An agricultural drought monitoring method based on multi-source soil moisture content fusion

Abstract

The application discloses the technical field of agricultural drought monitoring and a kind of agricultural drought monitoring method based on multi-source soil moisture fusion, the method includes the following steps: obtaining active remote sensing soil moisture data, passive remote sensing soil moisture data and VIC model soil moisture data;Scale recalibration is carried out to active remote sensing soil moisture data and passive remote sensing soil moisture data by constructing cumulative distribution function;Multi-source soil moisture fusion is carried out using TC algorithm by season;Soil relative humidity index is calculated based on fusion soil moisture data and field water capacity data;Drought grade is determined based on soil relative humidity index and soil texture data.The application uses the mode of multi-source data fusion by season, not only solves the problem of insufficient monitoring depth and space-time discontinuity of remote sensing soil moisture, but also solves the problem of simulation deviation caused by irrigation influence of model soil moisture, and realizes the accurate monitoring of national scale agricultural drought.

Description

A method for monitoring agricultural drought based on multi-source soil moisture content fusion Technical Field

[0001] This invention relates to the field of agricultural drought monitoring technology, specifically to an agricultural drought monitoring method based on the fusion of multi-source soil moisture content. Background Technology

[0002] Soil moisture content is the most critical state variable in agricultural drought monitoring and early warning. It is a comprehensive indicator reflecting numerous climate variables, vegetation, and soil characteristics, and the balance of soil moisture directly and decisively affects crop growth and production. There are three main methods for obtaining soil moisture content: site observation, model simulation, and satellite remote sensing retrieval. Among these, site observation provides the highest accuracy but is difficult to implement on a large scale, making it unsuitable for large-scale agricultural drought monitoring. Model simulation can obtain continuous spatiotemporal distribution data of soil moisture content, but it lacks ground observations, cannot identify changes in soil moisture content caused by irrigation, and performs poorly in areas without data, leading to biases in drought monitoring. Satellite remote sensing retrieval can obtain global-scale surface soil moisture distribution data, reflecting the true dryness and wetness of the underlying surface. However, in practical applications, it suffers from spatiotemporal discontinuity and accuracy uncertainties, resulting in low accuracy in drought monitoring.

[0003] Therefore, how to solve the problems of accuracy and spatiotemporal coverage of existing models and remote sensing soil moisture content, and improve the accuracy of large-scale agricultural drought monitoring and early warning, has become a technical problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0004] In view of this, the purpose of this invention is to provide an agricultural drought monitoring method based on the fusion of multi-source soil moisture content, so as to solve the technical problem of insufficient accuracy in existing large-scale agricultural drought monitoring methods.

[0005] The technical solution adopted in this invention is: an agricultural drought monitoring method based on multi-source soil moisture content fusion, comprising the following steps:

[0006] S100: Acquire active remote sensing soil moisture data, passive remote sensing soil moisture data, and VIC model soil moisture data;

[0007] S200: Using the grid corresponding to the soil moisture content data of the VIC model as the reference grid, match the active remote sensing soil moisture content data and the passive remote sensing soil moisture content data to the reference grid.

[0008] S300: Construct the cumulative distribution function of soil moisture data grid by grid, and perform scale recorrection on active remote sensing soil moisture data and passive remote sensing soil moisture data based on the soil moisture data of the VIC model to obtain active remote sensing soil moisture corrected data and passive remote sensing soil moisture corrected data.

[0009] S400: The TC algorithm is used seasonally to fuse the active remote sensing soil moisture correction data, passive remote sensing soil moisture correction data, and VIC model soil moisture data to obtain fused soil moisture data.

[0010] S500: Calculate the soil relative humidity index based on the fused soil moisture content data and field water holding capacity data;

[0011] S600: Based on the soil relative humidity index and soil texture data, the drought level is determined to achieve agricultural drought monitoring.

[0012] Preferably, S100 specifically includes: acquiring soil texture data, field water holding capacity data, ASCAT active remote sensing soil moisture content data, SMAP passive remote sensing soil moisture content data, and VIC model soil moisture content data.

[0013] Preferably, step S200 specifically includes: using the grid corresponding to the soil moisture content data of the VIC model as the reference grid, resampling and matching the active remote sensing soil moisture content data and passive remote sensing soil moisture content data to the reference grid through an inverse distance weighted interpolation method; matching the soil texture data to the reference grid through a grid averaging method; and matching the field water holding capacity data to the reference grid through a grid averaging method.

[0014] Preferably, step S300 specifically includes: firstly, constructing a cumulative distribution function of soil moisture content for each grid of the active remote sensing soil moisture content data, passive remote sensing soil moisture content data, and VIC model soil moisture content data; then, using the VIC model soil moisture content data corresponding to the reference grid as the reference data, interpolating the remote sensing soil moisture content values ​​at each cumulative frequency to the cumulative frequency corresponding to the simulated soil moisture content using a linear interpolation method, so that the active remote sensing soil moisture content data, passive remote sensing soil moisture content data, and VIC model soil moisture content data have the same cumulative distribution frequency curve, realizing the scale recorrection of the remote sensing soil moisture content data, and obtaining the active remote sensing soil moisture content corrected data and the passive remote sensing soil moisture content corrected data.

[0015] Preferably, step S400 specifically includes: dividing the soil moisture content correction data, passive remote sensing soil moisture content correction data, and VIC model soil moisture content data in the same reference grid for each season as a group of data to be fused, constructing fusion weights for the soil moisture content data using the TC algorithm, and then performing multi-source soil moisture content fusion on the active remote sensing soil moisture content correction data, passive remote sensing soil moisture content correction data, and VIC model soil moisture content data based on the fusion weights to obtain the fused soil moisture content data.

[0016] Preferably, the fusion weights are constructed as follows:

[0017] First, define the true soil moisture content of the grid as θ, the active remote sensing soil moisture content as θ1, the passive remote sensing soil moisture content as θ2, and the VIC model soil moisture content as θ3. The linear relationships between θ and θ1, and between θ2 and θ3 are as follows:

[0018]

[0019] Where β1, β2 and β3 are error deviation terms, α1, α2 and α3 are scaling coefficient terms, and ε1, ε2 and ε3 are unbiased random error terms;

[0020] Then, by eliminating θ through pairwise equations, we can obtain the following formula:

[0021]

[0022] in, Let i be the variance of the error corresponding to data product i. These are time series values ​​scaled according to reference data; <> represents the mean calculation.

[0023] The variances between the active remote sensing soil moisture content, the passive remote sensing soil moisture content, and the VIC model soil moisture content and the actual soil moisture content were calculated separately. The correlation coefficients between these values ​​and the actual soil moisture content were then determined.

[0024]

[0025] Where i, j, k represent the datasets of active remote sensing soil moisture content, passive remote sensing soil moisture content, and VIC model soil moisture content, respectively; R represents the corresponding correlation coefficient; σ i,j Represents the covariance of dataset i and dataset j. The variance of the error in dataset i is represented.

[0026] Finally, based on the correlation coefficient, a fusion weight w is constructed for the active remote sensing soil moisture content correction data, the passive remote sensing soil moisture content correction dataset, and the VIC model soil moisture content data. i for:

[0027]

[0028] Preferably, the soil texture data in S600 includes sand content, clay content, and silt content. The soil type classification criteria are as follows: if the sand content is greater than 50%, the clay content is less than 15%, and the silt content is between 15% and 50%, the soil type is sandy soil; if the sand content is between 25% and 50%, the clay content is between 15% and 40%, and the silt content is less than 50%, the soil type is loam; if the sand content is less than 40%, the clay content is greater than 40%, and the silt content is less than 20%, the soil type is clay.

[0029] Preferably, the criteria for classifying drought categories in S600 are as follows:

[0030]

[0031] The beneficial effects of this invention are:

[0032] This invention employs a multi-source data fusion approach combining remote sensing and model data. By performing remote sensing data scale correction, determining fusion weights using the TC algorithm, and seasonal fusion, it constructs high-precision, spatiotemporally continuous fused soil moisture content data, solving the problems of spatiotemporal discontinuity and insufficient accuracy of the original soil moisture content data. Based on this, it considers different soil types to formulate corresponding drought threshold classification standards, achieving accurate, efficient, and reliable agricultural drought monitoring.

[0033] This invention can effectively improve the efficiency and accuracy of monitoring soil moisture content and agricultural drought. Attached Figure Description

[0034] Figure 1 is a logic block diagram of the agricultural drought monitoring method based on multi-source soil moisture content fusion of the present invention;

[0035] Figure 2 is a box plot showing the correlation coefficients between active remote sensing soil moisture content, passive remote sensing soil moisture content, VIC model soil moisture content, fused soil moisture content, and station-observed soil moisture content in a specific embodiment.

[0036] Figure 3 is a scatter plot of drought area and drought intensity shown in a specific embodiment. Detailed Implementation

[0037] The specific embodiments of the present invention will be further described in detail below with reference to the accompanying drawings. These embodiments are for illustrative purposes only and are not intended to limit the scope of the invention.

[0038] As shown in Figure 1, an agricultural drought monitoring method based on multi-source soil moisture content fusion is described. This method can monitor drought over a large area of ​​soil and identify soil drought categories. The method includes the following steps:

[0039] S100: Acquire active remote sensing soil moisture content data, passive remote sensing soil moisture content data, and VIC model soil moisture content data.

[0040] S200: Using the grid corresponding to the soil moisture content data of the VIC model as the reference grid, the active remote sensing soil moisture content data and the passive remote sensing soil moisture content data are matched to the reference grid.

[0041] S300: Construct the cumulative distribution function of soil moisture data grid by grid, and perform scale recalibration on active remote sensing soil moisture data and passive remote sensing soil moisture data based on VIC model soil moisture data to obtain active remote sensing soil moisture correction data and passive remote sensing soil moisture correction data.

[0042] S400: The TC algorithm is used seasonally to fuse multi-source soil moisture data, including active remote sensing soil moisture correction data, passive remote sensing soil moisture correction data, and VIC model soil moisture data, to obtain fused soil moisture data.

[0043] S500: Soil relative humidity index is calculated based on the fusion of soil moisture content data and field water holding capacity data.

[0044] S600: Based on soil relative humidity index and soil texture data, drought level is determined to achieve agricultural drought monitoring.

[0045] This application adopts a multi-source data fusion approach. First, it constructs a cumulative distribution function for multi-source soil moisture content data to correct the multi-source soil moisture content data. Then, it uses the TC algorithm to fuse the multi-source soil moisture content data, solving the problems of spatiotemporal discontinuity and simulation bias. In addition, it uses a constructed drought threshold classification standard to identify drought levels, which can effectively improve the accuracy of soil moisture content monitoring and the reliability of agricultural drought monitoring.

[0046] As shown in Figures 1-3, a specific embodiment of an agricultural drought monitoring method based on multi-source soil moisture content fusion is presented. This method can monitor and identify agricultural drought across the entire country. The method includes the following steps:

[0047] S100: Data Collection; This data specifically includes on-site measured soil moisture content data, ASCAT active remote sensing soil moisture content data, SMAP passive remote sensing soil moisture content data, VIC model soil moisture content data, soil texture data, field water holding capacity data, and crop drought-affected area data. The data acquisition methods are as follows:

[0048] S110: Obtain the measured soil moisture content data at the site.

[0049] Specifically, the data was collected from the Information Center of the Ministry of Water Resources, which collected measured soil moisture content data from soil moisture monitoring stations. These stations are located in all provinces and cities across the country, totaling more than 2,300.

[0050] For all sampling points, the soil moisture content at depths of 0–10 cm, 10–20 cm, and 20–40 cm was measured. The average soil moisture content at each site was calculated using a depth-weighted average, as shown in the following formula:

[0051]

[0052] Where, θ i Represents the measured soil moisture content, h i h represents the soil layer thickness of the soil sample at the site. i The thicknesses are 10cm, 10cm, and 20cm respectively, with H representing the total soil layer thickness, which is 40cm.

[0053] S120: Acquire ASCAT active remote sensing soil moisture content data.

[0054] Specifically, the ASCAT sensor aboard the MetOp satellite is a true aperture radar capable of measuring the backscattered signal of a target, exhibiting good accuracy and stability in the C-band (5.3 GHz). The MetOp satellite is in a sun-synchronous orbit, crossing the equator at approximately 21:30 (ascending) and 09:30 (descending) local time. The ASCAT sensor has a spatial resolution of 25–34 km and a revisit frequency of 1–3 days for backscattered observations. The daily soil moisture content data used in this embodiment was obtained through inversion based on a time-varying detection algorithm, sourced from the EUMETSAT Satellite Applications Facility (H-SAF) project supporting operational hydrology and water management applications, with a data time series spanning from 2015 to 2021.

[0055] S130: Acquire SMAP passive remote sensing soil moisture content data.

[0056] Specifically, the SMAP satellite was launched by NASA on January 31, 2015, to acquire global-scale soil moisture data. The satellite carries an L-band microwave radiometer with a revisit time of 2–3 days, and is located in a 685 km sun-synchronous orbit, comprising two and a half orbits: an ascending orbit (6 PM local solar time) and a descending orbit (6 AM local solar time). This embodiment uses SMAP Level 3 data version V5, with a spatial resolution of 36 km (based on an equal-area scalable Earth grid v2, EASE-Grid 2.0); the target accuracy of SMAP is an unbiased root mean square error of 0.04 m. 3 ·m -3 The data time series spans from 2015 to 2021.

[0057] S140: Obtain soil moisture content data for the VIC model.

[0058] Specifically, the VIC model is a large-scale land surface hydrological model based on water and energy balance. It is one of the most widely used hydrological models in my country, exhibiting high accuracy among various hydrological models. Driven by four types of parameters (geographic, vegetation, soil, and hydrological) and meteorological forcing data such as precipitation, surface temperature, and wind speed, the VIC model can simulate various surface variables such as soil moisture content and runoff. This embodiment uses daily precipitation and temperature data from 756 meteorological stations across China from the China Meteorological Administration's meteorological data network to drive the model, simulating the spatial distribution of soil moisture content across a 10km grid from 1956 to 2021.

[0059] S150: Obtain soil texture data.

[0060] Specifically, this involves collecting data from the SoilGrids soil texture database, constructed by the International Soil Reference and Information Centre based on global soil profile and environmental variable data. This dataset is driven by globally distributed soil profile data, correlated with environmental covariates, and utilizes geostatistical and machine learning algorithms to create global soil property and classification maps. In 2016, based on the geostatistical version 2.0 grid cell soil dataset, this data provides 250m of soil physical properties (such as sand content, clay content, silt content, bulk density, and coarse particle content), soil biochemical properties (such as soil osmotic pressure (SOM), soil pH, and cation exchange capacity), and soil classification maps.

[0061] S160: Obtain field water holding capacity data.

[0062] Specifically, the field water holding capacity data comes from a fusion grid field water holding capacity constructed by a professor team from Hohai University using multi-source soil texture data and on-site measured field water holding capacity data. The resolution is 250m, and its accuracy has been verified to be better than other field water holding capacity products on the market.

[0063] S170: Data on the area of ​​crops affected by drought.

[0064] Specifically, the data collected statistical data on the area of ​​crops affected by drought in various provinces reflects the actual situation of agricultural drought. The data comes from the China Water Resources Yearbook and the China Flood and Drought Disaster Bulletin, and the data period is from 2000 to 2020.

[0065] S200: Data preprocessing.

[0066] Specifically: Define the 10km grid where the VIC model is located as the baseline grid;

[0067] ASCAT active remote sensing soil moisture data and SMAP passive remote sensing soil moisture data were resampled and matched to the corresponding reference grid using the inverse distance weighted interpolation method.

[0068] The soil moisture content data observed at the stations were matched to the corresponding reference grid using the nearest neighbor method.

[0069] Soil texture data were matched to the corresponding baseline grid using a grid averaging method.

[0070] Field water holding capacity data were matched to the corresponding baseline grid using a grid averaging method.

[0071] S200: Data recalibration.

[0072] The cumulative probability distribution (CDF) of soil moisture content was performed grid by grid. CDF functions were constructed for ASCAT active remote sensing soil moisture data, SMAP passive remote sensing soil moisture data, and VIC model soil moisture data, respectively. The benchmark data matched by the CDF function was the VIC soil moisture content of the grid.

[0073] By matching using the CDF function, the soil moisture values ​​of the grid's ASCAT soil moisture content, SMAP soil moisture content, and VIC simulated soil moisture content at the same cumulative frequency can be matched together, so that the three have the same cumulative distribution frequency curve. This is used to remove the systematic bias between remote sensing and simulation, and to keep the remote sensing and simulated soil moisture content at the same amplitude.

[0074] This represents the soil moisture observation value at time t located in grid k of VIC, and its cumulative frequency P. k,t The following formula can be used for calculation:

[0075]

[0076] in, Indicates that the total number of grids k is The observed time series is less than The soil moisture content value.

[0077] Based on the above formula, the cumulative frequency distribution of observed and VIC simulated soil moisture content at each time point can be calculated. Then, by using linear interpolation, the remotely sensed soil moisture content values ​​at each cumulative frequency can be interpolated to the cumulative frequency corresponding to the simulated soil moisture content.

[0078] The CDF matching method can be used to rescale ASCAT and SMAP soil moisture data and correct them to the corresponding VIC soil moisture numerical space.

[0079] S400: Seasonal data fusion.

[0080] Specifically, the Triple Collocation algorithm (TC algorithm) is widely used for error variance estimation of remote sensing products, such as remote sensing soil moisture products. In the TC algorithm, three sets of consistent and independent data for the research object are required. Therefore, the ASCAT soil moisture content, SMAP soil moisture content and VIC model soil moisture content in the same reference grid are used as a set of data for the TC algorithm for each grid.

[0081] First, define the true soil moisture content at a certain time in a certain grid as θ. Then, define the soil moisture content in ASCAT, SMAP, and VIC models as θ1, θ2, and θ3, respectively. The linear relationships between θ and θ1, and between θ2 and θ3 are as follows:

[0082]

[0083] Where β1, β2 and β3 are error deviation terms, α1, α2 and α3 are scaling coefficient terms, and ε1, ε2 and ε3 are unbiased random error terms.

[0084] Then, by eliminating θ through pairwise equations, we can obtain the formula for calculating the error variance:

[0085]

[0086] in, Let i be the variance of the error corresponding to data product i. These are time series values ​​scaled according to the reference data; <> represents the mean calculation.

[0087] Next, the error variances between the ASCAT soil moisture content, SMAP soil moisture content, and VIC model soil moisture content in the grid and the actual soil moisture content were calculated. Furthermore, the correlation coefficients between the ASCAT soil moisture content, SMAP soil moisture content, and VIC model soil moisture content and the actual soil moisture content were derived.

[0088]

[0089] Where i, j, and k represent the ASCAT soil moisture dataset, the SMAP soil moisture dataset, and the VIC model soil moisture dataset, respectively, R represents the correlation coefficient, and σ i,j Represents the covariance of dataset i and dataset j. This represents the error variance of dataset i.

[0090] Finally, a fusion weight w was constructed for the soil moisture content data from ASCAT active remote sensing, SMAP passive remote sensing, and VIC model. i :

[0091]

[0092] Considering that vegetation, temperature, and underlying surface conditions will change in different seasons, leading to changes in the error characteristics of active remote sensing soil moisture content, passive remote sensing soil moisture content, and model-simulated soil moisture content, in S300, the fusion period needs to be divided into four seasons: spring, summer, autumn, and winter.

[0093] The correlation coefficients between the observed soil moisture content at each station and the soil moisture content of ASCAT, SMAP, VIC, and fused soil moisture were calculated, and box plots of the correlation coefficients for all stations were drawn, as shown in Figure 2. It can be seen that the accuracy of the fused soil moisture content is improved to varying degrees compared with the original soil moisture content of ASCAT, SMAP, and VIC.

[0094] S500: Constructing a drought index.

[0095] The Soil Moisture Index (SMI) is an indicator of soil drought, directly reflecting the availability of water for crops and thus indicating agricultural drought conditions. The SMI is calculated as the percentage of soil moisture content to field capacity. A higher SMI indicates a better supply of water to crops, meaning the soil is relatively moist; a lower SMI indicates a worse supply of water to crops, meaning the soil is relatively dry.

[0096] The soil relative humidity index (SMI) is constructed based on the integrated soil moisture content, and is used as a drought index for agricultural drought monitoring; the calculation formula for SMI is as follows:

[0097]

[0098] Where SM is the combined soil moisture content and FC is the field holding capacity.

[0099] The soil relative humidity index comprehensively considers soil water-holding characteristics and dynamic features of soil moisture content, offering advantages such as clear physical meaning, simple data requirements, and accurate identification and quantification of drought. By fusing multi-source soil moisture content data to obtain gridded soil moisture content distribution, it offers better spatiotemporal continuity and large-scale drought monitoring capabilities compared to site-based soil moisture content, making it widely applicable to large-scale agricultural drought monitoring.

[0100] S600: Drought Identification and Drought Level Classification

[0101] Based on the obtained data on sand content, clay content, and silt content of the grid, and referring to the national standard "Soil Classification and Grading", the grid soil types are divided as follows:

[0102] If the sand content accounts for more than 50% of the total mass, the clay content accounts for less than 15% of the total mass, and the silt content accounts for between 15% and 50% of the total mass, then it is considered sandy soil.

[0103] If the sand content accounts for 25% to 50% of the total mass, the clay content accounts for 15% to 40% of the total mass, and the silt content accounts for less than 50% of the total mass, then it is considered loam.

[0104] If the sand content accounts for less than 40% of the total mass, the clay content accounts for more than 40% of the total mass, and the silt content accounts for less than 20% of the total mass, then it is clay.

[0105] Based on different soil types, the drought level classification standards corresponding to the soil relative humidity index under different soil types are shown in Table 1.

[0106] Table 1. Classification criteria for drought level corresponding to SMI drought index under different soil types.

[0107]

[0108] S700: Assessment of the effectiveness of drought monitoring.

[0109] Based on the obtained data on the area of ​​crops affected by drought in various provinces over the years, the effectiveness of drought monitoring was verified and evaluated.

[0110] The grids of the corresponding provinces are extracted from the provincial administrative division map, and the drought index time series of the corresponding grids are calculated. Drought identification and drought level classification under different soil types of the corresponding provinces are completed, and drought events and drought duration are determined.

[0111] Based on the determination of drought events and drought duration, the intensity of drought over the years is calculated to determine the severity of drought.

[0112] Based on the ranking of provinces by annual crop yield, the top 80% of provinces are designated as major crop-producing provinces, and the time series of drought-affected crop area and drought intensity over the years are calculated for all major crop-producing provinces.

[0113] A scatter plot of the area of ​​crops affected by drought and the intensity of drought was drawn, as shown in Figure 3. The correlation coefficient was calculated, and the effectiveness of the agricultural drought monitoring method in this embodiment was evaluated.

[0114] The results show that the agricultural drought monitoring method based on soil moisture content fusion in this embodiment has a correlation coefficient of up to 0.88 with the actual drought situation. Compared with other existing large-scale agricultural drought monitoring methods, it has the advantages of accuracy, efficiency and reliability. It can be applied to actual agricultural drought monitoring and provide scientific basis and technical support for drought prevention and control.

[0115] Compared with the prior art, this application has at least the following beneficial technical effects:

[0116] (1) Compared with the uncertainty of accuracy that may be caused by using remote sensing soil moisture data or model soil moisture data from a single source, this application adopts a seasonal data fusion method, which takes into account the influence of vegetation, temperature and underlying surface on the fusion process in different seasons. This not only solves the problems of insufficient depth and spatiotemporal discontinuity in remote sensing soil moisture monitoring, but also solves the problems of simulation deviation caused by irrigation and poor simulation accuracy in areas without data in model soil moisture, thus improving the monitoring accuracy of soil moisture.

[0117] (2) This application can conduct real-time monitoring of large-scale agricultural drought on a national scale by acquiring and fusing remote sensing and model data in real time. Based on high-precision fusion of soil moisture content data and field water holding capacity data, a drought index considering soil moisture dynamics and soil water holding capacity is constructed. This index fully considers the soil type of the underlying surface, achieving more accurate, efficient and reliable monitoring of agricultural drought.

[0118] (3) This application can be used to monitor agricultural drought in real time or near real time, provide scientific guidance for drought prevention and control, and has broad application prospects.

[0119] (4) This application adopts a multi-source data fusion approach. First, by constructing a cumulative distribution function of multi-source soil moisture data, the multi-source soil moisture data is corrected. Then, the multi-source soil moisture data is fused using the TC algorithm, which solves the problems of spatiotemporal discontinuity and simulation bias. In addition, the drought classification threshold is constructed to identify drought categories, which can effectively improve the accuracy of soil moisture monitoring and the reliability of drought monitoring.

[0120] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and substitutions can be made without departing from the technical principles of the present invention, and these improvements and substitutions should also be considered within the scope of protection of the present invention.

Claims

1. A method for monitoring agricultural drought based on multi-source soil moisture content fusion, characterized in that, The process includes the following steps: S100: Acquire active remote sensing soil moisture data, passive remote sensing soil moisture data, and VIC model soil moisture data; S200: Using the grid corresponding to the VIC model soil moisture data as the reference grid, match the active remote sensing soil moisture data and passive remote sensing soil moisture data to the reference grid; S300: Construct the cumulative distribution function of soil moisture data grid by grid, and perform scale recorrection on the active remote sensing soil moisture data and passive remote sensing soil moisture data based on the VIC model soil moisture data to obtain active remote sensing soil moisture corrected data and passive remote sensing soil moisture corrected data; S400: Use the TC algorithm to perform multi-source soil moisture fusion on the active remote sensing soil moisture corrected data, passive remote sensing soil moisture corrected data, and VIC model soil moisture data according to the season to obtain fused soil moisture data; S500: Based on the fused soil moisture data... S600: Calculate the soil relative humidity index based on the soil relative humidity index and soil texture data to achieve agricultural drought monitoring; S300 specifically includes: firstly, constructing the cumulative distribution function of soil moisture content for each grid of the active remote sensing soil moisture content data, passive remote sensing soil moisture content data, and VIC model soil moisture content data; then, using the VIC model soil moisture content data corresponding to the baseline grid as the baseline data, interpolating the remote sensing soil moisture content values ​​at each cumulative frequency to the cumulative frequency corresponding to the simulated soil moisture content through linear interpolation, so that the active remote sensing soil moisture content data, passive remote sensing soil moisture content data, and VIC model soil moisture content data have the same cumulative distribution frequency curve, realizing the scale recorrection of the remote sensing soil moisture content data, and obtaining the active remote sensing soil moisture content corrected data and the passive remote sensing soil moisture content corrected data.

2. The agricultural drought monitoring method based on multi-source soil moisture content fusion according to claim 1, characterized in that, Specifically, S100 includes: acquiring soil texture data, field water holding capacity data, ASCAT active remote sensing soil moisture content data, SMAP passive remote sensing soil moisture content data, and VIC model soil moisture content data.

3. The agricultural drought monitoring method based on multi-source soil moisture content fusion according to claim 2, characterized in that, Specifically, S200 includes: using the grid corresponding to the soil moisture content data of the VIC model as the reference grid, resampling and matching the ASCAT active remote sensing soil moisture content data and SMAP passive remote sensing soil moisture content data to the reference grid through the inverse distance weighted interpolation method; matching the soil texture data to the reference grid through the grid averaging method; and matching the field water holding capacity data to the reference grid through the grid averaging method.

4. The agricultural drought monitoring method based on multi-source soil moisture content fusion according to claim 1, characterized in that, Specifically, S400 includes: dividing the data into four seasons, taking the active remote sensing soil moisture content correction data, passive remote sensing soil moisture content correction data, and VIC model soil moisture content data from the same reference grid in each season as a group of data to be fused, constructing fusion weights for the soil moisture content data using the TC algorithm, and then performing multi-source soil moisture content fusion on the active remote sensing soil moisture content correction data, passive remote sensing soil moisture content correction data, and VIC model soil moisture content data based on the fusion weights to obtain the fused soil moisture content data.

5. The agricultural drought monitoring method based on multi-source soil moisture content fusion according to claim 4, characterized in that, The fusion weights are constructed as follows: First, define the true soil moisture content of the grid as θ, the active remote sensing soil moisture content as θ1, the passive remote sensing soil moisture content as θ2, and the VIC model soil moisture content as θ3. The linear relationships between θ and θ1, and between θ2 and θ3 are as follows: Where β1, β2, and β3 are error bias terms, α1, α2, and α3 are scaling coefficient terms, and ε1, ε2, and ε3 are unbiased random error terms; then, by eliminating θ through pairwise equations, the formula for calculating the error variance can be obtained: in, Let i be the variance of the error corresponding to data product i. The values ​​are time series values ​​scaled according to the reference data; <> represents the mean calculation. The variances between the active remote sensing soil moisture content, passive remote sensing soil moisture content, and VIC model soil moisture content and the actual soil moisture content are calculated separately. The correlation coefficients between the active remote sensing soil moisture content, passive remote sensing soil moisture content, and VIC model soil moisture content and the actual soil moisture content are then obtained. Where i, j, k represent the datasets of active remote sensing soil moisture content, passive remote sensing soil moisture content, and VIC model soil moisture content, respectively; R represents the corresponding correlation coefficient; σ i,j Represents the covariance of dataset i and dataset j. The error variance of dataset i is represented; finally, the fusion weight w is constructed based on the correlation coefficient of the active remote sensing soil moisture correction dataset, the passive remote sensing soil moisture correction dataset, and the VIC model soil moisture data. i for:

6. The agricultural drought monitoring method based on multi-source soil moisture content fusion according to claim 1, characterized in that, The soil texture data described in S600 includes sand content, clay content, and silt content. The soil type classification criteria are as follows: if the sand content is greater than 50%, the clay content is less than 15%, and the silt content is between 15% and 50%, the soil type is sandy soil; if the sand content is between 25% and 50%, the clay content is between 15% and 40%, and the silt content is less than 50%, the soil type is loam; if the sand content is less than 40%, the clay content is greater than 40%, and the silt content is less than 20%, the soil type is clay.

7. The agricultural drought monitoring method based on multi-source soil moisture content fusion according to claim 6, characterized in that, The criteria for classifying drought levels as described in S600 are as follows: 。

Images