A regional soil moisture condition prediction method and system based on numerical weather prediction

Abstract

The application discloses a kind of regional soil moisture content prediction method and system based on numerical weather prediction, it is related to farmland moisture content prediction technical field.The present application discloses a kind of regional soil moisture content prediction method and system based on numerical weather prediction, it is related to farmland moisture content prediction technical field.The present application discloses a kind of regional soil moisture content prediction method and system based on numerical weather prediction, it is related to farmland moisture content prediction technical field.The present application discloses a kind of regional soil moisture content prediction method and system based on numerical weather prediction, it is related to farmland moisture content prediction technical field.The present application discloses a kind of regional soil moisture content prediction method and system based on numerical weather prediction, it is related to farmland moisture content prediction technical field.The present application discloses a kind of regional soil moisture content prediction method and system based on numerical weather prediction, it is related to farmland moisture content prediction technical field.The present application discloses a kind of regional soil moisture content prediction method and system based on numerical weather prediction, it is related to farmland moisture content prediction technical field.The present application discloses a kind of regional soil moisture content prediction method and system based on numerical weather prediction, it is related to farmland moisture content prediction technical field.The present application discloses a kind of regional soil moisture content prediction method and system based on numerical weather prediction, it is related to farmland moisture content prediction technical fields.

Description

Technical Field

[0001] This invention relates to the field of farmland soil moisture prediction technology, and in particular to a regional soil moisture prediction method and system based on numerical weather prediction. Background Technology

[0002] Regional soil moisture is a crucial variable connecting meteorological, hydrological, soil, and agricultural production processes. It directly reflects the soil moisture supply status of farmland and serves as essential foundational data for agricultural drought mitigation, irrigation water allocation, crop water requirement forecasting, yield assessment, and the construction of digital twin irrigation districts. In recent years, influenced by climate change, extreme droughts, heavy rainfalls, and periodic waterlogging events have become frequent, making drought and flood disasters major types of natural disasters affecting agricultural production security and stable food supply. Therefore, timely understanding of the current status of regional soil moisture and prediction of future moisture trends are of great significance for achieving early warning, early decision-making, and rational allocation of irrigation district water resources.

[0003] With the development of numerical weather prediction technology, regional soil moisture is mainly obtained through moisture prediction methods based on meteorological forecasts. The implementation process includes: acquiring measured data from ground moisture stations and the spatial distribution of soil volumetric water content provided by land surface assimilation products such as CLDAS; eliminating regional systematic errors through bias correction to construct a high-precision initial moisture field; secondly, acquiring forecast data of meteorological elements such as precipitation, maximum / minimum temperature, and relative humidity for the next 1-10 days provided by numerical weather prediction products such as GRAPES_GFS; and then inputting the corrected initial moisture field and meteorological forecast-driven data into a soil moisture dynamic simulation model or a crop water requirement model for daily rolling simulation to obtain the regional future moisture forecast results.

[0004] However, in existing soil moisture prediction methods based on weather forecasts, the calculation of crop evapotranspiration often relies on relatively complete meteorological data input. When meteorological data is insufficient or there are significant differences in climate conditions in different regions, the model parameters lack localized correction, resulting in inaccurate estimation of evapotranspiration and reducing the accuracy of regional soil moisture prediction results. Summary of the Invention

[0005] Therefore, it is necessary to provide a method and system for predicting regional soil moisture based on numerical weather prediction to address the aforementioned technical problems.

[0006] This invention provides a method for predicting regional soil moisture based on numerical weather prediction, comprising: Acquire multi-source heterogeneous data for the study area, including: current soil moisture data, numerical weather prediction data, field water holding capacity data, and crop planting distribution data. The numerical weather prediction data includes: rainfall data, maximum temperature data, and minimum temperature data. A deviation correction model is constructed to correct the deviation of the current soil moisture data so that the current soil moisture data is close to the true state of the study area, thus obtaining the baseline soil moisture data. Based on numerical weather prediction data, the reference crop evapotranspiration is determined by the Hargreaves model, and the crop coefficients for different blocks and times within the study area are determined by combining crop planting distribution data and a pre-set crop coefficient database. The reference crop evapotranspiration and the crop coefficients are multiplied pixel by pixel to obtain a spatial distribution map of crop evapotranspiration in the future preset time period. Using baseline soil moisture data as the initial soil moisture content, rainfall data as the water supply item, and crop evapotranspiration at different time points as the water consumption item, the daily soil moisture content for the future preset time period is obtained through daily rolling updates. Based on field water holding capacity data, daily soil moisture content is converted into relative moisture content, and the corresponding drought level is determined according to the relative moisture content, resulting in a drought risk prediction distribution map of the study area.

[0007] Optionally, after acquiring multi-source heterogeneous data of the study area, a data preprocessing step is also included, specifically including: Spatial clipping of multi-source heterogeneous data is performed based on the boundary vector of the study area to adjust the coverage of various types of data to the same range. Multi-source heterogeneous data with different spatial resolutions are resampled to the target raster grid, and multi-source heterogeneous data with different coordinate systems are projected to the same spatial reference system. Time series data from multiple sources are uniformly aligned to ensure that each type of data is processed pixel-by-pixel within the same spatial raster grid system, resulting in preprocessed multi-source heterogeneous data.

[0008] Optionally, a deviation correction model is constructed, specifically including: Acquire historical soil moisture data for the study area at historical moments, as well as historical measured soil moisture data observed at ground moisture stations or manual sampling points at the same historical moment; Using historical measured soil moisture data as the dependent variable and historical soil moisture data as the independent variable, a mapping relationship between historical soil moisture data and historical measured soil moisture data is constructed by combining a quantile mapping model and a linear regression model. This is used to correct regional deviations in soil moisture among different blocks within the study area, resulting in a deviation correction model.

[0009] Optionally, the preset crop coefficient database includes: different crop types and their crop coefficients at different growth stages; By combining crop planting distribution data and a pre-set crop coefficient database, crop coefficients for different blocks and times within the study area are determined. The reference crop evapotranspiration is multiplied pixel-by-pixel with the crop coefficient to obtain a spatial distribution map of crop evapotranspiration over a future pre-set time period, specifically including: Identify the crop types and current growth stages of each block within the study area based on crop planting distribution data; The surface reflectance images of the study area from the Sentinel-2 satellite during the corresponding growth period were acquired, and qualified images with cloud cover below a set threshold were selected. The normalized difference vegetation index was determined based on the near-infrared and red light bands in the qualified images. For different crop types, the normalized difference vegetation index is converted into the basic crop coefficient through the linear conversion rule between the normalized difference vegetation index and the basic crop coefficient, and the soil evaporation coefficient is determined according to the current growth stage. The basic crop coefficient and the soil evaporation coefficient are superimposed to obtain the crop coefficient for each type of crop. Interpolation or missing data completion processing is performed on the crop coefficients in spatially discontinuous areas within the study area to obtain a spatial distribution map of crop coefficients within the forecast time period; the spatial distribution map of crop coefficients is multiplied pixel by pixel with the reference crop evapotranspiration to obtain a spatial distribution map of crop evapotranspiration within the future preset time period.

[0010] Optionally, the baseline soil moisture data is used as the initial soil moisture content, the rainfall data is used as the water recharge item, and the crop evapotranspiration at different time points is used as the water consumption item. The daily soil moisture content for the future preset period is obtained through a rolling daily update method, specifically including: The target soil layer depth is determined based on the soil layer range corresponding to the current soil moisture data, the depth of crop root distribution, or the preset predicted layer. For each day within the forecast period, the following iterative operation is performed based on the principle of water balance: the soil moisture content of the previous day is used as the starting state, the rainfall of the day is added to the soil moisture system as a water supply item, the evapotranspiration of the crop on the day is deducted from the soil moisture system as a water consumption item, and the soil moisture content is homogenized throughout the entire soil depth to obtain the updated soil moisture content of the day. Based on the following formula, the updated daily soil moisture content is used as the starting state for the next day's iteration. The daily soil moisture content is iterated day by day to obtain the predicted raster data of soil moisture content for the next preset time period: SWC _{ n +1}=( SWC _ n × h + P _ n - ETc _ n ) / h ; in, SWC _ n For the first n Daily soil moisture content h The depth of the soil layer. SWC _{ n +1} is the first n +1 day soil moisture content, P _ n For the first n Daily rainfall data, ETc _ n For the first n Daily crop evapotranspiration.

[0011] Optionally, based on field water holding capacity data, daily soil moisture content is converted into relative moisture content, and the corresponding drought level is determined according to the relative moisture content to obtain a drought risk prediction distribution map of the study area, specifically including: Obtain field water holding capacity data for the corresponding predicted soil depth in the study area; For each forecast day within the forecast period, the predicted soil moisture content for that day is compared with the field water holding capacity data at the same location to obtain the spatial distribution of relative moisture content for that day. Based on the appropriate soil moisture control index for crop growth period, a classification relationship between relative water content and drought level is preset, with drought levels including no drought, mild drought, moderate drought and severe drought. The spatial distribution of daily relative water content was classified according to the hierarchical correspondence to obtain the daily drought risk prediction distribution map of the study area during the forecast period.

[0012] This invention provides a regional soil moisture prediction system based on numerical weather prediction, comprising: The data acquisition module is used to acquire multi-source heterogeneous data of the study area. The multi-source heterogeneous data includes: current soil moisture data, numerical weather prediction data, field water holding capacity data, and crop planting distribution data. The numerical weather prediction data includes: rainfall data, maximum temperature data, and minimum temperature data. The deviation correction module is used to construct a deviation correction model, which corrects the deviation of the current soil moisture data to make the current soil moisture data closer to the real state of the study area, thus obtaining the baseline soil moisture data. The evapotranspiration module is used to determine the reference crop evapotranspiration based on numerical weather prediction data and the Hargreaves model, and to determine the crop coefficients for different blocks and times within the study area by combining crop planting distribution data and a preset crop coefficient database; the reference crop evapotranspiration and the crop coefficients are multiplied pixel by pixel to obtain a spatial distribution map of crop evapotranspiration in the future preset time period. The soil moisture content prediction module is used to take the baseline soil moisture data as the initial soil moisture content, the rainfall data as the water supply item, and the crop evapotranspiration at different time points as the water consumption item, and obtain the daily soil moisture content in the future preset period through daily rolling updates. The drought level prediction module is used to convert daily soil moisture content into relative moisture content based on field water holding capacity data, and determine the corresponding drought level according to the relative moisture content, thereby obtaining a drought risk prediction distribution map of the study area.

[0013] The regional soil moisture prediction method and system based on numerical weather prediction provided in this invention have the following advantages compared with the prior art: This invention constructs a differentiated generation mechanism for crop coefficients in spatial and temporal dimensions by combining crop planting distribution data and a pre-set crop coefficient database. It achieves pixel-by-pixel fine-grained assignment based on actual planting structure and growth stage differentiation, and completes dynamic correction of localized parameters in the evapotranspiration calculation process. This effectively overcomes the limitation of poor applicability of soil moisture prediction methods based on weather forecasts in different climate zones, making water consumption estimation more consistent with the actual regional evapotranspiration patterns and significantly improving the accuracy of regional soil moisture prediction results. Attached Figure Description

[0014] Figure 1 This is a flowchart illustrating a regional soil moisture prediction method based on numerical weather prediction provided in one embodiment; Figure 2 This is a data processing flowchart of a regional soil moisture prediction method based on numerical weather prediction provided in one embodiment; Figure 3 This is a schematic diagram of the forecast output of a regional soil moisture prediction method based on numerical weather prediction provided in one embodiment. Detailed Implementation

[0015] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0016] This invention provides a regional soil moisture prediction method based on numerical weather prediction to improve the accuracy of the soil moisture baseline field, enhance the ability to predict soil moisture for the next 1-10 days, and achieve continuous spatial prediction of regional soil moisture and output drought level forecasts. Figure 2 As shown, the method includes:

[0017] The study area was acquired using multi-source heterogeneous data, including current soil moisture data, numerical weather prediction data, field water holding capacity data, and crop planting distribution data. The numerical weather prediction data included rainfall data, maximum temperature data, and minimum temperature data.

[0018] A bias correction model is constructed to correct the bias in the current soil moisture data so that the current soil moisture data closely reflects the true state of the study area, thus obtaining baseline soil moisture data.

[0019] Based on numerical weather prediction data, a reference crop evapotranspiration was determined using the Hargreaves model. Combined with crop planting distribution data and a pre-defined crop coefficient database, crop coefficients for different blocks and times within the study area were determined. The reference crop evapotranspiration and the crop coefficients were multiplied pixel-by-pixel to obtain a spatial distribution map of crop evapotranspiration over a pre-defined future time period.

[0020] Using baseline soil moisture data as the initial soil moisture content, rainfall data as the water supply, and crop evapotranspiration at different time points as the water consumption, the daily soil moisture content for the future preset time period is obtained through daily rolling updates.

[0021] Based on field water holding capacity data, daily soil moisture content is converted into relative moisture content, and the corresponding drought level is determined according to the relative moisture content, resulting in a drought risk prediction distribution map of the study area.

[0022] Preferably, after acquiring multi-source heterogeneous data of the study area, a data preprocessing step is also included, specifically including: spatially clipping the multi-source heterogeneous data based on the boundary vector of the study area to adjust the coverage of various data types to the same range; resampling the multi-source heterogeneous data with different spatial resolutions onto the target raster grid; and projecting the multi-source heterogeneous data with different coordinate systems onto the same spatial reference frame. The time series of the multi-source heterogeneous data are uniformly aligned to ensure that various data types are processed pixel-by-pixel within the same spatial raster grid system, resulting in preprocessed multi-source heterogeneous data.

[0023] Preferably, a deviation correction model is constructed, specifically including: Historical soil moisture data for the study area at specific historical moments, as well as historical measured soil moisture data observed at ground moisture monitoring stations or manual sampling points at the same historical moment, were acquired. Using historical measured soil moisture data as the dependent variable and historical soil moisture data as the independent variable, a mapping relationship between historical soil moisture data and historical measured soil moisture data was constructed through a combination of quantile mapping and linear regression models. This was used to correct regional moisture biases between different blocks within the study area, resulting in a bias correction model.

[0024] Preferably, the preset crop coefficient database includes: different crop types and their crop coefficients at different growth stages.

[0025] By combining crop planting distribution data and a pre-defined crop coefficient database, crop coefficients for different blocks and times within the study area are determined. The reference crop evapotranspiration is multiplied pixel-by-pixel with the crop coefficient to obtain a spatial distribution map of crop evapotranspiration over a future pre-defined time period, specifically including:

[0026] The crop types and current growth stages of each block within the study area were identified based on crop planting distribution data. Sentinel-2 satellite surface reflectance images of the study area at the corresponding growth stages were acquired, and qualified images with cloud cover below a set threshold were selected. The normalized difference vegetation index was determined based on the near-infrared and red bands in the qualified images.

[0027] For different crop types, the Normalized Difference Vegetation Index (NDV) is converted into a basic crop coefficient using a linear transformation rule between the NDC and the basic crop coefficient. The soil evaporation coefficient is then determined based on the current growth stage. The basic crop coefficient and the soil evaporation coefficient are then superimposed to obtain the crop coefficient for each crop type.

[0028] Interpolation or missing data completion is performed on crop coefficients in spatially discontinuous areas within the study region to obtain a spatial distribution map of crop coefficients for the forecast time period. The spatial distribution map of crop coefficients is then multiplied pixel-by-pixel with the reference crop evapotranspiration to obtain a spatial distribution map of crop evapotranspiration for the future preset time period.

[0029] Preferably, the baseline soil moisture data is used as the initial soil moisture content, the rainfall data is used as the water replenishment item, and the crop evapotranspiration at different time points is used as the water consumption item. The daily soil moisture content for the future preset period is obtained through a rolling daily update method, specifically including: The target soil depth is determined based on the soil layer range corresponding to the current soil moisture data, the depth of crop root distribution, or the preset predicted layer. For each day within the forecast period, the following iterative operation is performed based on the water balance principle: using the previous day's soil moisture content as the starting state, the current day's rainfall is added to the soil moisture system as a water supply item, and the current day's crop evapotranspiration is deducted from the soil moisture system as a water consumption item. This process is then homogenized across the entire soil depth to obtain the updated daily soil moisture content.

[0030] Based on the following formula, the updated daily soil moisture content is used as the starting state for the next day's iteration. The daily soil moisture content is iterated day by day to obtain the predicted raster data of soil moisture content for the next preset time period: SWC _{ n +1}=( SWC _n × h + P _ n - ETc _ n ) / h ; in, SWC _ n For the first n Daily soil moisture content h The depth of the soil layer. SWC _{ n +1} is the first n +1 day soil moisture content, P _ n For the first n Daily rainfall data, ETc _ n For the first n Daily crop evapotranspiration.

[0031] Preferably, based on field water holding capacity data, daily soil moisture content is converted into relative moisture content, and the corresponding drought level is determined according to the relative moisture content to obtain a drought risk prediction distribution map of the study area, specifically including: Field water holding capacity (FBC) data for the corresponding predicted soil depth in the study area were obtained. For each forecast day within the forecast period, the predicted soil moisture content for that day was compared with the FBC data at the same location to obtain the spatial distribution of relative moisture content for that day. Based on suitable soil moisture control indicators for crop growth stages, a hierarchical correspondence between relative moisture content and drought level was pre-defined, with drought levels including no drought, mild drought, moderate drought, and severe drought. The spatial distribution of daily relative moisture content was classified according to the hierarchical correspondence to obtain the daily drought risk prediction distribution map of the study area within the forecast period.

[0032] like Figure 1 As shown, a specific embodiment of the present invention is provided: S10 Multi-source data acquisition.

[0033] Acquire real-time CLDAS soil moisture data for the study area (as the initial source of current soil moisture values ​​for the region, used to provide current or historical soil moisture status), GRAPES_GFS numerical weather prediction data (including at least: rainfall data (from GRAPES_GFS precipitation forecast products, used to characterize water recharge), maximum temperature data, minimum temperature data, relative humidity and other relevant factors, used to provide input factors for the Hargreaves model in future periods), and crop coefficients K for different crop types. c Database (used for inverting crop coefficient K for specific growth stages of crop types in different regions) cField water holding capacity data (used to convert soil volumetric water content into relative water content) and crop planting distribution data (including the spatial distribution of major crop types such as wheat and corn, used to extract soil moisture forecast values ​​within crop planting areas using a mask).

[0034] The GRAPES_GFS numerical weather prediction data comes from the GRAPES_GFS global medium-range numerical weather prediction model system product (spatial resolution 0.25°×0.25°), and the forecast data includes precipitation, maximum temperature, minimum temperature, relative humidity, etc. for 1-10 days.

[0035] S20 Data Preprocessing and Bias Correction.

[0036] A. Data preprocessing.

[0037] Spatially clipping various data types based on the shapefile boundary vector of the study area ensures consistent coverage. Data with different spatial resolutions are uniformly resampled to the target resolution, and data from different coordinate systems are uniformly projected to the same spatial reference frame. Time-series data are aligned at the same time interval, and missing data checks, outlier handling, and quality control are performed to ensure that all types of data can be processed pixel-by-pixel within the same spatial raster grid system.

[0038] B. CLDAS baseline moisture deviation correction.

[0039] Since CLDAS gridded soil moisture data may have regional systematic biases, a linear bias correction model was constructed based on measured soil volumetric water content data and CLDAS soil moisture data to correct the CLDAS soil moisture data, thereby reducing systematic errors and obtaining baseline soil moisture data.

[0040] The reference data consists of ground-based soil moisture observations or higher-precision soil moisture products. The correction methods employed are linear regression correction and quantile mapping correction. Specifically, using the reference data as the true value, the CLDAS product is corrected using historical observation / site data from the same period or multi-year, multi-point statistical relationships to make the baseline soil moisture more closely reflect the actual regional conditions. By establishing a mapping relationship between CLDAS soil moisture and the true value, correction parameters are obtained and applied to the CLDAS grid in the study area to obtain the baseline soil moisture data SWC_0 (or as the initial water content field).

[0041] The linear regression model based on historical soil moisture data and measured soil moisture data is as follows: y =0.8057 x +0.0335 (r) 2 =0.8136); In the formula, x Historical soil moisture data, yThis is historical measured soil moisture data.

[0042] In this embodiment, K is determined for different blocks and different times within the study area. c Values, including: The study area and corresponding crop growth period were defined. Sentinel-2 imagery was loaded using the Google Earth Engine (GEE) platform, and qualified data with cloud cover below 20% were selected to obtain Sentinel-2 surface reflectance data (COPERNICUS / S2_SR). Then, the Normalized Difference Vegetation Index (NDVI) for different growth stages was calculated using the imagery's B8 and B4 bands. Based on crop planting distribution data, the crop type and current growth stage of the study area were identified, starting from K... c Match the corresponding K in the database c The value is then converted from NDVI to K using a linear formula. c Then, constrain the numerical range, and then select sample points to plot K. c Time series plots, generating visual K c Spatial distribution, output K c result.

[0043] The formula for calculating the Normalized Difference Vegetation Index (NDVI) is as follows: NDVI=(NIR Red) / (NIR Red) In the formula, NIR corresponds to the 8th band (B8) of Sentinel-2, and Red corresponds to the 4th band (B4).

[0044] Crop coefficient is divided into Kcb That is, the basic crop coefficient (related only to crop transpiration, dominated by the canopy) and K e That is, the soil evaporation coefficient (which is only related to surface exposure and soil moisture). Summer maize K c Calculation: Some scholars have proposed a concise linear equation for crops such as corn and soybeans, and the model validation results are very good (r 2 =0.910). It is also clearly stated that this linearity... K e The model performs consistently on irrigated maize, rainfed maize, and soybeans, and is applicable to regional-scale water resource management. K can be directly retrieved via MODIS NDVI. c It has been verified that summer corn Kcb The inversion model is as follows:

[0045] ; winter wheat Kcb The inversion model is as follows: 1); Kcb This refers to the basic crop coefficient, K. c and K cb The largest difference occurs during the initial growth stage, when transpiration is minimal and evaporation is dominant. During the mid-growth stage, as the crop canopy approaches or completely covers the ground, evaporation from below the canopy has a smaller impact on crop transpiration. K cb Value and K c Almost the same. Basic crop coefficient K cb Maybe only K c The value is small, ranging from 0.05 to 0.10, depending on the ground cover. High canopy cover is beneficial in the later stages of growth. ，K cb Value and K c Similar values, K c ≈ K cb 。 According to Allen Richard and others K cb With K c The following relationship exists:

[0046] ; Therefore, during the winter wheat seedling stage... K cb The value is increased by 0.1. K e During the overwintering period K cb The value is increased by 0.05, and the subsequent reproductive period is set by default. K cb Value and K c The values ​​are similar, that is, calculate the K value for winter wheat. c The method is to first calculate after obtaining the NDVI. K cb Then through K e Directly calculate K c .

[0047] S30 Evapotranspiration / Water Consumption Calculation.

[0048] Reference crop evapotranspiration ET0 is calculated based on the Hargreaves model and numerical weather prediction data. The ET0 calculation can utilize the T value predicted by GRAPES_GFS. max T min and radiation R a(or R can be estimated from latitude and longitude, annual accumulated days, and the relationship between outer atmospheric radiation) a ).

[0049] Based on crop coefficient K c By combining the database with information on crop planting distribution and timing in the study area, K was determined for different regions and at different times. c The value is used to obtain the crop evapotranspiration ET. c .

[0050] In this embodiment, crop evapotranspiration ET c It can be calculated as follows: ET c =ET0 K c ; Or use the following formula: ; Furthermore, historical data from multiple years and locations can be used to make localized corrections to the evapotranspiration calculation parameters, making the evapotranspiration estimation more consistent with regional climate and crop physiological characteristics.

[0051] S40 Soil Moisture Rolling Forecast for the Next 1-10 Days.

[0052] Based on baseline soil moisture data SWC_ 0 is used as the initial soil moisture content for the future... n The rainfall data is updated daily (n=1…10). P_n As a water supply item, crop evapotranspiration ET c _n As a water consumption item, the update yields... SWC_n ; and will SWC_n As the initial value for the next day's update, the daily SWC prediction raster for the next 1-10 days is obtained.

[0053] Furthermore, in practical implementation, upper and lower limits can be set for SWC to avoid non-physical outliers.

[0054] Obtain the daily SWC prediction raster and corresponding ET. c After grating, the predicted SWC is converted into relative water content (RWC) by combining it with the regional field water holding capacity spatial distribution map, and a regional soil moisture spatial distribution map is generated. The generated RWC or SWC gratings are then resampled and masked as necessary to obtain a spatially continuous and boundary-consistent daily soil moisture forecast product for the study area.

[0055] To ensure unit consistency, this embodiment uses 400mm (i.e., the planned wetting layer depth of 0-40cm) as the unit conversion constant for SWC and ET. cThe data is converted to the same dimension, where n is the number of days to be predicted.

[0056] Calculated pixel-by-pixel and day-by-day using the following formula: ; After the calculation is completed, for The results were resampled, and a mask was extracted according to the crop planting distribution map of the study area to obtain a raster of daily soil moisture changes in the study area for the next 1-10 days, outputting a spatially continuous soil moisture distribution map.

[0057] S50 drought level forecast.

[0058] like Figure 3 As shown, based on the appropriate soil moisture control index for crop growth period, a grading threshold system from RWC or SWC to drought level is established. The prediction results are divided into no drought, mild drought, moderate drought, and severe drought levels, and drought level forecast maps or regional statistical results are output to provide decision-making basis for agricultural drought resistance management.

[0059] As can be seen from the embodiments, the beneficial effects that the present invention can achieve include, but are not limited to: 1) By correcting the deviation of CLDAS soil moisture data, the accuracy of the soil moisture reference field is improved, and the systematic error is reduced from the source.

[0060] 2) By integrating GRAPES_GFS numerical weather forecast and rainfall forecast data and introducing future-driven approaches, dynamic soil moisture forecasts for 1-10 days can be achieved.

[0061] 3) Matching K based on crop distribution c Furthermore, localized corrections were made to make evapotranspiration estimates more suitable for different crops and growth stages, thereby improving the applicability of soil moisture forecasts.

[0062] 4) Combine field water holding capacity to calculate RWC and form a spatial continuous grid product, which can output high-precision regional soil moisture distribution maps and drought level forecast results, making it convenient for operational applications.

[0063] Based on the same inventive concept, embodiments of the present invention also provide a regional soil moisture prediction system, the system comprising: The data acquisition module is used to acquire multi-source heterogeneous data of the study area. The multi-source heterogeneous data includes: current soil moisture data, numerical weather prediction data, field water holding capacity data, and crop planting distribution data. The numerical weather prediction data includes: rainfall data, maximum temperature data, and minimum temperature data.

[0064] The deviation correction module is used to construct a deviation correction model, which corrects the deviation of the current soil moisture data to make the current soil moisture data closer to the actual state of the study area, thus obtaining baseline soil moisture data.

[0065] The evapotranspiration module is used to determine the reference crop evapotranspiration based on numerical weather prediction data using the Hargreaves model, and to determine the crop coefficients for different blocks and times within the study area by combining crop planting distribution data and a pre-set crop coefficient database. The reference crop evapotranspiration is then multiplied pixel-by-pixel with the crop coefficients to obtain a spatial distribution map of crop evapotranspiration for the future preset time period.

[0066] The soil moisture content prediction module uses baseline soil moisture data as the initial soil moisture content, rainfall data as the water supply item, and crop evapotranspiration at different time points as the water consumption item. It obtains the daily soil moisture content for a future preset period by updating the data daily.

[0067] The drought level prediction module is used to convert daily soil moisture content into relative moisture content based on field water holding capacity data, and determine the corresponding drought level according to the relative moisture content, thereby obtaining a drought risk prediction distribution map of the study area.

[0068] The embodiments described above are merely examples of several implementations of the present invention, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements all fall within the scope of protection of the present invention.

Claims

1. A method and system for predicting regional soil moisture based on numerical weather prediction, characterized in that, include: Acquire multi-source heterogeneous data for the study area, including: current soil moisture data, numerical weather prediction data, field water holding capacity data, and crop planting distribution data. The numerical weather prediction data includes: rainfall data, maximum temperature data, and minimum temperature data. A deviation correction model is constructed to correct the deviation of the current soil moisture data so that the current soil moisture data is close to the true state of the study area, thereby obtaining baseline soil moisture data. Based on the numerical weather forecast data, the reference crop evapotranspiration is determined by the Hargreaves model, and the crop coefficients for different blocks and times within the study area are determined by combining the crop planting distribution data and the preset crop coefficient database. The reference crop evapotranspiration is multiplied pixel by pixel to obtain a spatial distribution map of crop evapotranspiration in the future preset time period. Using the baseline soil moisture data as the initial soil moisture content, the rainfall data as the water supply item, and the crop evapotranspiration at different time points as the water consumption item, the daily soil moisture content for the future preset period is obtained by rolling updates every day. Based on the field water holding capacity data, the daily soil moisture content is converted into relative moisture content, and the corresponding drought level is determined according to the relative moisture content to obtain a drought risk prediction distribution map of the study area.

2. The method for predicting regional soil moisture based on numerical weather prediction as described in claim 1, characterized in that, After acquiring multi-source heterogeneous data of the study area, a data preprocessing step is also included, specifically: Spatial clipping of the multi-source heterogeneous data is performed based on the boundary vector of the study area to adjust the coverage of various types of data to the same range. The multi-source heterogeneous data with different spatial resolutions are resampled to the target raster grid, and the multi-source heterogeneous data with different coordinate systems are projected to the same spatial reference system; Time series data from multiple sources are uniformly aligned to ensure that each type of data is processed pixel-by-pixel within the same spatial raster grid system, resulting in preprocessed multi-source heterogeneous data.

3. The method for predicting regional soil moisture based on numerical weather prediction as described in claim 1, characterized in that, The construction of the deviation correction model specifically includes: Acquire historical soil moisture data for the study area at historical moments, as well as historical measured soil moisture data observed at ground moisture stations or manual sampling points at the same historical moment; Using the historical measured soil moisture data as the dependent variable and the historical soil moisture data as the independent variable, a mapping relationship between the historical soil moisture data and the historical measured soil moisture data is constructed by combining a quantile mapping model and a linear regression model. This is used to correct the regional deviation of soil moisture between different blocks in the study area, thus obtaining the deviation correction model.

4. The method for predicting regional soil moisture based on numerical weather prediction as described in claim 1, characterized in that, The preset crop coefficient database includes: different crop types and their crop coefficients at different growth stages; The process involves combining the crop planting distribution data with a pre-defined crop coefficient database to determine crop coefficients for different blocks and times within the study area; multiplying the reference crop evapotranspiration by the crop coefficients pixel-by-pixel to obtain a spatial distribution map of crop evapotranspiration over a future pre-defined time period, specifically including: The crop types and current growth stages of each block within the study area are identified based on the crop planting distribution data. The surface reflectance images of the study area from the Sentinel-2 satellite during the corresponding growth period were acquired, and qualified images with cloud cover below a set threshold were selected. The normalized difference vegetation index was determined based on the near-infrared and red light bands in the qualified images. For different crop types, the normalized difference vegetation index is converted into the basic crop coefficient through a linear conversion rule between the normalized difference vegetation index and the basic crop coefficient, and the soil evaporation coefficient is determined according to the current growth stage; the basic crop coefficient and the soil evaporation coefficient are superimposed to obtain the crop coefficient for each type of crop. Interpolation or missing data completion processing is performed on the crop coefficients in spatially discontinuous areas within the study area to obtain a spatial distribution map of crop coefficients within the forecast time period; the spatial distribution map of crop coefficients is multiplied pixel by pixel with the reference crop evapotranspiration to obtain a spatial distribution map of crop evapotranspiration within the future preset time period.

5. The method for predicting regional soil moisture based on numerical weather prediction as described in claim 1, characterized in that, The method of using the baseline soil moisture data as the initial soil moisture content, the rainfall data as the water replenishment item, and the crop evapotranspiration at different time points as the water consumption item, and obtaining the daily soil moisture content for a future preset period through a daily rolling update, specifically includes: The target soil layer depth is determined based on the soil layer range corresponding to the current soil moisture data, the depth of crop root distribution, or the preset predicted layer. For each day within the forecast period, the following iterative operation is performed based on the principle of water balance: the soil moisture content of the previous day is used as the starting state, the rainfall of the day is added to the soil moisture system as a water supply item, the evapotranspiration of the crop on the day is deducted from the soil moisture system as a water consumption item, and the soil moisture content is homogenized throughout the entire soil depth to obtain the updated soil moisture content of the day. Based on the following formula, the updated daily soil moisture content is used as the starting state for the next day's iteration. The daily soil moisture content is iterated day by day to obtain the predicted raster data of soil moisture content for the next preset time period: SWC _{ n +1}=( SWC _ n × h + P _ n － ETc _ n ) / h ； in, SWC _ n For the first n Daily soil moisture content h The depth of the soil layer. SWC _{ n +1} is the first n +1 day soil moisture content, P _ n For the first n Daily rainfall data, ETc _ n For the first n Daily crop evapotranspiration.

6. The method for predicting regional soil moisture based on numerical weather prediction as described in claim 1, characterized in that, Based on the field water holding capacity data, the daily soil moisture content is converted into relative moisture content, and the corresponding drought level is determined according to the relative moisture content to obtain a drought risk prediction distribution map of the study area, specifically including: Obtain field water holding capacity data for the corresponding predicted soil depth in the study area; For each forecast day within the forecast period, the predicted soil moisture content for that day is compared with the field water holding capacity data at the same location to obtain the spatial distribution of relative moisture content for that day. Based on the appropriate soil moisture control index for crop growth period, a pre-defined correspondence between relative water content and drought level is established, wherein the drought level includes no drought, mild drought, moderate drought and severe drought. The spatial distribution of daily relative water content is classified according to the aforementioned hierarchical correspondence to obtain the daily drought risk prediction distribution map of the study area during the forecast period.

7. A regional soil moisture prediction system based on numerical weather prediction, characterized in that, include: The data acquisition module is used to acquire multi-source heterogeneous data of the study area. The multi-source heterogeneous data includes: current soil moisture data, numerical weather prediction data, field water holding capacity data, and crop planting distribution data. The numerical weather prediction data includes: rainfall data, maximum temperature data, and minimum temperature data. The deviation correction module is used to construct a deviation correction model, and to correct the deviation of the current soil moisture data through the deviation correction model, so that the current soil moisture data is close to the real state of the study area, and obtain the baseline soil moisture data. The evapotranspiration module is used to determine the reference crop evapotranspiration based on the numerical weather forecast data using the Hargreaves model, and to determine the crop coefficients for different blocks and times within the study area by combining the crop planting distribution data and a preset crop coefficient database; the reference crop evapotranspiration is multiplied pixel by pixel to obtain a spatial distribution map of crop evapotranspiration in the future preset time period; The soil moisture content prediction module is used to take the baseline soil moisture data as the initial soil moisture content, the rainfall data as the water supply item, and the crop evapotranspiration at different time points as the water consumption item, and obtain the daily soil moisture content in the future preset period by rolling updates. The drought level prediction module is used to convert the daily soil moisture content into relative moisture content based on the field water holding capacity data, and determine the corresponding drought level according to the relative moisture content to obtain a drought risk prediction distribution map of the study area.

Images