates most irradiance variations related to instrument calibration, sun angle, terrain, cloud shadows, and atmospheric conditions, enhancing its responsiveness to vegetation. It is the most widely used of the dozens of vegetation indices currently available. Therefore, the Normalized Difference Vegetation Index (NDVI) was selected as the remote sensing evaluation indicator.

[0101] (1) Single-time normalized difference vegetation index

[0102] The normalized difference vegetation index (NDVI) is calculated for single-time satellite data, see formula (1).

[0103]

[0104] Where:

[0105] I NDVI ---The NDVI of wheat pixels monitored by a satellite at a single time;

[0106] R NIR ---Reflectance of the pixel in the near-infrared band;

[0107] R RED---The reflectivity of the red light band of the pixel.

[0108] (2) NDVI synthesis

[0109] For each single normalized vegetation index within a certain time period, the maximum value of the vegetation index of the same pixel is selected as the value of the pixel after multi-time synthesis, see formula (2).

[0110] I NDVI (i)=max(I NDVI (i,1),I NDVI (i,2),…,I NDVI (i,t))............(2)

[0111] Where:

[0112] I NDVI (i) --- NDVI of the i-th winter wheat pixel after synthesis;

[0113] i---sequence number of winter wheat pixel in the region;

[0114] I NDVI (i,t)---NDVI of the i-th winter wheat pixel at time t;

[0115] t---The total number of observation times for the pixel within a given observation time period.

[0116] (3) NDVI mean

[0117] The mean NDVI is the average value of the maximum NDVI values ​​of all winter wheat pixels in the region, see formula (3).

[0118]

[0119] Where:

[0120] ---Regional NDVI mean;

[0121] m---the total number of winter wheat pixels in the area;

[0122] i---sequence number of winter wheat pixel in the region;

[0123] I i ---The NDVI after synthesis of the i-th winter wheat pixel in the region;

[0124] r---region code.

[0125] Taking Shandong Province before wintering from 2008 to 2019 as an example, the maximum composite values ​​of land surface NDVI in Shandong Province before wintering from 2008 to 2019 were calculated respectively, and winter wheat pixels were extracted using winter wheat distribution vector data of the corresponding years to establish the NDVI data of winter wheat in Shandong Province before wintering from 2008 to 2019. The average NDVI values ​​of winter wheat pixels in Shandong Province from 2008 to 2019 were calculated respectively. Figure 2 As shown in the curve, the NDVI value is between 219.74 and 277.25. The NDVI fluctuated greatly from year to year from 2008 to 2012, and the average NDVI during this period was 250.35; after that, the curve fluctuation became smaller, and the average value from 2013 to 2019 was 271.35.

[0126] Building a remote sensing monitoring model: In biology, certain characteristics of a population, such as plant height and ear length, generally follow a normal distribution. NDVI has gained widespread recognition in inferring vegetation growth, and the NDVI series of satellite remote sensing wheat pixels in a specific region approximates a normal distribution.

[0127] Normal distribution, also known as "normal distribution" or Gaussian distribution, has a bell-shaped curve with low values ​​at both ends and high values ​​in the middle, and is symmetrical on both sides. Because of its bell-shaped curve, people often call it a bell curve. Figure 3 If the random variable X obeys a mathematical expectation of μ and variance of σ 2 Normal distribution, denoted as N(μ, σ 2 The expected value μ of a normal distribution determines its location, and its standard deviation σ determines the amplitude of the distribution. When μ = 0 and σ = 1, the normal distribution is called the standard normal distribution.

[0128] The probability density function of the normal distribution is:

[0129]

[0130] By calculating the NDVI thresholds corresponding to wheat seedling classification percentages from field surveys conducted over 10 years (2008-2017), we found a consistent relationship between the NDVI thresholds for a given region and the mean and standard deviation of the NDVI values ​​for wheat pixels in that region. Therefore, a parametric regression equation can be established between the NDVI thresholds for a given region and the corresponding mean NDVI values, enabling the dynamic determination of the thresholds for winter wheat seedling classification.

[0131] S105. Analyze MODIS data by multiple regression analysis;

[0132] S106. Output the analysis and fitting results.

[0133] The multivariate regression analysis method in step 5 of the present invention includes the following: From the expression of the normal distribution, it can be seen that in addition to the mean, the standard deviation is also a parameter expressing the normal distribution. Therefore, the standard deviation of the NDVI data of all pixels in a certain area can be added as another variable in the regression model to perform a multivariate fit.

[0134] The multiple regression relationship model is formula (5);

[0135]

[0136] Where: T is the NDVI critical value for wheat seedling classification in a certain area; the classification critical values ​​of vigorous seedlings and first-class seedlings, first-class seedlings and second-class seedlings, and second-class seedlings and third-class seedlings are calculated separately, denoted as T 0 / 1 ,T 1 / 2 and T 2 / 3 ;

[0137] is the average NDVI value of a certain area;

[0138] σ is the standard deviation of NDVI in a certain area;

[0139] a, b, c are regression coefficients.

[0140] A multiple regression model was established, and the model parameters, goodness of fit, and F-test results are shown in Table 3.

[0141] Table 3 Multiple regression model parameters, goodness of fit and Significance F

[0142]

[0143]

[0144]

[0145]

[0146] The present invention uses the province-wide data as a whole to establish a unified regression model. The expression is:

[0147] The critical value between vigorous seedlings and first-class seedlings is: R 2 : 0.7839;

[0148] Critical values ​​for Class I and Class II seedlings: R 2 : 0.8543;

[0149] Critical values ​​for Class II and Class III seedlings: R 2 :0.7976.

[0150] To verify the effectiveness of the regression model in estimating wheat seedling classification, the present invention used data from 2018, 2019, and 2020 to validate the regression model. Overall, the gap between the estimated values ​​of the seedling classification model and the ground survey values ​​in 16 prefecture-level cities was within a relatively reasonable range.

[0151] In order to quantitatively describe the validation effect of the fitting model, we can analyze it from the perspective of the error between the fitting estimate and the actual value. First, calculate the percentage of Class I seedlings, Class II seedlings, Class III seedlings, and vigorous seedlings corresponding to the regression model fitting value, and then calculate the absolute value of the absolute error between the above value and the ground survey value. ai |, considering that there are certain differences in the planting area of ​​winter wheat in each prefecture-level city in Shandong Province, the contribution of each prefecture-level city's seedling classification percentage to the province's seedling classification percentage is different. Therefore, when describing the model verification error, the absolute error of each prefecture-level city's seedling classification is multiplied by the area coefficient S of the corresponding year. i (S i =(the planting area of ​​a certain city divided by the planting area of ​​the whole province) as the weighted error |E ai |S i When analyzing the average error of various seedlings in the province, the sum of the weighted errors of the 16 cities should be taken: ∑|E ai |S i .

[0152] In 2018, the average error for all seedling types in the province using binary regression analysis was 3.31%, which was smaller than the 5.08% obtained using univariate regression analysis. In 2019, the average absolute errors for univariate and binary regression analyses were 6.64% and 4.52%, respectively, and in 2020, these values ​​were 5.06% and 4.54%, respectively. This indicates that both univariate and binary regression models provide good fit for seedling classification, but overall, binary regression provides better results than univariate regression.

[0153] From the perspective of the four categories of seedlings (Class I, Class II, Class III, and prosperous seedlings), in 2018, the average values ​​of the fitting errors for each category using univariate regression analysis (16 prefecture-level cities) were 0.34 for Class I, 0.36 for Class II, 0.13 for Class III, and 0.23 for prosperous seedlings. Correspondingly, the medians of the fitting results using binary regression analysis were 0.33 for Class I, 0.29 for Class II, 0.11 for Class III, and 0.12 for prosperous seedlings. This indicates that for Class I, Class II, Class III, and prosperous seedlings, the fitting errors of the binary regression model were smaller than those of the univariate regression model.

[0154] The wheat seedling condition assessment method based on the fitting model provided by the present invention is a unit and algorithm step of each example described in combination with the embodiments disclosed herein, and can be implemented by electronic hardware, computer software or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in terms of function in the above description. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of the present invention.

[0155] Those skilled in the art will appreciate that various aspects of the wheat seedling condition assessment method based on a fitting model provided herein can be implemented as a system, method, or program product. Therefore, various aspects of the present disclosure can be specifically implemented in the following forms: a complete hardware implementation, a complete software implementation (including firmware, microcode, etc.), or an implementation that combines hardware and software aspects, which may be collectively referred to herein as a "circuit," "module," or "system."

[0156] The above description of the disclosed embodiments is intended to enable one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be readily apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention is not limited to the embodiments shown herein but is intended to conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A wheat seedling condition assessment method based on a fitting model, characterized in that: Methods include: Step 1: Receive MODIS data of winter wheat before wintering within a preset time period through a satellite remote sensing receiving system; Step 2: calibrate, geometrically correct, project and convert the MODIS data into a coordinate system; Step 3: Collect and analyze ground data; Calculate the ratio of the area of ​​vigorous seedlings, first-class seedlings, second-class seedlings, and third-class seedlings in the seedling condition inspection data to the preset sowing area to obtain the percentage of vigorous seedlings, first-class seedlings, second-class seedlings, and third-class seedlings in each planting area within the preset time period; Step 4: Select the Normalized Difference Vegetation Index (NDVI) as the remote sensing evaluation indicator and construct a remote sensing monitoring model; The Normalized Difference Vegetation Index (NDVI) is calculated based on single-time satellite data. （1） Where: Monitor the NDVI of wheat pixels for a satellite at a single time; is the reflectivity of the pixel in the near-infrared band; is the reflectivity of the pixel in the red light band; Based on the single normalized vegetation index within a certain time period, the maximum value of the vegetation index of the same pixel is selected as the composite value of the pixel for calculation. The calculation method is as follows: （2） Where: For the The NDVI after the synthesis of winter wheat pixels; is the pixel number of winter wheat in the area; For the Winter wheat pixel NDVI of the hour; is the total number of observation times of the pixel in a given observation period; Calculate the mean NDVI. The mean NDVI is the average value of the maximum NDVI values ​​of all winter wheat pixels in the region. The calculation method is as shown in formula (3). （3） Where: ---Regional NDVI mean; ---Total number of winter wheat pixels in the region; ---sequence number of winter wheat pixel in the region; ---The first in the region The NDVI after the synthesis of winter wheat pixels; ---Area code; Step 5: Analyze MODIS data through multiple regression analysis; Multiple regression analysis methods include: In the linear regression model, the standard deviation of the NDVI data of all pixels in a certain area is set as a variable to perform multivariate fitting; The multiple regression relationship model is formula (5); （5） Where: T is the NDVI critical value for wheat seedling classification in a certain area. The critical values ​​for classification between vigorous seedlings and first-class seedlings, first-class seedlings and second-class seedlings, and second-class seedlings and third-class seedlings are calculated separately and recorded as T 0 / 1 ,T 1 / 2 and T 2 / 3 ; is the average NDVI value of a certain area; is the standard deviation of NDVI in a certain area; a, b, c are regression coefficients; Step 6: Output the fitting results after analysis.

2. The wheat seedling condition assessment method based on the fitting model according to claim 1, wherein In step 1, the MODIS data received is updated at least twice a day during the day and twice a day at night.

3. The wheat seedling condition assessment method based on the fitting model according to claim 1, wherein Step 2 also includes: (1) Unpack MODIS data, interpret the CCSDS format of the data, extract time, scan line, various field of view data frames, lost packets, telemetry data, engineering data, satellite attitude information, ephemeris information and detection data of various resolutions, and store them in layers and categories in HDF format; (2) Calculate the geographic latitude and longitude of each detection data based on MODIS data and store it in HDF format; (3) Calculate the reflectivity or radiation value of each channel detection data based on the satellite calibration coefficient sent with the data and store it in HDF format.

4. The wheat seedling condition assessment method based on the fitting model according to claim 1, wherein In the method, the critical values ​​of prosperous seedlings and first-class seedlings, the critical values ​​of first-class seedlings and second-class seedlings, and the critical values ​​of second-class seedlings and third-class seedlings are set respectively; The critical value between vigorous seedlings and first-class seedlings is: T 0 / 1 , R 2 : 0.7839; Critical value between Class I and Class II seedlings: T 1 / 2 , R 2 : 0.8543; Critical value between Class II and Class III seedlings: T 2 / 3 , R 2 :0.7976.

5. A wheat seedling condition assessment system based on a fitting model, characterized in that: The system adopts the wheat seedling condition assessment method based on the fitting model as described in any one of claims 1 to 4; The system includes: winter wheat pre-wintering data acquisition module, MODIS data preprocessing module, ground data analysis module, remote sensing monitoring module, data analysis module and result output module; The winter wheat pre-wintering data acquisition module is used to receive MODIS data of the winter wheat pre-wintering period within a preset time period through a satellite remote sensing receiving system; MODIS data preprocessing module is used to perform calibration, geometric correction, projection and coordinate system conversion on MODIS data; The ground data analysis module is used to collect and analyze ground data; calculate the ratio of the area of ​​vigorous seedlings, first-class seedlings, second-class seedlings, and third-class seedlings in the seedling condition inspection data to the preset sowing area, and calculate the percentage of vigorous seedlings, first-class seedlings, second-class seedlings, and third-class seedlings in each planting area within a preset time period; Remote sensing monitoring module, used to select the Normalized Difference Vegetation Index (NDVI) as a remote sensing evaluation indicator and build a remote sensing monitoring model; Data analysis module, used to analyze MODIS data through multiple regression analysis; The result output module is used to output the fitting results after analysis.

Images