Method and device for determining irrigation water consumption in arid region

Abstract

The invention provides a method and device for determining the irrigation water consumption of an arid region, and relates to the technical field of agricultural irrigation, and the method comprises the steps: firstly obtaining various evapotranspiration data, ground observation rainfall data, satellite observation rainfall data and soil environment characteristic data of the arid region in an irrigation period; carrying out weighted fusion on the plurality of evapotranspiration data to obtain first data; correcting the satellite observation rainfall data pixel by pixel based on the ground observation data to obtain second data; inputting the soil environment characteristic data into a first deep learning model to obtain first non-evaporation loss data of the arid region in the irrigation period; then correcting the first non-evaporation loss data to obtain second non-evaporation loss data; and finally, based on the first data, the second data and the second non-evaporation loss data, determining the irrigation water consumption of the arid region in the irrigation period. According to the method for determining the irrigation water consumption of the arid region, accurate detection of the irrigation water consumption of the arid region is achieved.

Description

Technical Field

[0001] This invention relates to the field of agricultural irrigation technology, and in particular to a method and apparatus for determining irrigation water consumption in arid areas. Background Technology

[0002] Accurately detecting irrigation water usage in arid regions is crucial for ensuring sustainable agricultural and ecological development as well as water resource management.

[0003] The main method for detecting irrigation water consumption using existing technologies is to calculate irrigation water consumption by inverting the water balance method. Specifically, remote sensing observation is combined with water balance. First, hydrological variables such as precipitation, soil moisture content, and evapotranspiration are obtained, and then irrigation water consumption is derived based on the water balance formula.

[0004] However, the existing soil moisture content products for arid regions have relatively coarse resolution, and the performance of evapotranspiration products is unstable in different regions, leading to unreliable results in irrigation water consumption monitoring in arid areas. Furthermore, the water balance formula contains some variables that cannot currently be directly obtained through remote sensing, such as infiltration and groundwater recharge, resulting in insufficient data and consequently low accuracy in monitoring irrigation water consumption in arid regions. Summary of the Invention

[0005] This invention provides a method and apparatus for determining irrigation water consumption in arid areas, which solves the technical problem of inaccurate detection of irrigation water consumption in arid areas in the prior art, and realizes accurate detection of irrigation water consumption in arid areas.

[0006] This invention provides a method for determining irrigation water consumption in arid areas, comprising the following steps.

[0007] Acquire various evapotranspiration data, ground-based precipitation data, satellite-based precipitation data, and soil characteristic environmental data for arid regions during the irrigation period; The various evapotranspiration data are weighted and fused to obtain the first data; Based on the ground observation data, the satellite observation precipitation data is corrected pixel by pixel to obtain the second data; The soil characteristic environmental data is input into the first deep learning model to perform non-evaporation loss detection, and the first non-evaporation loss data of the arid area during the irrigation period is obtained from the output of the first deep learning model. The first non-evaporation loss data is corrected according to the vegetation type of the arid area to obtain the second non-evaporation loss data; Based on the first data, the second data, and the second non-evaporation loss data, the irrigation water consumption in the arid area during the irrigation period is determined.

[0008] According to the method for determining irrigation water consumption in arid areas provided by the present invention, before acquiring various evapotranspiration data, surface-observed precipitation data, satellite-observed precipitation data, and soil characteristic environmental data of the arid area during the irrigation period, the method further includes: Acquire satellite-observed surface temperature data for arid regions; The arid region was simulated using an Earth system model to obtain simulated surface temperature data. Based on the satellite-observed surface temperature data and the simulated surface temperature data, temperature difference data is obtained; Phenological data were obtained based on the leaf area index of the vegetation in the arid region. Based on the temperature difference data and the phenological data, the irrigation period and non-irrigation period of the arid area are determined.

[0009] According to the method for determining irrigation water consumption in arid areas provided by the present invention, the training steps of the first deep learning model include: Acquire various evapotranspiration data, ground-based precipitation data, satellite-based precipitation data, and soil environmental characteristic data for the arid region during the non-irrigation period; The various evapotranspiration data are weighted and fused to obtain the third data. Based on the ground observation data, the satellite observation precipitation data is corrected pixel by pixel to obtain the fourth data. Based on the third and fourth data, non-evaporation loss data of the arid area during the non-irrigation period are obtained; The first deep learning model is trained using the non-evaporation loss data as label data and the soil environmental characteristic data as feature data.

[0010] According to the present invention, a method for determining irrigation water consumption in arid areas includes soil environmental characteristic data such as evapotranspiration data, surface soil moisture content, leaf area index of vegetation, soil texture, soil slope, root depth of vegetation, saturated hydraulic conductivity, and soil porosity.

[0011] According to a method for determining irrigation water consumption in arid areas provided by the present invention, the step of obtaining the surface soil moisture content includes: Acquire high-resolution surface soil moisture characteristic data; the surface soil moisture characteristic data includes precipitation data, evapotranspiration data, surface temperature data, leaf area index of vegetation, soil elevation, and soil texture; The high-resolution surface soil moisture characteristic data is input into the second deep learning model for downscaling to obtain the surface soil moisture content output by the second deep learning model.

[0012] According to the method for determining irrigation water consumption in arid areas provided by the present invention, the training steps of the second deep learning model include: Acquire historical surface soil moisture characteristics data at first resolution and the corresponding surface soil moisture content observed by satellite; The resolution of the historical surface soil moisture characteristic data at the first resolution is adjusted to be consistent with the surface soil moisture content to obtain historical surface soil moisture characteristic data at the second resolution. The second deep learning model is trained using historical surface soil moisture characteristic data at the second resolution as feature data and the corresponding satellite-observed surface soil moisture content as label data.

[0013] The present invention also provides a device for determining irrigation water consumption in arid areas, comprising the following modules: The acquisition module is used to acquire various evapotranspiration data, ground-based precipitation data, satellite-based precipitation data, and soil environmental characteristic data in arid areas during the irrigation period; The fusion module is used to perform weighted fusion of the various evapotranspiration data to obtain the first data; The correction module is used to perform pixel-by-pixel correction on the satellite precipitation data based on the ground observation data to obtain the second data; The detection module is used to input the soil environmental characteristic data into the first deep learning model to perform non-evaporation loss detection and obtain the first non-evaporation loss data of the arid area during the irrigation period output by the first deep learning model. The correction module is used to correct the first non-evaporation loss data according to the vegetation type of the arid area to obtain the second non-evaporation loss data; The determination module is used to determine the irrigation water consumption of the arid area during the irrigation period based on the first data, the second data, and the second non-evaporation loss data.

[0014] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the method for determining irrigation water consumption in arid areas as described above.

[0015] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method for determining irrigation water consumption in arid areas as described above.

[0016] The present invention also provides a computer program product, including a computer program that, when executed by a processor, implements the method for determining irrigation water consumption in arid areas as described above.

[0017] This invention provides a method for determining irrigation water consumption in arid regions. First, it acquires various evapotranspiration data, ground-based precipitation data, satellite-based precipitation data, and soil environmental characteristic data for the arid region during the irrigation period. Then, it performs weighted fusion of the various evapotranspiration data to obtain first data, thereby improving the accuracy and reliability of the evapotranspiration data. Next, based on the ground-based observation data, it performs pixel-by-pixel correction on the satellite precipitation data to obtain second data, thereby improving the overall spatial accuracy and local reliability of the satellite precipitation data. Finally, it inputs the soil environmental characteristic data into a first deep learning model for non-evaporation loss detection to obtain the first deep learning model's output for the arid region during the irrigation period. First, non-evaporation loss data is used to accurately and quickly detect non-evaporation loss data in arid areas during the irrigation period through a model, achieving unbiased, data-driven inversion of key data and breaking through the data acquisition limitations of traditional remote sensing observations. Then, the first non-evaporation loss data is corrected according to the vegetation type of the arid area to obtain second non-evaporation loss data. This allows for differentiated correction of non-evaporation loss data for different vegetation types through vegetation classification information, improving the accuracy of non-evaporation loss data. Finally, based on the first, second, and second non-evaporation loss data, the irrigation water consumption in arid areas during the irrigation period is determined, achieving accurate detection of irrigation water consumption in arid areas and providing decision support for large-scale agricultural water resource management. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0019] Figure 1 This is a flowchart illustrating a method for determining irrigation water consumption in arid regions, provided by the present invention.

[0020] Figure 2 This is a comparative diagram of the statistical data provided by the present invention.

[0021] Figure 3 This is a schematic diagram of the irrigation identification results provided by the present invention.

[0022] Figure 4 This is a schematic diagram of the network structure of a long short-term memory neural network model provided by the present invention.

[0023] Figure 5 This is a schematic diagram of the structure of a device for determining irrigation water consumption in arid areas provided by the present invention.

[0024] Figure 6 This is a schematic diagram of the structure of the electronic device provided by the present invention. Detailed Implementation

[0025] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0026] The following is combined with Figures 1 to 6 This invention describes a method and apparatus for determining irrigation water consumption in arid regions.

[0027] Figure 1 This is a flowchart illustrating a method for determining irrigation water consumption in arid areas provided by the present invention, as shown below. Figure 1 As shown, the method includes the following steps: Step 101: Obtain various evapotranspiration data, ground-based precipitation data, satellite-based precipitation data, and soil environmental characteristic data for arid areas during the irrigation period.

[0028] Specifically, various evapotranspiration data in arid regions during the irrigation period can be acquired through a variety of remote sensing evapotranspiration data acquisition devices or models. These devices or models include the Moderate-resolution Imaging Spectrometer (MODIS), the Global Land Evaporation Amsterdam Model (GLEAM), and the Penman-Monteith-Leuning (PML).

[0029] Precipitation data observed on the ground during the irrigation period in arid areas can be collected from precipitation observations recorded at meteorological stations.

[0030] Satellite precipitation data during the irrigation season in arid regions can be collected through remote sensing inversion satellite precipitation products such as the Global Precipitation Measurement Mission (GPM) and the Tropical Rainfall Measuring Mission (TRMM).

[0031] Soil environmental characteristics data in arid areas during the irrigation period include evapotranspiration data, topsoil moisture content, leaf area index of vegetation, soil texture, soil slope, root depth of vegetation, saturated hydraulic conductivity, and soil porosity.

[0032] Step 102: Weight and fuse the various evapotranspiration data to obtain the first data.

[0033] Specifically, fusing data acquired by multi-source remote sensing evapotranspiration acquisition devices or models (MODIS, GLEAM, PML, etc.) can yield more stable and regionally adaptable evapotranspiration data, which is then used as the primary data.

[0034] Current mainstream remote sensing evapotranspiration products have different uncertainties due to the limitations of their own models. Therefore, this embodiment of the invention uses the Bayesian tripod method to perform weighted fusion of different evapotranspiration products.

[0035] The Bayesian triangular hat method can be used to fuse different models without any prior knowledge. It mainly consists of two parts: the classic triangular hat part, which is used to assess the uncertainty of different models, and the Bayesian part, which is used to weight each product.

[0036] The triangular hat method is used to assess the uncertainty of three or more data sequences. For example, for Given a set of different observed sequences, the relationship between each sequence and the true sequence can be represented as follows: In the formula, It is the sequence length. yes Observation sequence, It is a real sequence. yes and The difference between them. Because of... It is unknown, from Randomly select one of the observation sequences as the reference sequence. Then the reference sequence Other observation sequences The relationship between them can be represented as: In the formula, yes OK The difference matrix of columns, covariance matrix Defined as: In the formula, It is an identity matrix minus a column vector. yes OK The noise covariance matrix of the column, express The transpose of .

[0037] In the formula, , , ..., This represents the covariance matrix for each observation sequence.

[0038] objective function Defined to satisfy the noise covariance matrix Positive definiteness: In the formula, Constraints The initial conditions for the iteration are defined as follows: Building upon the triangular hat method, this invention further proposes using a Bayesian triangular hat approach to integrate different models. For example, for the first... The evaporation data provided by each evaporation product ( ), its probability density function It can be represented as: In the formula, It is the true value of evapotranspiration. It is white noise with zero mean. yes The error variance, This represents the likelihood equation.

[0039] for The evaporation data provided by the evaporation estimation product. The maximum likelihood equation can be expressed as: In order to calculate The maximum likelihood value, the cost equation can be expressed as: make If it is zero, then It can be represented as: In the formula, yes The weight.

[0040] The error variance of the evapotranspiration data for each evapotranspiration product can be obtained using the triangular cap method. Based on the error variance, multiple evapotranspiration data are weighted and fused to obtain more stable and regionally adaptable evapotranspiration data as the first data, thereby improving the accuracy and reliability of evapotranspiration data.

[0041] Based on the above embodiments, in some embodiments, the present invention may also employ other strategies for evapotranspiration product fusion, such as weighted average method (evaluating weights based on the historical performance of the model), ensemble learning methods based on machine learning (such as random forest regression, XGBoost weighted fusion), and Bayesian network models based on confidence propagation for dynamic fusion of time-series evapotranspiration products, etc., all of which can achieve the same or similar effects, and the present invention does not limit them.

[0042] Step 103: Based on the ground observation data, perform pixel-by-pixel correction on the satellite observation precipitation data to obtain the second data.

[0043] Specifically, precipitation observations recorded by meteorological stations can collect ground-based precipitation data for arid regions during the irrigation period. While meteorological station-recorded precipitation data is generally considered the most accurate and reliable ground-based data, its limited quantity and uneven spatial distribution result in insufficient spatial representation capabilities for large-scale arid regions. In contrast, satellite precipitation products retrieved through remote sensing (such as GPM and TRMM) offer better spatial continuity and wide-area coverage. However, due to limitations in satellite observation mechanisms, sensor performance, orbital characteristics, and uncertainties in retrieval algorithms, satellite precipitation data exhibits significant errors in accuracy and local representation. Therefore, this invention effectively fuses these two products, compensating for each other's shortcomings and achieving pixel-by-pixel correction of satellite precipitation data. This yields higher-quality and spatially continuous precipitation data as secondary data, supporting subsequent irrigation water usage retrieval. The specific correction process is as follows: (1) Calculate the precipitation error between meteorological stations and satellite observations. For any weather station and at any time Define precipitation error The difference between ground-based observation data and satellite-based precipitation data is expressed as: In the formula, For the first Each weather station at time The measured precipitation (ground observation data). The corresponding pixel precipitation data represents the precipitation observed by the satellite (satellite-observed precipitation data).

[0044] (2) Constructing the spatial error field Based on precipitation error data from all meteorological stations, a spatial interpolation method is used to extend the station-scale error field to the entire study area, obtaining precipitation error estimates for any location within the study area. The spatial interpolation methods include ordinary kriging interpolation, kriging interpolation with a trend term, inverse distance-weighted interpolation co-kriging interpolation, and spatial interpolation using neural networks (such as spatial convolutional networks or graph neural networks) to improve accuracy. In this embodiment, ordinary kriging interpolation is used as an example, and the basic form of the precipitation error estimate can be expressed as: In the formula, The spatial coordinates of the location to be estimated Represents the Kriging interpolation weighting coefficients. This represents the number of meteorological stations participating in the interpolation.

[0045] (3) Introduce adaptive weighted interpolation method To more accurately reflect the error characteristics of different meteorological stations, this embodiment of the invention further employs a weighted Kriging interpolation method, wherein the weight of each meteorological station is... The system dynamically adjusts the data based on spatial distance and local error characteristics. Specifically, adjustments based on spatial distance increase the weight of stations closer to the location to be interpolated and decrease it from the opposite location. Adjustments based on the magnitude of precipitation error at each station decrease the weight of a station with a large error (above a preset threshold) to prevent it from having an excessive impact on the error field.

[0046] This invention improves the interpolation process by introducing a distance attenuation function and an error adaptive mechanism, thereby enhancing accuracy. The specific weighting function is expressed as follows: In the formula, Representative weather stations To the position to be interpolated distance, The standardized error of a station is used to reflect the magnitude of the error at that meteorological station. In this way, the weights of stations can be adaptively adjusted, making the construction of the error field more flexible and accurate.

[0047] (4) Constructing integrated precipitation products The error field obtained by Kriging interpolation is superimposed on the satellite-observed precipitation data, and pixel-by-pixel correction is performed on each pixel in the satellite-observed precipitation data to obtain high-quality fused precipitation data as the second data. , is represented as: (5) Post-optimization processing To further improve the physical validity and numerical stability of the second data, it is also necessary to ensure that the corrected precipitation is non-negative. The following numerical optimizations can be performed: The embodiments of the present invention employ a method combining error correction and spatial interpolation to spatially expand the systematic deviation between the measured precipitation values ​​of meteorological stations and satellite precipitation products, thereby achieving pixel-by-pixel correction of satellite-observed precipitation data and improving the overall spatial accuracy and local reliability of satellite-observed precipitation data.

[0048] Step 104: Input the soil environmental characteristic data into the first deep learning model to perform non-evaporation loss detection, and obtain the first non-evaporation loss data of the arid area during the irrigation period output by the first deep learning model.

[0049] Specifically, since irrigation can be considered a special type of precipitation, the relationship between non-evaporation loss and soil environmental factors is stable between irrigation and non-irrigation periods. Based on this stable correspondence, a deep learning model can be established. Through training, the model can detect the corresponding non-evaporation loss based on objectively existing soil environmental factors. Non-evaporation loss mainly includes soil water storage (or soil moisture content) and infiltration.

[0050] This invention uses non-evaporation loss data in arid regions during the non-irrigation period, obtained through the water balance formula, as training labels and selects relevant soil environmental characteristic data as feature data to establish and train a first deep learning model. The trained first deep learning model is then used to detect non-evaporation loss during the irrigation period. The first deep learning model can be an artificial neural network, a backpropagation network, or a convolutional neural network, etc.

[0051] In this embodiment of the invention, an artificial neural network is selected as the first deep learning model for model training. After the model training is completed, soil environmental characteristic data during the irrigation period are input into the model to obtain the first non-evaporation loss data of the arid area during the irrigation period. Thus, the model can accurately and quickly detect the non-evaporation loss data of the arid area during the irrigation period, achieving unbiased, data-driven inversion of key data. This breaks through the data acquisition limitations of traditional remote sensing observation and solves the variable estimation problem that traditional remote sensing methods cannot obtain.

[0052] Step 105: Correct the first non-evaporation loss data according to the vegetation type of the arid area to obtain the second non-evaporation loss data.

[0053] Specifically, since natural vegetation plays a crucial role in the water cycle, especially in influencing soil moisture content and infiltration, and different vegetation types have different effects on soil moisture content and infiltration (for example, different vegetation types such as forests and grasslands have different effects on physical processes such as soil infiltration), it is necessary to consider the spatial distribution of vegetation, especially the influence of adjacent natural vegetation, when estimating non-evaporative losses.

[0054] To address this, the present invention designs a post-processing module that incorporates vegetation type in arid regions as an important environmental feature into the calculation of non-evaporative loss. Specifically, firstly, based on remote sensing image data, such as the Normalized Difference Vegetation Index (NDVI) or Leaf Area Index (LAI), vegetation types in arid regions are classified to identify different types of natural vegetation, such as forests, grasslands, and farmland. Then, based on vegetation water use efficiency, the relationship between vegetation type and soil moisture is combined, thereby determining the characteristics of different vegetation types. Correct the first non-evaporation loss data. Obtain the second non-evaporation loss data. : In the formula, This is the vegetation type correction factor.

[0055] This invention improves the accuracy of non-evaporation loss data by using vegetation classification information to differentiate and correct non-evaporation loss data for regions with different vegetation types.

[0056] Step 106: Based on the first data, the second data, and the second non-evaporation loss data, determine the irrigation water consumption of the arid area during the irrigation period.

[0057] Specifically, the irrigation amount during the irrigation period can be obtained using the complete water balance formula: In the formula, This represents the amount of irrigation water used in arid areas during the irrigation season. This represents the second non-evaporation loss data. Indicates the second data. This indicates the first data point.

[0058] Based on the above method, this invention takes a drought-stricken area from 2004 to 2013 as the research object, estimates the irrigation water consumption during the irrigation season in each year, and compares and verifies the estimated value with the actual irrigation water consumption in the records of each year. Figure 2 This is a comparative diagram of the statistical data provided by the present invention, such as... Figure 2 As shown, the irrigation water consumption estimate obtained by the irrigation water consumption inversion method for arid areas proposed in this invention has high accuracy and high consistency with the actual statistical data in historical records.

[0059] Based on first data, second data, and second non-evaporation loss data, this invention determines the irrigation water consumption in arid areas during the irrigation season by combining water balance formulas. It constructs a complete integrated inversion system of "remote sensing observation + deep learning + water balance" to achieve accurate detection of irrigation water consumption in arid areas, thereby providing decision support for large-scale agricultural water resource management.

[0060] This invention provides a method for determining irrigation water consumption in arid regions. First, it acquires various evapotranspiration data, ground-based precipitation data, satellite-based precipitation data, and soil environmental characteristic data for the arid region during the irrigation period. Then, it performs weighted fusion of the various evapotranspiration data to obtain first data, thereby improving the accuracy and reliability of the evapotranspiration data. Next, based on the ground-based observation data, it performs pixel-by-pixel correction on the satellite precipitation data to obtain second data, thereby improving the overall spatial accuracy and local reliability of the satellite precipitation data. Finally, it inputs the soil environmental characteristic data into a first deep learning model to perform non-evaporation loss detection, obtaining the first non-evaporation loss data of the arid region during the irrigation period output by the first deep learning model. The model accurately and rapidly detects non-evaporative loss data in arid regions during the irrigation period, achieving unbiased, data-driven inversion of key data and overcoming the data acquisition limitations of traditional remote sensing observations. Then, it corrects the first non-evaporative loss data based on the vegetation type of the arid region to obtain a second non-evaporative loss data. This allows for differentiated correction of non-evaporative loss data for different vegetation types through vegetation classification information, improving the accuracy of the non-evaporative loss data. Finally, based on the first, second, and third non-evaporative loss data, the irrigation water consumption in arid regions during the irrigation period is determined, enabling accurate detection of irrigation water consumption in arid regions. This provides decision support for large-scale agricultural water resource management and is crucial for ensuring sustainable agricultural and ecological development and water resource management.

[0061] Furthermore, before acquiring various evapotranspiration data, surface-observed precipitation data, satellite-observed precipitation data, and soil environmental characteristic data for arid areas during the irrigation period, the following steps are also included: Acquire satellite-observed surface temperature data for arid regions; The arid region was simulated using an Earth system model to obtain simulated surface temperature data. Based on the satellite-observed surface temperature data and the simulated surface temperature data, temperature difference data is obtained; Phenological data were obtained based on the leaf area index of the vegetation in the arid region. Based on the temperature difference data and the phenological data, the irrigation period and non-irrigation period of the arid area are determined.

[0062] Specifically, for Earth's surface temperature, data can be obtained either through simulations using Earth system models or through satellite observations. However, simulated surface temperature data cannot reflect the impact of human activities on surface temperature, while satellite observation data can capture these effects.

[0063] Human irrigation has a significant impact on land surface temperature. Since simulated land surface temperature data cannot capture this effect, discrepancies arise between simulated and satellite-observed land surface temperature data. To address this, this invention calculates the temperature difference based on both satellite-observed and simulated land surface temperature data. Specifically, the satellite-observed land surface temperature data is scaled down to eliminate systematic errors between it and the simulated data. : In the formula, This represents satellite-observed Earth surface temperature data. This represents simulated surface temperature data. and These represent the standard deviation and the mean, respectively. Through this scaling method, the two sets of data will have the same mean and standard deviation. At this point, the temperature difference data... It can be represented as: To further improve the accuracy of irrigation period identification, this invention also utilizes LAI dynamics to divide the growing season and non-growing season of vegetation in arid areas.

[0064] First, trend analysis is performed on the time series of LAI for each pixel to identify long-term trends, seasonal cycles, and short-term fluctuations in the data: In the formula, HP represents the Hodrick-Prescott filter.

[0065] Then, the similarity between the LAI curve of each pixel and the typical growth curve is calculated using the Dynamic Time Warping (DTW) algorithm. DTW can identify the growth characteristics of different geographical regions and vegetation types. In the formula, and These are time series of LAI from two different regions. It is a dynamic time-aligned mapping.

[0066] Then, Change Point Detection algorithms (such as those based on CUSUM statistics or Bayesian methods) are used to identify the points of change at the beginning and end of the growing season in the LAI time series: This allows for the accurate determination of the start and end times of the growing season, especially under complex climatic conditions, avoiding errors caused by simply basing the division on fixed dates and further improving accuracy.

[0067] Finally, an adaptive threshold algorithm is applied to automatically adjust the threshold for the growth period based on variations in different regions and years. The growing season and non-growing season are divided to obtain phenological data (whether it is the growing season or not): In the formula, It is an adaptive adjustment coefficient that reflects the characteristics of LAI fluctuations.

[0068] Based on the above temperature difference data Based on phenological data, each time step is divided into three possible periods to determine the irrigation and non-irrigation periods in arid areas: Based on the determination that the phenological period is the growing season, by further determining whether the temperature difference is positive or negative, this time point can be identified as the irrigation period or the non-irrigation period.

[0069] This invention selects eight irrigation records from a typical irrigation station as samples to verify the above algorithm. Figure 3 This is a schematic diagram of the irrigation identification results provided by the present invention, as shown below. Figure 3 As shown, 6 out of 8 irrigation events were successfully classified into the irrigation period, verifying the reliability and accuracy of the algorithm.

[0070] Traditional irrigation period delineation relies heavily on statistical data or fixed growing season assumptions, making it difficult to dynamically adapt to complex planting systems and climate changes. This invention combines satellite-observed surface temperature data with simulated surface temperature data, along with LAI phenological identification, to achieve dynamic, pixel-level identification of irrigation periods. This improves the temporal accuracy and timeliness of irrigation period identification during irrigation water consumption monitoring in arid regions.

[0071] Furthermore, the training steps of the first deep learning model include: Acquire various evapotranspiration data, ground-based precipitation data, satellite-based precipitation data, and soil environmental characteristic data for the arid region during the non-irrigation period; The various evapotranspiration data are weighted and fused to obtain the third data. Based on the ground observation data, the satellite observation precipitation data is corrected pixel by pixel to obtain the fourth data. Based on the third and fourth data, non-evaporation loss data of the arid area during the non-irrigation period are obtained; The first deep learning model is trained using the non-evaporation loss data as label data and the soil environmental characteristic data as feature data.

[0072] Specifically, since the impact of irrigation can be ignored during the non-irrigation period, and the relationship between non-evaporation loss and soil environmental factors is stable between the irrigation and non-irrigation periods, it is easier to explore the accurate relationship between non-evaporation loss and soil environmental factors during the non-irrigation period.

[0073] Multiple evapotranspiration data for arid regions during the non-irrigation period were acquired using various remote sensing evapotranspiration data acquisition devices or models (such as MODIS, GLEAM, PML, etc.); ground-based precipitation data for arid regions during the non-irrigation period were collected using precipitation observations recorded by meteorological stations; and satellite-based precipitation data for arid regions during the non-irrigation period were collected using satellite precipitation products retrieved through remote sensing (such as GPM, TRMM, etc.). Soil environmental characteristics data for arid regions during the non-irrigation period included evapotranspiration data, topsoil moisture content, leaf area index of vegetation, soil texture, soil slope, root depth of vegetation, saturated hydraulic conductivity, and soil porosity.

[0074] By using the Bayesian triplet hat method, the weighted average method (evaluating weights based on the model's historical performance), machine learning-based ensemble learning methods (such as random forest regression and XGBoost weighted fusion), or a Bayesian network model based on confidence propagation to weight and fuse multiple evapotranspiration data, a third data point can be obtained, thus yielding more stable and regionally adaptable evapotranspiration data.

[0075] According to the method in step 103, using ground observation data, pixel-by-pixel correction is performed on satellite-observed precipitation data, and the fourth data can be obtained, thereby improving the overall spatial accuracy and local reliability of satellite-observed precipitation data.

[0076] Based on the third data and the fourth data, the non-evaporation loss data in the non-irrigation period in the arid area is obtained. Since the influence of irrigation can be ignored in the non-irrigation period, the water balance formula in the non-irrigation period can be expressed as: In the formula, represents the non-evaporation loss data in the non-irrigation period, represents the change in soil water content at time represents the infiltration amount at time represents the fourth data, represents the third data. Only the non-evaporation loss amount in the non-irrigation period needs to be obtained based on the difference between precipitation and evapotranspiration.

[0077] And the magnitude of the non-evaporation loss amount depends on a series of environmental factors, such as soil texture, root density, topography, etc. In the embodiments of the present invention, environmental factors are reflected by soil environmental characteristic data.

[0078] In the formula, etc. represent different environmental factors, represents the potential mapping relationship between the non-evaporation loss amount and soil environmental factors, represents the infiltration amount, represents the saturated hydraulic conductivity, represents the effective saturation, and are two empirical parameters, represents the soil water content, represents the soil water content from 0 to h centimeters underground.

[0079] This invention uses non-evaporation loss data as label data and soil environmental characteristic data as feature data to jointly constitute sample data. The sample data obtained during the non-irrigation period is divided into a training set, a validation set, and a test set in an 8:1:1 ratio. In the validation set, the hyperparameters of the first deep learning model are determined through trial and error, enabling the model to learn in advance the potential mapping relationship between non-evaporation loss and soil environmental factors. This allows for accurate and rapid detection of non-evaporation loss based on soil environmental characteristic data from arid areas during the irrigation period. The first deep learning model can be various models such as artificial neural networks, backpropagation networks, and convolutional neural networks.

[0080] Furthermore, the soil environmental characteristic data includes evapotranspiration data, surface soil moisture content, vegetation leaf area index, soil texture, soil slope, vegetation root depth, saturated hydraulic conductivity, and soil porosity.

[0081] Specifically, soil environmental characteristic data is used to reflect environmental factors in arid areas during irrigation or non-irrigation periods, enabling the model to detect non-evaporation losses based on the potential mapping relationship between non-evaporation losses and soil environmental factors.

[0082] In this embodiment of the invention, data such as evapotranspiration, surface soil moisture content, leaf area index of vegetation, soil texture, soil slope, root depth of vegetation, saturated hydraulic conductivity, and soil porosity in arid areas are selected as feature data to fully reflect the environmental factors in arid areas and improve the accuracy of model detection.

[0083] Furthermore, the step of obtaining the surface soil moisture content includes: Acquire high-resolution surface soil moisture characteristic data; the surface soil moisture characteristic data includes precipitation data, evapotranspiration data, surface temperature data, leaf area index of vegetation, soil elevation, and soil texture; The high-resolution surface soil moisture characteristic data is input into the second deep learning model for downscaling to obtain the surface soil moisture content output by the second deep learning model.

[0084] Specifically, the spatial resolution of existing surface soil moisture data retrieved from satellites is usually above 10 km, which is too coarse for application at the field scale, especially in arid regions where irrigated vegetation is relatively fragmented.

[0085] This invention employs a second deep learning model to learn a coarse-resolution mapping relationship between soil moisture and related features, thereby downscaling satellite surface soil moisture data to obtain high-resolution surface soil moisture content. The second deep learning model can be a long short-term memory neural network model, a bidirectional long short-term memory neural network model, a Transformer encoder, a convolutional neural network model, or a hybrid model combining convolutional neural networks and long short-term memory neural networks.

[0086] In addition, in order to fully characterize soil moisture features, the embodiments of the present invention select data such as precipitation data, evapotranspiration data, surface temperature data, leaf area index of vegetation, soil elevation and soil texture, so that the model can fully learn soil moisture features and improve the accuracy of the high-resolution soil moisture content output after downscaling.

[0087] Furthermore, the training steps of the second deep learning model include: Acquire historical surface soil moisture characteristics data at first resolution and the corresponding surface soil moisture content observed by satellite; The resolution of the historical surface soil moisture characteristic data at the first resolution is adjusted to be consistent with the surface soil moisture content to obtain historical surface soil moisture characteristic data at the second resolution. The second deep learning model is trained using historical surface soil moisture characteristic data at the second resolution as feature data and the corresponding satellite-observed surface soil moisture content as label data.

[0088] Specifically, to improve model performance, the model can be pre-trained before being used to downscale the surface soil moisture content observed by satellite. This embodiment of the invention uses a long short-term memory neural network model as an example to illustrate the training process of the second deep learning model.

[0089] Long Short-Term Memory (LSTM) neural network models are an improved type of recurrent neural network specifically designed for processing time-series data and problems with long dependencies. By introducing storage units and gating mechanisms (input gate, forget gate, output gate) to control the flow of information, the model effectively retains useful information while discarding irrelevant information. Figure 4 This is a schematic diagram of the network structure of a long short-term memory neural network model provided by the present invention, as shown below. Figure 4 As shown, the long short-term memory neural network model used in this invention includes an input layer, three hidden layers, a fully connected layer, and an output layer.

[0090] In gating mechanisms, the forget gate is used to determine which information should be discarded from the storage unit, and its output ranges from 0 to 1. In the formula, It is the weight matrix of the forget gate. This indicates the connection to the hidden state of the previous time step. The vector representing the input at the current time step. It is the bias vector of the forget gate. This represents the sigmoid function.

[0091] The input gate is used to determine the location where new information should be merged into the storage unit, and its output is also in the range of 0 to 1, used to indicate the degree of update required for the new information.

[0092] Finally, update the cell state based on the outputs of the forget gate and the input gate: In the formula, and Representing different weights, and These represent different biases. and These represent the neuron states at the current and previous time steps, respectively. It is the hyperbolic tangent function. Represents the input gate vector. Indicates the candidate memory cell value. This represents the forget gate vector.

[0093] The output gate is used to determine the current time. Hidden state : In the formula, and These represent the weights and biases of the output gate, respectively. The hidden state represents the current time step.

[0094] Since environmental variables (such as evapotranspiration and precipitation) affect soil moisture with a lag effect, this embodiment of the invention prefers a long short-term memory neural network model to capture the complex nonlinear relationship between them compared to other models.

[0095] To fully characterize soil moisture features, this embodiment of the invention collected historical surface soil moisture feature data at a first resolution, including precipitation data, evapotranspiration data, surface temperature data, leaf area index of vegetation, soil elevation and soil texture data, as well as the corresponding surface soil moisture content observed by satellite.

[0096] The historical surface soil moisture characteristic data of the first resolution are unified to the same spatiotemporal resolution as the surface soil moisture content data observed by satellite, and only the data on irrigated vegetation are retained in space to ensure that the model has better targeting, thus obtaining the historical surface soil moisture characteristic data of the second resolution.

[0097] Using historical surface soil moisture feature data at the second resolution as feature data and corresponding satellite-observed surface soil moisture content as label data, sample data is constructed together. The sample data is divided into training set, validation set and test set in a ratio of 8:1:1. The hyperparameters of the model are determined by trial and error on the validation set to train the second deep learning model, so that the model can fully learn the coarse-resolution mapping relationship between soil moisture content and related features.

[0098] After obtaining the trained model, high-resolution surface soil moisture characteristic data can be input into the model to obtain the downscaled surface soil moisture content. To address the issue that the low spatial resolution of existing soil moisture products limits the feasibility of field-level irrigation monitoring, this invention utilizes a deep learning model to downscale satellite-observed surface soil moisture content using high-resolution remote sensing features. This downscales coarse-resolution (e.g., 25km) moisture content data to high-resolution (e.g., 500m) data, effectively solving the problem of insufficient spatial resolution. This improves the detail and regional adaptability of irrigation water usage detection in arid regions, thereby enhancing the accuracy of irrigation water usage detection in arid areas.

[0099] The following describes a device for determining irrigation water consumption in arid areas provided by the present invention. The device for determining irrigation water consumption in arid areas described below can be referred to in correspondence with the method for determining irrigation water consumption in arid areas described above.

[0100] Based on any of the above embodiments Figure 5 This is a schematic diagram of the structure of a device for determining irrigation water consumption in arid areas provided by the present invention, as shown below. Figure 5 As shown. This embodiment of the invention provides a device for determining irrigation water consumption in arid areas, including an acquisition module 501, a fusion module 502, a correction module 503, a detection module 504, a correction module 505, and a determination module 506, wherein: The acquisition module 501 is used to acquire various evapotranspiration data, ground-observed precipitation data, satellite-observed precipitation data, and soil environmental characteristic data of the arid region during the irrigation period; the fusion module 502 is used to perform weighted fusion of the various evapotranspiration data to obtain first data; the correction module 503 is used to perform pixel-by-pixel correction on the satellite-observed precipitation data based on the ground-observed precipitation data to obtain second data; the detection module 504 is used to input the soil environmental characteristic data into a first deep learning model to perform non-evaporation loss detection and obtain the first non-evaporation loss data of the arid region during the irrigation period output by the first deep learning model; the correction module 505 is used to correct the first non-evaporation loss data according to the vegetation type of the arid region to obtain second non-evaporation loss data; and the determination module 506 is used to determine the irrigation water consumption of the arid region during the irrigation period based on the first data, the second data, and the second non-evaporation loss data.

[0101] This invention provides a device for determining irrigation water consumption in arid areas. First, it acquires various evapotranspiration data, ground-based precipitation data, satellite-based precipitation data, and soil environmental characteristic data for the arid area during the irrigation period. Then, it performs weighted fusion of the various evapotranspiration data to obtain first data, thereby improving the accuracy and reliability of the evapotranspiration data. Next, based on the ground-based observation data, it performs pixel-by-pixel correction on the satellite precipitation data to obtain second data, thereby improving the overall spatial accuracy and local reliability of the satellite precipitation data. Finally, it inputs the soil environmental characteristic data into a first deep learning model for non-evaporation loss detection to obtain the first deep learning model's output for the arid area during the irrigation period. First, non-evaporation loss data is used to accurately and quickly detect non-evaporation loss data in arid areas during the irrigation period through a model, achieving unbiased, data-driven inversion of key data and breaking through the data acquisition limitations of traditional remote sensing observations. Then, the first non-evaporation loss data is corrected according to the vegetation type of the arid area to obtain second non-evaporation loss data. This allows for differentiated correction of non-evaporation loss data for different vegetation types through vegetation classification information, improving the accuracy of non-evaporation loss data. Finally, based on the first, second, and second non-evaporation loss data, the irrigation water consumption in arid areas during the irrigation period is determined, achieving accurate detection of irrigation water consumption in arid areas and providing decision support for large-scale agricultural water resource management.

[0102] Figure 6 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 6As shown, the electronic device may include: a processor 610, a communications interface 620, a memory 630, and a communication bus 640, wherein the processor 610, the communications interface 620, and the memory 630 communicate with each other via the communication bus 640. The processor 610 can call logical instructions in the memory 630 to execute a method for determining irrigation water usage in arid areas, the method including: Acquire various evapotranspiration data, ground-based precipitation data, satellite-based precipitation data, and soil characteristic environmental data for arid regions during the irrigation period; The various evapotranspiration data are weighted and fused to obtain the first data; Based on the ground observation data, the satellite observation precipitation data is corrected pixel by pixel to obtain the second data; The soil characteristic environmental data is input into the first deep learning model to perform non-evaporation loss detection, and the first non-evaporation loss data of the arid area during the irrigation period is obtained from the output of the first deep learning model. The first non-evaporation loss data is corrected according to the vegetation type of the arid area to obtain the second non-evaporation loss data; Based on the first data, the second data, and the second non-evaporation loss data, the irrigation water consumption in the arid area during the irrigation period is determined.

[0103] Furthermore, the logical instructions in the aforementioned memory 630 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0104] On the other hand, the present invention also provides a computer program product, the computer program product comprising a computer program that can be stored on a non-transitory computer-readable storage medium, wherein when the computer program is executed by a processor, the computer is able to execute the method for determining irrigation water consumption in arid areas provided by the above methods, the method comprising: Acquire various evapotranspiration data, ground-based precipitation data, satellite-based precipitation data, and soil characteristic environmental data for arid regions during the irrigation period; The various evapotranspiration data are weighted and fused to obtain the first data; Based on the ground observation data, the satellite observation precipitation data is corrected pixel by pixel to obtain the second data; The soil characteristic environmental data is input into the first deep learning model to perform non-evaporation loss detection, and the first non-evaporation loss data of the arid area during the irrigation period is obtained from the output of the first deep learning model. The first non-evaporation loss data is corrected according to the vegetation type of the arid area to obtain the second non-evaporation loss data; Based on the first data, the second data, and the second non-evaporation loss data, the irrigation water consumption in the arid area during the irrigation period is determined.

[0105] In another aspect, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to perform the method for determining irrigation water consumption in arid areas provided by the methods described above, the method comprising: Acquire various evapotranspiration data, ground-based precipitation data, satellite-based precipitation data, and soil characteristic environmental data for arid regions during the irrigation period; The various evapotranspiration data are weighted and fused to obtain the first data; Based on the ground observation data, the satellite observation precipitation data is corrected pixel by pixel to obtain the second data; The soil characteristic environmental data is input into the first deep learning model to perform non-evaporation loss detection, and the first non-evaporation loss data of the arid area during the irrigation period is obtained from the output of the first deep learning model. The first non-evaporation loss data is corrected according to the vegetation type of the arid area to obtain the second non-evaporation loss data; Based on the first data, the second data, and the second non-evaporation loss data, the irrigation water consumption in the arid area during the irrigation period is determined.

[0106] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0107] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0108] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element. Furthermore, it should be noted that the scope of the methods and apparatuses in the embodiments of this application is not limited to performing functions in the order shown or discussed, but may also include performing functions substantially simultaneously or in the reverse order, depending on the functions involved. For example, the described methods may be performed in a different order than described, and various steps may be added, omitted, or combined. Additionally, features described with reference to certain examples may be combined in other examples.

[0109] It should also be noted that the terms "first," "second," etc., used in this invention are used to distinguish similar objects, and not to describe a specific order or sequence. It should be understood that such terms can be used interchangeably where appropriate so that embodiments of this application can be implemented in orders other than those illustrated or described herein, and the objects distinguished by "first" and "second" are generally of the same class, and the number of objects is not limited; for example, the first object can be one or more.

[0110] In this application's embodiments, "determine B based on A" means that factor A must be considered when determining B. It is not limited to "B can be determined based solely on A," but should also include: "determine B based on A and C," "determine B based on A, C, and E," "determine C based on A, and further determine B based on C," etc. Additionally, it can include using A as a condition for determining B, for example, "when A meets the first condition, determine B using the first method"; another example, "when A meets the second condition, determine B," etc.; another example, "when A meets the third condition, determine B based on the first parameter," etc. Of course, it can also be a condition where A is a factor in determining B, for example, "when A meets the first condition, determine C using the first method, and further determine B based on C," etc.

[0111] In this invention, the term "multiple" refers to two or more, and other quantifiers are similar.

[0112] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for determining irrigation water consumption in arid areas, characterized in that, include: Acquire various evapotranspiration data, ground-based precipitation data, satellite-based precipitation data, and soil characteristic environmental data for arid regions during the irrigation period; The various evapotranspiration data are weighted and fused to obtain the first data; Based on the ground observation data, the satellite observation precipitation data is corrected pixel by pixel to obtain the second data; The soil characteristic environmental data is input into the first deep learning model to perform non-evaporation loss detection, and the first non-evaporation loss data of the arid area during the irrigation period is obtained from the output of the first deep learning model. The first non-evaporation loss data is corrected according to the vegetation type of the arid area to obtain the second non-evaporation loss data; Based on the first data, the second data, and the second non-evaporation loss data, the irrigation water consumption in the arid area during the irrigation period is determined.

2. The method for determining irrigation water consumption in arid areas according to claim 1, characterized in that, Before acquiring various evapotranspiration data, surface-observed precipitation data, satellite-observed precipitation data, and soil characteristic environmental data for arid areas during the irrigation period, the following steps are also included: Acquire satellite-observed surface temperature data for arid regions; The arid region was simulated using an Earth system model to obtain simulated surface temperature data. Based on the satellite-observed surface temperature data and the simulated surface temperature data, temperature difference data is obtained; Phenological data were obtained based on the leaf area index of the vegetation in the arid region. Based on the temperature difference data and the phenological data, the irrigation period and non-irrigation period of the arid area are determined.

3. The method for determining irrigation water consumption in arid areas according to claim 1, characterized in that, The training steps of the first deep learning model include: Acquire various evapotranspiration data, ground-based precipitation data, satellite-based precipitation data, and soil environmental characteristic data for the arid region during the non-irrigation period; The various evapotranspiration data are weighted and fused to obtain the third data. Based on the ground observation data, the satellite observation precipitation data is corrected pixel by pixel to obtain the fourth data. Based on the third and fourth data, non-evaporation loss data of the arid area during the non-irrigation period are obtained; The first deep learning model is trained using the non-evaporation loss data as label data and the soil environmental characteristic data as feature data.

4. The method for determining irrigation water consumption in arid areas according to claim 1 or 3, characterized in that, The soil environmental characteristics data include evapotranspiration data, surface soil moisture content, vegetation leaf area index, soil texture, soil slope, vegetation root depth, saturated hydraulic conductivity, and soil porosity.

5. The method for determining irrigation water consumption in arid areas according to claim 4, characterized in that, The steps for obtaining the surface soil moisture content include: Acquire high-resolution surface soil moisture characteristic data; the surface soil moisture characteristic data includes precipitation data, evapotranspiration data, surface temperature data, leaf area index of vegetation, soil elevation, and soil texture; The high-resolution surface soil moisture characteristic data is input into the second deep learning model for downscaling to obtain the surface soil moisture content output by the second deep learning model.

6. The method for determining irrigation water consumption in arid areas according to claim 5, characterized in that, The training steps for the second deep learning model include: Acquire historical surface soil moisture characteristics data at first resolution and the corresponding surface soil moisture content observed by satellite; The resolution of the historical surface soil moisture characteristic data at the first resolution is adjusted to be consistent with the surface soil moisture content to obtain historical surface soil moisture characteristic data at the second resolution. The second deep learning model is trained using historical surface soil moisture characteristic data at the second resolution as feature data and the corresponding satellite-observed surface soil moisture content as label data.

7. A device for determining irrigation water consumption in arid areas, characterized in that, include: The acquisition module is used to acquire various evapotranspiration data, ground-based precipitation data, satellite-based precipitation data, and soil environmental characteristic data in arid areas during the irrigation period; The fusion module is used to perform weighted fusion of the various evapotranspiration data to obtain the first data; The correction module is used to perform pixel-by-pixel correction on the satellite precipitation data based on the ground observation data to obtain the second data; The detection module is used to input the soil environmental characteristic data into the first deep learning model to perform non-evaporation loss detection and obtain the first non-evaporation loss data of the arid area during the irrigation period output by the first deep learning model. The correction module is used to correct the first non-evaporation loss data according to the vegetation type of the arid area to obtain the second non-evaporation loss data; The determination module is used to determine the irrigation water consumption of the arid area during the irrigation period based on the first data, the second data, and the second non-evaporation loss data.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the method for determining irrigation water consumption in arid areas as described in any one of claims 1 to 6.

9. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the method for determining irrigation water consumption in arid areas as described in any one of claims 1 to 6.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the method for determining irrigation water consumption in arid areas as described in any one of claims 1 to 6.

Images