Remote sensing multispectral imaging system for agricultural monitoring

Abstract

The invention provides a remote sensing multispectral imaging system for agricultural monitoring, which relates to the technical field of space remote sensing, and realizes accurate selection by indicating an economic crop determination module, analyzing historical soil salinity and water shortage data and selecting sensitive crops; the correlation evaluation module calculates a water shortage index change rate, analyzes correlation between historical data and current data, and improves prediction accuracy; the value range determination module divides soil salinity intervals and standardizes management data in combination with evaluation coefficients; the indication area determination module realizes accurate monitoring by using a multispectral image and enhances representativeness; the humidity evaluation module analyzes local spectral data to quantify the soil humidity; the prediction model construction module predicts the salinity content by using machine learning, and can accurately predict dynamic changes of soil salinity and moisture; the correlation analysis module verifies real-time monitoring data in combination with the prediction model; and the division module optimizes a soil management strategy according to the water shortage index and the humidity coefficient.

Description

Technical Field

[0001] The invention relates to the field of aerospace remote sensing technology, in particular to a remote sensing multi-spectral imaging system for agricultural monitoring. Background Art

[0002] With the development of agricultural production towards refinement and intelligence, traditional ground monitoring methods can no longer meet the requirements of large-scale, fast and accurate data acquisition. Multispectral imaging technology can provide richer and more detailed ground information than single-spectral imaging technology by capturing information from multiple spectral bands, which helps to deeply analyze the specific conditions of soil, water, vegetation and other elements.

[0003] In the prior art, the publication number is CN117274585A, and the name is a semantic segmentation method for agricultural multimodal remote sensing images based on transformer channel feature enhancement, which includes the following steps: S1: divide the agricultural semantic segmentation image data set; S2: define the preprocessing scheme in data reading; S3: define a random weight sampler and a distributed data loader; S4: load data from the data loader, and start feature extraction using a model of an encoder-decoder architecture; S5: use a transformer-based channel enhancement module to enhance the fusion features in the decoder; S6 calculate the model's prediction map xpred, xc for each pixel point in the image; S7: calculate the loss function; S8: repeat steps S2 to S7 until the model training is completed.

[0004] The existing agricultural remote sensing monitoring system has shortcomings in the following aspects: first, traditional methods usually rely on ground sampling and experimental analysis, which is not only time-consuming and laborious, but also difficult to achieve large-scale real-time monitoring; second, the current multispectral imaging system often lacks full use of historical data, making it difficult to accurately predict the dynamic changes of soil salinity and moisture, resulting in insufficient scientificity and effectiveness of agricultural decision-making; in addition, the existing technology has limitations in the correlation analysis of soil and crop physiological status, and cannot fully explore the potential relationship between historical data and current monitoring data;

[0005] The above information disclosed in the above Background section is only for enhancement of understanding of the background of the present disclosure and therefore it may contain information that does not form the prior art that is already known to one of ordinary skill in the art. Summary of the invention

[0006] The object of the present invention is to provide a remote sensing multispectral imaging system for agricultural monitoring to solve the problems raised in the above background technology.

[0007] To achieve the above object, the present invention provides the following technical solutions:

[0008] A remote sensing multispectral imaging system for agricultural monitoring, specifically comprising:

[0009] Indicator cash crop determination module: used to obtain the historical soil salinity change data of the target monitoring area in the previous year, as well as the soil water shortage index corresponding to the historical soil salinity change data; based on the historical soil salinity change data, determine the indicator cash crop that is most sensitive to the soil salinity change in the target monitoring area;

[0010] Correlation evaluation module: used to calculate the change rate of soil water shortage index between two consecutive months in the previous year, and to perform correlation analysis on the historical soil salinity change data and the change rate of soil water shortage index to obtain the correlation evaluation coefficient;

[0011] Value range determination module: used to divide the soil salinity level into several intervals according to the historical soil salinity change data, and determine the value range of the soil water shortage index in each soil salinity level interval by combining the correlation evaluation coefficient and the change rate of the soil water shortage index;

[0012] Indicator area determination module: used to obtain multispectral image data of the target monitoring area and then perform preprocessing, and determine the most representative indicator area in the image data based on historical soil salinity change data and correlation evaluation coefficients, as well as reference multispectral data corresponding to the indicator area. The indicator area is used to plant the determined indicator economic crops, and the indicator area can represent the soil salinity level of the target monitoring area;

[0013] Humidity evaluation module: used to receive the pre-processed reference multispectral data, extract the local spectral data used to evaluate soil moisture in the reference multispectral data, analyze and process the local spectral data, and construct a soil moisture evaluation coefficient. The soil moisture evaluation coefficient is used to evaluate the humidity of the target monitoring area;

[0014] Prediction model building module: used to obtain multispectral data that indicates the one-to-one correspondence between economic crops and historical soil salinity change data, and after data processing, build a machine learning model, and then input the processed multispectral data into the machine learning model for training and verification, and finally generate a salt content prediction model;

[0015] Correlation analysis module: used to substitute the pre-processed reference multispectral data into the salt content prediction model to obtain the prediction result of soil salinity level, and match the prediction result with several divided soil salinity level intervals to determine the soil water shortage index of the current target monitoring area;

[0016] Division module: It is used to build a soil division model by combining the soil water shortage index and soil moisture evaluation coefficient of the current target monitoring area. The model is used to divide the soil moisture demand level to generate farmland management strategies.

[0017] Furthermore, the change rates of soil water shortage indicators between two consecutive months are calculated in turn, including:

[0018] Calculate the change rate of soil moisture content, field water holding capacity, permanent wilting point and soil water tension from each month to the next month in turn; set the current month as j and the next month as j+1, and the change rate formula is as follows:

[0019]

[0020] In formula (1), ΔZB represents the change rate of soil water shortage index, ZBj represents the soil water shortage index value in the jth month, and ZBj +1 represents the soil water shortage index value in month j+1;

[0021] Pair the historical soil salinity change data with the monthly change rates of various soil water stress indicators;

[0022] Calculate the Pearson correlation coefficient between historical soil salinity change data and the change rate of each soil water shortage index;

[0023] The Pearson correlation coefficient calculated for each soil water shortage index forms a correlation evaluation coefficient matrix FJ:

[0024]

[0025] In formula (2), r SMC 、r FC 、r PWP 、r SWT They are the Pearson correlation coefficients between soil salinity change data and the corresponding change rates of soil moisture content, field water holding capacity, permanent wilting point and soil water tension;

[0026] Based on historical soil salinity change data, the soil salinity level is divided into several intervals, including:

[0027] According to the percentage of soil salt content, it is divided into low salt interval (0, 10%), medium salt interval [10%-15%], and high salt interval (15%-100%);

[0028] Combined with the correlation evaluation coefficient matrix FJ and the change rate of each soil water shortage index, the value range of the soil water shortage index in each soil salinity level interval is determined, including:

[0029] The calculation formula for setting the soil water shortage index is Through the linear regression analysis of historical data, the regression coefficient a′ of each soil water shortage index is obtained. ZB and the intercept b′ ZB ; YFp is the salt level;

[0030] The correlation coefficient r in the correlation evaluation coefficient matrix FJ SMC 、r FC 、r PWP 、r SWT After standardization, the following calculation formula is obtained:

[0031]

[0032] w SMC 、w FC 、w PWP 、w SWT They are used as weight coefficients for soil moisture content, field capacity, permanent wilting point and soil water tension respectively;

[0033] The calculation formula for soil moisture content is:

[0034] SMCbh=w SMC ×(a SMC ×YFp+b SMC ); (5)

[0035] In formula (5), a SMC 、b SMC are the regression coefficient and intercept of soil moisture content respectively; YFp takes the endpoint values ​​on both sides of the current salt partition for calculation;

[0036] The calculation formula for field water holding capacity is:

[0037] FCbh=w FC ×(a FC ×YFp+b FC ); (6)

[0038] In formula (6), a FC 、b FC are the regression coefficient and intercept of soil moisture content, respectively;

[0039] The calculation formula for the permanent wilting point is:

[0040] PWPbh=w PWP ×(a PWP ×YFp+b PWP ); (7)

[0041] In formula (7), a PWP 、b PWP are the regression coefficient and intercept of soil moisture content, respectively;

[0042] The calculation formula for the value of soil water tension is:

[0043] SWTbh=wSWT×(a SWT ×YFp+b SWT ); (8)

[0044] In formula (8), a SWT 、b SWT are the regression coefficient and intercept of soil moisture content, respectively.

[0045] Furthermore, a machine learning model is constructed, and the processed multispectral data is input into the machine learning model for training and verification, and finally a salt content prediction model is generated, which specifically includes:

[0046] Extract key spectral features from historical multispectral data, including but not limited to: vegetation index, soil conditioning index, red edge position;

[0047] According to the time series data, the spectral characteristic change rate and soil salinity change rate at each time point are calculated to form the time series characteristics;

[0048] A method based on multi-source data fusion is used to fuse spectral features and soil salinity features; a multi-layer perceptron neural network is used as a data fusion model;

[0049] The extracted spectral features and soil salinity features are input into the MLP neural network for feature fusion;

[0050] Use historical data to train the MLP neural network;

[0051] The performance of the trained model is verified through the validation set, and the prediction accuracy and error of the model are calculated;

[0052] According to the verification results, adjust the parameters of the MLP neural network;

[0053] The optimized MLP neural network model is saved as the final salt content prediction model.

[0054] Furthermore, the pre-processed reference multispectral data is substituted into the salt content prediction model to obtain the predicted result of soil salinity level, which is then matched with several divided soil salinity level intervals to determine the soil water shortage index of the current target monitoring area, including:

[0055] Match the predicted salt content with the pre-divided salt interval to determine the salt level of the current target monitoring area;

[0056] According to the percentage of soil salt content, the predicted salt content is matched with the low salt interval (0, 10%), the medium salt interval [10%-15%], and the high salt interval (15%-100%) to determine the salt interval to which it belongs;

[0057] Based on the confirmed salinity range, the value ranges of soil water shortage indicators including soil moisture content, field water holding capacity, permanent wilting point and soil water tension are determined.

[0058] Furthermore, it is used to combine the soil water shortage index and soil moisture evaluation coefficient of the current target monitoring area to build a soil classification model, which is used to classify the soil moisture demand level to generate farmland management strategies, including:

[0059] The calculation formula for defining the soil partition model is as follows:

[0060] WDL=f(SMCbh,FCbh,PWPbh,SWTbh,H); (12)

[0061] In formula (12), H is the soil moisture evaluation coefficient, SMCbh, FCbh, PWPbh, SWTbh are soil water shortage indicators including soil moisture content, field water holding capacity, permanent wilting point and soil water tension, respectively, and WDL is the division value;

[0062] Based on the calculated soil water shortage index {SMCbh, FCbh, PWPbh, SWTbh} and soil moisture evaluation coefficient H, the soil water demand is divided into three levels; the three levels are low water demand level, medium water demand level and high water demand level;

[0063] For low moisture demand grades;

[0064] When H < 0.35, and the soil salinity interval is the low salinity interval (0-10%);

[0065] No additional irrigation or simple retention measures are required;

[0066] For medium moisture demand levels;

[0067] When 0.35≤H<0.64, and the soil salinity range is the medium salinity range [10%-15%];

[0068] Irrigate according to actual needs to prevent soil water shortage or over-irrigation;

[0069]

[0070] In formula (13), MWDmax and MWDmin are the maximum and minimum irrigation amounts for the medium water demand level, respectively, and Water Amount is the required water amount;

[0071] For high moisture demand grades;

[0072] When H≥0.64, and the soil salinity interval is the high salinity interval (15%-100%);

[0073] Take active irrigation measures to maintain soil moisture, pay attention to salt accumulation, and take soil improvement measures;

[0074] Water Amount=HWDmax×(1+Saltincrease); (14)

[0075] In formula (14), HWDmax is the maximum irrigation amount for the high water demand level, and Saltincrease is the percentage of salt accumulation increase.

[0076] Compared with the prior art, the beneficial effects of the present invention are: through the indicator economic crop determination module, historical soil salinity change data and soil water shortage indicators are obtained and analyzed, the most sensitive indicator economic crops are determined, accurate crop selection is achieved, and monitoring targeting is improved;

[0077] Through the correlation evaluation module, the change rate of soil water shortage index is calculated to obtain the correlation evaluation coefficient, so as to realize the correlation analysis between historical data and current monitoring data and enhance the prediction accuracy. Through the value range determination module, the soil salinity level is divided into intervals, and the value range of the index is determined in combination with the evaluation coefficient, so as to realize standardized management and improve data processing efficiency.

[0078] Through the indicator area determination module, the representative area is identified based on multispectral image data to achieve accurate monitoring and improve data representativeness; through the humidity evaluation module, the local spectral data is analyzed to construct the humidity evaluation coefficient, realize the quantitative assessment of soil humidity, and enhance the accuracy of moisture management;

[0079] Through the prediction model building module, machine learning is used to process multispectral data to establish a salt content prediction model, realize dynamic prediction, and improve foresight; through the correlation analysis module, the current soil water shortage index is determined by combining the prediction model and the standard interval, and real-time monitoring and prediction verification are realized to enhance system reliability;

[0080] By dividing the modules, combining the water shortage index and humidity evaluation coefficient, a soil division model is constructed to achieve the classification of soil moisture demand levels, optimize farmland management strategies, and improve resource utilization efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0081] Figure 1 It is a schematic diagram of the overall system flow of the present invention;

[0082] Figure 2 It is a system module block diagram of the present invention. DETAILED DESCRIPTION

[0083] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with specific embodiments.

[0084] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in the present invention should be understood by people with ordinary skills in the field to which the present invention belongs. The words "first", "second" and similar words used in the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. "Include" or "comprise" and similar words mean that the elements or objects appearing before the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connect" or "connected" and similar words are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative positional relationships. When the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0085] Embodiment 1:

[0086] See also Figure 1 and Figure 2 , the present invention provides a technical solution:

[0087] A remote sensing multispectral imaging system for agricultural monitoring, specifically comprising:

[0088] Indicator cash crop determination module: used to obtain the historical soil salinity change data of the target monitoring area in the previous year, as well as the soil water shortage index corresponding to the historical soil salinity change data; based on the historical soil salinity change data, determine the indicator cash crop that is most sensitive to the soil salinity change in the target monitoring area;

[0089] Correlation evaluation module: used to calculate the change rate of soil water shortage index between two consecutive months in the previous year, and to perform correlation analysis on the historical soil salinity change data and the change rate of soil water shortage index to obtain the correlation evaluation coefficient;

[0090] Value range determination module: used to divide the soil salinity level into several intervals according to the historical soil salinity change data, and determine the value range of the soil water shortage index in each soil salinity level interval by combining the correlation evaluation coefficient and the change rate of the soil water shortage index;

[0091] Indicator area determination module: used to obtain multispectral image data of the target monitoring area and then perform preprocessing, and determine the most representative indicator area in the image data based on historical soil salinity change data and correlation evaluation coefficients, as well as reference multispectral data corresponding to the indicator area. The indicator area is used to plant the determined indicator economic crops, and the indicator area can represent the soil salinity level of the target monitoring area;

[0092] Humidity evaluation module: used to receive the pre-processed reference multispectral data, extract the local spectral data used to evaluate soil moisture in the reference multispectral data, analyze and process the local spectral data, and construct a soil moisture evaluation coefficient. The soil moisture evaluation coefficient is used to evaluate the humidity of the target monitoring area;

[0093] Prediction model building module: used to obtain multispectral data that indicates the one-to-one correspondence between economic crops and historical soil salinity change data, and after data processing, build a machine learning model, and then input the processed multispectral data into the machine learning model for training and verification, and finally generate a salt content prediction model;

[0094] Correlation analysis module: used to substitute the pre-processed reference multispectral data into the salt content prediction model to obtain the prediction result of soil salinity level, and match the prediction result with several divided soil salinity level intervals to determine the soil water shortage index of the current target monitoring area;

[0095] Division module: It is used to build a soil division model by combining the soil water shortage index and soil moisture evaluation coefficient of the current target monitoring area. The model is used to divide the soil moisture demand level to generate farmland management strategies.

[0096] It is further explained that the soil salinity change data of the target monitoring area in the previous year are directly extracted from the agricultural monitoring database, and the acquired historical soil salinity change data and soil water shortage index are preprocessed; the data preprocessing includes data cleaning, normalization and interpolation processing for subsequent analysis;

[0097] Data cleaning: remove noise and outliers from the data;

[0098] Data normalization: scaling the data to a standard range of 0 to 1 to facilitate comparison of different data sources;

[0099] Data interpolation: Interpolate missing data to ensure data integrity.

[0100] Soil water shortage indicators include soil moisture content, field water holding capacity, permanent wilting point and soil water tension, and soil moisture content, field water holding capacity, permanent wilting point and soil water tension are marked as SMC, FC, PWP and SWT respectively;

[0101] Soil moisture content: It indicates the proportion of water in the soil, expressed as a percentage;

[0102] Field Capacity: The maximum amount of water that soil can hold under gravity, expressed as a percentage of volume;

[0103] Permanent Wilting Point: The lowest soil moisture content at which the plant cannot absorb water from the soil, expressed as a percentage by volume;

[0104] Soil Water Tension: The attraction of water in the soil, specifically comparing the current soil water tension to the maximum water tension, expressed as a percentage of Pascals (Pa).

[0105] The Pearson correlation coefficient was used to calculate the correlation between soil salinity changes and the corresponding multispectral data of cash crops; to determine the sensitivity of historical soil salinity change data to the growth of different cash crops, including:

[0106] Extract multispectral data of each economic crop in the target monitoring area;

[0107] The Pearson correlation coefficient between the historical soil salinity change data and the corresponding multispectral data of economic crops was calculated using the following formula:

[0108]

[0109] In formula (15), Xi ′ is the i′th soil salinity change data point, Yi ′ is the multispectral data point corresponding to the i′th economic crop, and They are the means of soil salinity change data and multispectral data corresponding to economic crops, respectively.

[0110] According to the results of correlation analysis, the cash crops with positive correlation with soil salinity change and the largest correlation coefficient were selected as indicator cash crops;

[0111] Rank the correlation coefficients for each cash crop;

[0112] Select the cash crop with the largest absolute value of the correlation coefficient;

[0113] Confirm that the cash crop is an indicator cash crop;

[0114] The model was validated on selected indicator cash crops to ensure their sensitivity to soil salinity changes; in actual monitoring, the model parameters were adjusted through set periodic verification and feedback to further improve the accuracy of the model.

[0115] Specific steps:

[0116] Verify the forecast accuracy of selected indicator cash crops through data monitoring in the current year;

[0117] Based on the verification results, the model parameters are adjusted and feedback optimization is performed.

[0118] The selection of indicated cash crops includes the first and second categories;

[0119] The first category selects a single type of indicator cash crop whose multispectral data can respond to different soil salinity levels;

[0120] The second category selects multiple types of indicator cash crops and analyzes the multispectral data of specific indicator cash crops under different soil salinity levels.

[0121] Further explanation: the change rate of soil water shortage index between two consecutive months is calculated in sequence, including:

[0122] Calculate the change rate of soil moisture content, field water holding capacity, permanent wilting point and soil water tension from each month to the next month in turn; set the current month as j and the next month as j+1, and the change rate formula is as follows:

[0123]

[0124] In formula (1), ΔZB represents the change rate of soil water shortage index, ZBj represents the soil water shortage index value in the jth month, and ZBj +1 represents the soil water shortage index value in month j+1;

[0125] The change rate can directly reflect the fluctuation of each soil water shortage index in the time dimension;

[0126] Pair the historical soil salinity change data with the monthly change rates of various soil water stress indicators;

[0127] Calculate the Pearson correlation coefficient between the historical soil salinity change data and the change rate of each soil water shortage index; the calculation formula is the same as the Pearson correlation coefficient in the indicator cash crop determination module, so it is not repeated here:

[0128] The Pearson correlation coefficient calculated for each soil water shortage index forms a correlation evaluation coefficient matrix FJ:

[0129]

[0130] In formula (2), r SMC 、r FC 、r PWP 、r SWT They are the Pearson correlation coefficients between soil salinity change data and the corresponding change rates of soil moisture content, field water holding capacity, permanent wilting point and soil water tension;

[0131] Based on historical soil salinity change data, the soil salinity level is divided into several intervals, including:

[0132] According to the percentage of soil salt content, it is divided into low salt interval (0, 10%), medium salt interval [10%-15%], and high salt interval (15%-100%);

[0133] Combined with the correlation evaluation coefficient matrix FJ and the change rate of each soil water shortage index, the value range of the soil water shortage index in each soil salinity level interval is determined; this embodiment uses linear regression or other statistical methods to calculate the change range of each soil water shortage index under different soil salinity levels;

[0134] The soil water shortage index value range calculated by linear regression is as follows:

[0135] The calculation formula for setting the soil water shortage index is Through the linear regression analysis of historical data, the regression coefficient a′ of each soil water shortage index is obtained. ZB and the intercept b′ ZB ; YFp is the salt level;

[0136] The correlation coefficient r in the correlation evaluation coefficient matrix FJ SMC 、r FC 、r PWP 、r SWT After standardization, the following calculation formula is obtained:

[0137]

[0138] w SMC 、w FC 、w PWP 、w SWT They are used as weight coefficients for soil moisture content, field capacity, permanent wilting point and soil water tension respectively;

[0139] The calculation formula for soil moisture content is:

[0140] SMCbh=w SMC ×(a SMC ×YFp+bSMC ); (5)

[0141] In formula (5), a SMC 、b SMC are the regression coefficient and intercept of soil moisture content respectively; YFp takes the endpoint values ​​on both sides of the current salt partition for calculation;

[0142] For example: in the low salt interval (0, 10%), YFp is taken as 0 and 10% respectively, and the SMCbh calculated twice is obtained respectively, and the two calculation results are used as the value range of the current soil moisture content; the same below, no further description is given;

[0143] The calculation formula for field water holding capacity is:

[0144] FCbh=w FC ×(a FC ×YFp+b FC )

[0145] Among them, a FC 、b FC are the regression coefficient and intercept of soil moisture content, respectively;

[0146] The calculation formula for the permanent wilting point is:

[0147] PWPbh=w PWP ×(a PWP ×YFp+b PWP )

[0148] Among them, a PWP 、b PWP are the regression coefficient and intercept of soil moisture content, respectively;

[0149] The calculation formula for the value of soil water tension is:

[0150] SWTbh=w SWT ×(a SWT ×YFp+b SWT )

[0151] Among them, a SWT 、b SWT are the regression coefficient and intercept of soil moisture content, respectively;

[0152] For the low salinity interval (0, 10%), irrigation measures were taken to maintain soil moisture;

[0153] The regression coefficient and intercept were determined based on the experimental data, and the following value ranges were obtained;

[0154] The soil moisture content is within the range of 4.5% ≤ SMC ≤ 5.2%;

[0155] The field water capacity is in the range of 3.5%≤FC≤4.0%;

[0156] The permanent wilting point is in the range of 3.0%≤PWP≤3.5%;

[0157] The soil water tension value is in the range of 5.5% ≤ SWT ≤ 6.0%;

[0158] For the medium salinity range [10%-15%], irrigation is carried out according to actual needs to prevent soil water shortage or over-irrigation; at the same time, based on the soil water shortage index value range determined in the low salinity range, the soil moisture content is taken as

[0159] The soil moisture content is in the range of 4.8% ≤ SMC ≤ 5.4%;

[0160] The field water capacity is in the range of 4.0%≤FC≤4.5%;

[0161] The permanent wilting point is in the range of 3.5%≤PWP≤4.0%;

[0162] The soil water tension value is in the range of 5.8% ≤ SWT ≤ 6.2%;

[0163] For the high-salinity range (15%-100%), attention should be paid to soil salt accumulation, and soil improvement measures should be taken to increase organic matter and improve irrigation technology to improve soil structure and water retention capacity.

[0164] The soil moisture content is within the range of 5.0% ≤ SMC ≤ 5.6%;

[0165] The field water capacity is in the range of 4.5%≤FC≤5.0%;

[0166] The permanent wilting point is in the range of 4.0%≤PWP≤4.5%;

[0167] The soil water tension value is in the range of 6.0% ≤ SWT ≤ 6.5%;

[0168] The mapping relationship table is shown in Table 1 below:

[0169] Table 1

[0170]

[0171] According to the above mapping relationship, the soil water shortage index value range of the target monitoring area is determined under different soil salinity levels:

[0172] Low salinity level: When the soil salinity level is below 10%, the soil moisture content should be maintained between 4.5% and 5.2%, the field capacity should be between 3.5% and 4.0%, the permanent wilting point should be between 3.0% and 3.5%, and the soil water tension should be between 5.5% and 6.0%; in this case, irrigation measures can be taken to maintain soil moisture;

[0173] Medium salinity level: When the soil salinity level is between 10% and 15%, the soil moisture content should be controlled between 4.8% and 5.4%, the field water holding capacity should be between 4.0% and 4.5%, the permanent wilting point should be between 3.5% and 4.0%, and the soil water tension should be maintained between 5.8% and 6.2%; at this time, appropriate irrigation should be carried out according to actual needs to prevent soil water shortage or over-irrigation;

[0174] High salinity level: When the soil salinity level is higher than 15%, the soil moisture content should be maintained between 5.0% and 5.6%, the field water holding capacity should be between 4.5% and 5.0%, the permanent wilting point should be between 4.0% and 4.5%, and the soil water tension should be between 6.0% and 6.5%; in this case, it is necessary to focus on the accumulation of salt in the soil, take necessary soil improvement measures, increase organic matter, improve irrigation technology, etc., to improve soil structure and water retention capacity.

[0175] Further explanation: the multispectral image data of the target monitoring area is obtained and then preprocessed, and the most representative indication area is determined in the image data based on the historical soil salinity change data and the correlation evaluation coefficient, and the reference multispectral data corresponding to the indication area is used to plant the determined indication economic crops, specifically including:

[0176] Use remote sensing satellites to obtain multispectral image data of the target monitoring area;

[0177] Multispectral images include multiple bands such as red, green, blue, and near-infrared to facilitate subsequent image processing and analysis;

[0178] Preprocessing the acquired multispectral image data, including radiation correction, geometric correction and denoising operations, to ensure the accuracy and consistency of the image data; radiation correction, geometric correction and denoising operations are existing technologies and will not be described in detail;

[0179] Radiometric correction is to eliminate the impact of sensors and atmosphere on image data to ensure that the reflectivity of the ground objects represented by each pixel is accurate;

[0180] Geometric correction is to eliminate the geometric distortion caused by the change of sensor angle or terrain undulation during image acquisition, so that the image is aligned with the actual geographic location;

[0181] Denoising is to remove noise from the image and enhance the clarity and usability of the image;

[0182] Using historical soil salinity change data, the pre-processed multispectral images are analyzed to extract current soil salinity level information;

[0183] By analyzing the near-infrared band and red band in the multispectral image, the soil salt index is calculated. The calculation formula is as follows:

[0184]

[0185] Among them, NIR is the reflectance of the near infrared band, R is the reflectance of the red band; SSI is the soil salinity index; this formula is used to quantify the soil salinity level of each pixel in the image;

[0186] Obtain the correlation evaluation coefficient matrix FJ, and determine the most representative indicator area in the target monitoring area based on the soil salinity index and the correlation evaluation coefficient matrix FJ, including:

[0187] The target monitoring area is divided into several small areas, each of which contains several pixels;

[0188] For each pixel in a small area, calculate the weighted average of the soil salinity index:

[0189]

[0190] Among them, SSIi represents the soil salinity index of the i-th pixel, wi represents the corresponding weight, and SSIavg is the weighted average of the soil salinity index;

[0191] The weighted average calculation of soil water shortage index in each small area is:

[0192]

[0193] Among them, ZBi represents the soil water shortage index value of the i-th pixel, wi′ represents the corresponding weight, and ZBavg is the weighted average value of the soil water shortage index; wi′ is determined by the expert group through experimental data and will not be repeated here;

[0194] Combining the weighted average of soil salinity index and soil water shortage index, and using the correlation evaluation coefficient matrix FJ, the comprehensive score of each small area is calculated:

[0195]

[0196] Among them, S represents the comprehensive score, SSIavg is the weighted average of the soil salinity index; ZBavg is the weighted average of the soil water shortage index; the small area with the highest comprehensive score is selected as the indicator area that best represents the soil salinity level and water shortage conditions in the target monitoring area.

[0197] Further explanation: construct a soil moisture evaluation coefficient, which is used to evaluate the humidity of the target monitoring area; specifically includes:

[0198] Extract local spectral data for soil moisture evaluation from the preprocessed reference multispectral data, including:

[0199] Select the spectral bands sensitive to soil moisture for analysis and processing;

[0200] Select the red edge band, shortwave infrared band, and near infrared band; these bands are more sensitive to changes in soil moisture;

[0201] Extracting the reflectance values ​​of the red edge band, the short-wave infrared band and the near infrared band from the reference multispectral data as local spectral data;

[0202] The soil moisture index is calculated using the extracted local spectral data;

[0203]

[0204] Among them, RE represents the reflectivity of the red edge band, SWI R represents the reflectivity of the short-wave infrared band, NIR′ represents the reflectivity of the near-infrared band, RED represents the reflectivity of the red band, a1, a2, and a3 are all positive adjustment coefficients, and a1, a2, and a3 are adjusted based on actual conditions and experimental data;

[0205] Divide the indicated area into several small areas, each of which contains several pixels;

[0206] For each pixel in a small area, calculate the weighted average of the soil moisture index:

[0207]

[0208] Among them, SMIu represents the soil moisture index of the u-th pixel, wu represents the corresponding weight; wi′ is determined by the expert group through experimental data and will not be described in detail;

[0209] The soil moisture evaluation coefficient is defined as H, and the calculation formula is as follows:

[0210]

[0211] Among them, b1 and b2 are both positive adjustment coefficients, SMIavg is the weighted average of the soil moisture index, and Jy is the average rainfall in the target monitoring area in the current month; b1 and b2 are determined by the expert group based on experimental data;

[0212] According to the value of soil moisture evaluation coefficient H, the humidity conditions in the target monitoring area are divided into low humidity, medium humidity and high humidity; specifically, they include:

[0213] The effective value range of H is limited to (0,1). If the value range of H is in the interval of (0,0.35), the humidity condition of the target monitoring area is low humidity; the corresponding average rainfall is between 0 and A1 mm, excluding A1;

[0214] If the value of H is in the interval [0.35, 0.64), the humidity in the target monitoring area is medium humidity; the corresponding average rainfall is between A1 and A2 mm;

[0215] If the value range of H is in the interval [0.64,1), the humidity condition in the target monitoring area is high humidity; the corresponding average rainfall exceeds A2 mm.

[0216] To further illustrate, a machine learning model is constructed, and then the processed multispectral data is input into the machine learning model for training and verification, and finally a salt content prediction model is generated, specifically including:

[0217] Using geographic information system (GIS) tools, multispectral data and soil salinity data were spatially aligned through spatial interpolation and resampling techniques to ensure the spatial consistency of the two datasets;

[0218] Extract key spectral features from historical multispectral data, including but not limited to: vegetation index, soil conditioning index, red edge position;

[0219] According to the time series data, the spectral characteristic change rate and soil salinity change rate at each time point are calculated to form the time series characteristics;

[0220] A method based on multi-source data fusion is used to fuse spectral features and soil salinity features. A multi-layer perceptron (MLP) neural network is used as a data fusion model. MLP is selected because it can handle complex nonlinear relationships and has good generalization ability.

[0221] The extracted spectral features and soil salinity features are input into the MLP neural network for feature fusion. The number of nodes in the network input layer is the total number of spectral features and soil salinity features, the number of hidden layer nodes is optimized according to the experimental results, and the number of output layer nodes is the predicted value of soil salinity.

[0222] Use historical data to train the MLP neural network; use the cross-validation method to divide the data set into a training set and a validation set. The training set is used to train the model, and the validation set is used to evaluate the model performance. The mean square error (MSE) is used as the loss function, and the Adam optimizer is used as the optimization algorithm.

[0223] The performance of the trained model is verified through the validation set, and the prediction accuracy and error of the model are calculated; the formula is as follows:

[0224]

[0225] Among them, M′ is the number of samples, ym ′ is the actual value of the m′th sample, is the predicted value of the m′th sample;

[0226] According to the verification results, adjust the parameters of the MLP neural network, including the number of hidden layer nodes, learning rate, number of training rounds, etc., to gradually optimize the model performance;

[0227] The optimized MLP neural network model is saved as the final salt content prediction model. This model can process new multispectral data in real time and predict the soil salt content in the corresponding area.

[0228] Further explanation: the pre-processed reference multispectral data is substituted into the salt content prediction model to obtain the predicted result of soil salinity level, which is matched with several divided soil salinity level intervals to determine the soil water shortage index of the current target monitoring area, including:

[0229] Match the predicted salt content with the pre-divided salt interval to determine the salt level of the current target monitoring area;

[0230] Specifically, according to the percentage of soil salt content, the predicted salt content is matched with the low salt interval (0, 10%), the medium salt interval [10%-15%], and the high salt interval (15%-100%) to determine the salt interval to which it belongs;

[0231] Based on the confirmed salinity range, the value ranges of soil water shortage indicators including soil moisture content, field water holding capacity, permanent wilting point and soil water tension are determined.

[0232] Further explanation: It is used to combine the soil water shortage index and soil moisture evaluation coefficient of the current target monitoring area to build a soil classification model, which is used to classify the soil moisture demand level to generate farmland management strategies, including:

[0233] The calculation formula for defining the soil partition model is as follows:

[0234] WDL=f(SMCbh,FCbh,PWPbh,SWTbh,H)

[0235] Among them, H is the soil moisture evaluation coefficient, SMCbh, FCbh, PWPbh, SWTbh are soil water shortage indicators including soil moisture content, field water holding capacity, permanent wilting point and soil water tension, and WDL is the division value;

[0236] Based on the calculated soil water shortage index (SMCbh, FCbh, PWPbh, SWTbh) and soil moisture evaluation coefficient H, the soil water demand is divided into three levels: low water demand level, medium water demand level and high water demand level;

[0237] For low moisture demand grades;

[0238] When H < 0.35, and the soil salinity interval is the low salinity interval (0-10%);

[0239] No additional irrigation or simple retention measures are required;

[0240] For medium moisture demand levels;

[0241] When 0.35≤H<0.64, and the soil salinity range is the medium salinity range [10%-15%];

[0242] Irrigate according to actual needs to prevent soil water shortage or over-irrigation;

[0243]

[0244] Among them, MWDmax and MWDmin are the maximum and minimum irrigation amounts for the medium water demand level, respectively, and WaterAmount is the required water amount;

[0245] For high moisture demand grades;

[0246] When H≥0.64, and the soil salinity interval is the high salinity interval (15%-100%);

[0247] Take active irrigation measures to maintain soil moisture, pay attention to salt accumulation, and take soil improvement measures;

[0248] Water Amount=HWDmax×(1+Saltincrease)

[0249] Among them, HWDmax is the maximum irrigation amount for the high water demand level, and Saltincrease is the percentage of salt accumulation increase.

[0250] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0251] The above embodiments may be implemented in whole or in part by software, hardware, firmware or any other combination thereof. When implemented by software, the above embodiments may be implemented in whole or in part in the form of a computer program product. Those skilled in the art may appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein may be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed by hardware or software methods depends on the specific application and design constraints of the technical solution.

[0252] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, and may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0253] The above description is only a specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application.

Claims

1. A remote sensing multispectral imaging system for agricultural monitoring, characterized in that: Specifically include: Indicator cash crop determination module: used to obtain the historical soil salinity change data of the target monitoring area in the previous year, as well as the soil water shortage index corresponding to the historical soil salinity change data; based on the historical soil salinity change data, determine the indicator cash crop that is most sensitive to the soil salinity change in the target monitoring area; Correlation evaluation module: used to calculate the change rate of soil water shortage index between two consecutive months in the previous year, and to perform correlation analysis on the historical soil salinity change data and the change rate of soil water shortage index to obtain the correlation evaluation coefficient; Value range determination module: used to divide the soil salinity level into several intervals according to the historical soil salinity change data, and determine the value range of the soil water shortage index in each soil salinity level interval by combining the correlation evaluation coefficient and the change rate of the soil water shortage index; Indicator area determination module: used to obtain multispectral image data of the target monitoring area and then perform preprocessing, and determine the most representative indicator area in the image data based on historical soil salinity change data and correlation evaluation coefficients, as well as reference multispectral data corresponding to the indicator area. The indicator area is used to plant the determined indicator economic crops, and the indicator area can represent the soil salinity level of the target monitoring area; Humidity evaluation module: used to receive the pre-processed reference multispectral data, extract the local spectral data used to evaluate soil moisture in the reference multispectral data, analyze and process the local spectral data, and construct a soil moisture evaluation coefficient. The soil moisture evaluation coefficient is used to evaluate the humidity of the target monitoring area; Prediction model building module: used to obtain multispectral data that indicates the one-to-one correspondence between economic crops and historical soil salinity change data, and after data processing, build a machine learning model, and then input the processed multispectral data into the machine learning model for training and verification, and finally generate a salt content prediction model; Correlation analysis module: used to substitute the pre-processed reference multispectral data into the salt content prediction model to obtain the prediction result of soil salinity level, and match the prediction result with several divided soil salinity level intervals to determine the soil water shortage index of the current target monitoring area; Division module: It is used to build a soil division model by combining the soil water shortage index and soil moisture evaluation coefficient of the current target monitoring area. The model is used to divide the soil moisture demand level to generate farmland management strategies.

2. A remote sensing multispectral imaging system for agricultural monitoring according to claim 1, characterized in that: Preprocess the acquired historical soil salinity change data and soil water shortage index; Soil water shortage indicators include soil moisture content, field water holding capacity, permanent wilting point and soil water tension, and soil moisture content, field water holding capacity, permanent wilting point and soil water tension are marked as SMC, FC, PWP and SWT respectively; The Pearson correlation coefficient was used to calculate the correlation between soil salinity changes and the corresponding multispectral data of economic crops; To determine the historical soil salinity change data and the growth sensitivity of different cash crops, including: Extract multispectral data of each economic crop in the target monitoring area; Calculate the Pearson correlation coefficient between historical soil salinity change data and corresponding multispectral data of economic crops; According to the results of correlation analysis, the cash crops with positive correlation with soil salinity changes and the largest correlation coefficient were selected as indicator cash crops; The selection of indicated cash crops includes the first and second categories; The first category selects a single type of indicator cash crop whose multispectral data can respond to different soil salinity levels; The second category selects multiple types of indicator cash crops and analyzes the multispectral data of specific indicator cash crops under different soil salinity levels.

3. A remote sensing multispectral imaging system for agricultural monitoring according to claim 2, characterized in that: The change rates of soil water shortage indicators between two consecutive months are calculated in sequence, including: Calculate the change rate of soil moisture content, field water holding capacity, permanent wilting point and soil water tension from each month to the next month in turn; set the current month as j and the next month as j+1, and the change rate formula is as follows: In formula (1), ΔZB represents the change rate of soil water shortage index, ZB j represents the soil water shortage index value in the jth month, ZB j+1 represents the soil water shortage index value in month j+1; Pair the historical soil salinity change data with the monthly change rates of various soil water stress indicators; Calculate the Pearson correlation coefficient between historical soil salinity change data and the change rate of each soil water shortage index; The Pearson correlation coefficients calculated for each soil water shortage index form a correlation evaluation coefficient matrix FJ: In formula (2), r SMC 、r FC 、r PWP 、r SWT They are the Pearson correlation coefficients between soil salinity change data and the corresponding change rates of soil moisture content, field water holding capacity, permanent wilting point and soil water tension; Based on historical soil salinity change data, soil salinity levels are divided into several intervals, including: According to the percentage of soil salt content, it is divided into low salt interval (0, 10%), medium salt interval [10%-15%], and high salt interval (15%-100%); Combined with the correlation evaluation coefficient matrix FJ and the change rate of each soil water shortage index, the value range of the soil water shortage index in each soil salinity level interval is determined, including: The calculation formula for setting the soil water shortage index is Through the linear regression analysis of historical data, the regression coefficient a′ of each soil water shortage index is obtained. ZB and the intercept b′ ZB ; YFp is the salt level; The correlation coefficient r in the correlation evaluation coefficient matrix FJ SMC 、r FC 、r PWP 、r SWT After standardization, the following calculation formula is obtained: w SMC 、w FC 、w PWP 、w SWT They are used as weight coefficients for soil moisture content, field capacity, permanent wilting point and soil water tension respectively; The calculation formula for soil moisture content is: SMC bh =w SMC ×(a SMC ×YFp+b SMC ); (5) In formula (5), a SMC , b SMC are the regression coefficient and intercept of soil moisture content respectively; YFp takes the endpoint values ​​on both sides of the current salt partition for calculation; The calculation formula for field water holding capacity is: FC bh =w FC ×(a FC ×YFp+b FC ); (6) In formula (6), a FC , b FC are the regression coefficient and intercept of soil moisture content, respectively; The calculation formula for the permanent wilting point is: PWP bh =w PWP ×(a PWP ×YFp+b PWP ); (7) In formula (7), a PWP 、b PWP are the regression coefficient and intercept of soil moisture content, respectively; The calculation formula for the value of soil water tension is: SWT bh =w SWT ×(a SWT ×YFp+b SWT ); (8) In formula (8), a SWT 、b SWT are the regression coefficient and intercept of soil moisture content, respectively.

4. A remote sensing multispectral imaging system for agricultural monitoring according to claim 3, characterized in that: Use remote sensing satellites to obtain multispectral image data of the target monitoring area; Using historical soil salinity change data, the pre-processed multispectral images are analyzed to extract current soil salinity level information; By analyzing the near-infrared band and red band in the multispectral image, the soil salt index is calculated. The calculation formula is as follows: In formula (9), NIR is the near-infrared band reflectance, R is the red band reflectance; SSI is the soil salinity index; Obtain the correlation evaluation coefficient matrix FJ, and determine the most representative indicator area in the target monitoring area based on the soil salinity index and the correlation evaluation coefficient matrix FJ, including: The target monitoring area is divided into several small areas, each of which contains several pixels; For each pixel in a small area, calculate the weighted average of the soil salinity index: The weighted average calculation of soil water shortage index in each small area is: Combining the weighted average of soil salinity index and soil water shortage index, the comprehensive score of each small area is calculated using the correlation evaluation coefficient matrix FJ: In formula (10), S represents the comprehensive score, SSI avg is the weighted average value of soil salinity index; ZB avg It is the weighted average of the soil water shortage index; the small area with the highest comprehensive score is selected as the indicator area that best represents the soil salinity level and water shortage conditions in the target monitoring area.

5. A remote sensing multispectral imaging system for agricultural monitoring according to claim 4, characterized in that: Construct a soil moisture evaluation coefficient, which is used to evaluate the humidity of the target monitoring area, including: Extracting local spectral data for soil moisture evaluation from the preprocessed reference multispectral data, and calculating the soil moisture index using the extracted local spectral data; Divide the indicated area into several small areas, each of which contains several pixels; For each pixel in a small area, calculate the weighted average of the soil moisture index: The soil moisture evaluation coefficient is defined as H, and the calculation formula is as follows: In formula (11), b1 and b2 are both positive adjustment coefficients, SMI avg is the weighted average of the soil moisture index, Jy is the average rainfall in the target monitoring area in the current month; the effective value range of H is limited to (0,1); According to the value of the soil moisture evaluation coefficient H, the humidity conditions in the target monitoring area are divided into low humidity, medium humidity and high humidity.

6. A remote sensing multispectral imaging system for agricultural monitoring according to claim 5, characterized in that: Build a machine learning model, then input the processed multispectral data into the machine learning model for training and verification, and finally generate a salt content prediction model, including: Extract key spectral features from historical multispectral data, including but not limited to: vegetation index, soil conditioning index, red edge position; According to the time series data, the spectral characteristic change rate and soil salinity change rate at each time point are calculated to form the time series characteristics; A method based on multi-source data fusion is used to fuse spectral features and soil salinity features; a multi-layer perceptron neural network is used as a data fusion model; The extracted spectral features and soil salinity features are input into the MLP neural network for feature fusion; Use historical data to train the MLP neural network; The performance of the trained model is verified through the validation set, and the prediction accuracy and error of the model are calculated; According to the verification results, adjust the parameters of the MLP neural network; The optimized MLP neural network model is saved as the final salt content prediction model.

7. A remote sensing multispectral imaging system for agricultural monitoring according to claim 6, characterized in that: Substitute the pre-processed reference multispectral data into the salt content prediction model to obtain the predicted results of soil salinity level. The predicted results are matched with several divided soil salinity level intervals to determine the soil water shortage index of the current target monitoring area, including: Match the predicted salt content with the pre-divided salt interval to determine the salt level of the current target monitoring area; According to the percentage of soil salt content, the predicted salt content is matched with the low salt interval (0, 10%), the medium salt interval [10%-15%], and the high salt interval (15%-100%) to determine the salt interval to which it belongs; Based on the confirmed salinity range, the value ranges of soil water shortage indicators including soil moisture content, field water holding capacity, permanent wilting point and soil water tension are determined.

8. A remote sensing multispectral imaging system for agricultural monitoring according to claim 7, characterized in that: It is used to build a soil classification model based on the soil water shortage index and soil moisture evaluation coefficient of the current target monitoring area. The model is used to classify the soil water demand level to generate farmland management strategies, including: The calculation formula for defining the soil partition model is as follows: WDL=f(SMC bh ,FC bh ,PWP bh ,SWT bh ,H); (12) In formula (12), H is the soil moisture evaluation coefficient, SMC bh ,FC bh ,PWP bh ,SWT bh They are soil water shortage indicators including soil moisture content, field water holding capacity, permanent wilting point and soil water tension, and WDL is the division value; Based on the calculated soil water shortage index {SMC bh ,FC bh ,PWP bh ,SWT bh } and soil moisture evaluation coefficient H, the soil moisture demand is divided into three levels; the three levels are low moisture demand level, medium moisture demand level and high moisture demand level; For low moisture demand grades; When H < 0.35, and the soil salinity interval is the low salinity interval (0-10%); No additional irrigation or simple retention measures are required; For medium moisture demand levels; When 0.35≤H<0.64, and the soil salinity range is the medium salinity range [10%-15%]; Irrigate according to actual needs to prevent soil water shortage or over-irrigation; In formula (13), MWD max and MWD min They are the maximum and minimum irrigation amounts for the medium water demand level, respectively, and WaterAmount is the required water amount; For high moisture demand grades; When H≥0.64, and the soil salinity interval is the high salinity interval (15%-100%); Take active irrigation measures to maintain soil moisture, pay attention to salt accumulation, and take soil improvement measures; Water Amount=HWD max ×(1+Salt increase ); (14) In formula (14), HWD max The maximum irrigation amount for high water demand level, Salt increase is the percentage increase in salt accumulation.

Images