Method and system for determining project scale layout of water network irrigation area of agricultural and pastoral area

Abstract

The invention discloses an agricultural and pastoral area water network irrigation area project scale layout determination method and system. The method comprises the following steps: collecting agricultural and pastoral area data, gridding an agricultural and pastoral area, constructing a topographic feature matrix, calculating the rainfall of each grid, calculating an evapotranspiration value, performing water-soil balance calculation, extracting a plurality of drought centers and dividing the agricultural and pastoral area into a plurality of sub-areas; calculating the water inflow of each sub-region, drawing a water layer distribution diagram, determining a water intake, and constructing a water source engineering layout scheme of each sub-region; extracting a water source project water outlet of each sub-region, determining an irrigation system and constructing a drainage project in combination with irrigation area topographic data, and obtaining an irrigation area project layout scheme of each sub-region; an evaluation index system is constructed, an economic evaluation model is constructed, comprehensive benefit scores of the water network irrigation area project layout schemes are calculated, and the scheme with the highest score is the agricultural and pastoral area water network irrigation area project scale layout scheme. The water resource utilization efficiency is improved, and reasonable configuration and sustainable utilization of water resources in agricultural and pastoral areas are achieved.

Description

Technical Field

[0001] The present invention relates to a method for determining the scale layout of water network irrigation projects in rural and pastoral areas. Background Art

[0002] The scale layout of water network irrigation projects is crucial for the sustainable development of agriculture and animal husbandry. A reasonable layout of water network irrigation projects can effectively allocate water resources, improve irrigation efficiency, promote agricultural and animal husbandry production increase and income, and at the same time protect the ecological environment. However, there are many deficiencies in the existing methods for determining the scale layout of water network irrigation projects in rural and pastoral areas.

[0003] On the one hand, when calculating the regional rainfall and evapotranspiration values, traditional methods often use simple statistical or empirical formulas, without fully considering the influence of complex factors such as terrain, resulting in low accuracy of the calculation results and being difficult to accurately reflect the actual water resource situation. On the other hand, when determining the layout of water source projects, irrigation systems and drainage projects, there is a lack of systematic and comprehensive planning, without fully combining the actual needs and characteristics of rural and pastoral areas, making the project layout unreasonable, the water resource utilization efficiency low, and unable to meet the growing water demand in rural and pastoral areas. In addition, when evaluating the project layout scheme with the existing methods, the evaluation indicators are single, without comprehensively considering the benefits in multiple aspects such as economy, society and ecology, and it is difficult to select the optimal project scale layout scheme. Therefore, there is an urgent need for a scientific, accurate and comprehensive method and system for determining the scale layout of water network irrigation projects in rural and pastoral areas.

[0004] The present invention proposes a method and system for determining the scale layout of water network irrigation projects in rural and pastoral areas, solves the above problems existing at present, improves the water resource utilization efficiency, and realizes the reasonable allocation and sustainable utilization of water resources in rural and pastoral areas. Summary of the Invention

[0005] Object of the Invention: To provide a method for determining the scale layout of water network irrigation projects in rural and pastoral areas to solve the above problems existing in the prior art. On the other hand, to provide a system for determining the scale layout of water network irrigation projects in rural and pastoral areas.

[0006] Technical Solution: The method for determining the scale layout of water network irrigation projects in rural and pastoral areas includes the following steps: Step S1, collect rural and pastoral area data, grid the rural and pastoral areas, construct a terrain feature matrix, calculate the rainfall of each grid, calculate the evapotranspiration value using a pre-constructed long short-term memory model, and perform soil and water balance calculation, extract several drought centers and divide the rural and pastoral areas into several sub-regions; Step S2, calculate the water inflow of each sub-region, draw a water layer distribution map, calculate the water quality guarantee rate based on the water source quantity data and water demand, determine the water intake, and construct the water source project layout scheme of each sub-region based on the spatial distribution of drought centers; Step S3: Extract the water source project outlets of each sub-region, determine the irrigation system in combination with the irrigation district terrain data, construct a drainage project based on the rainstorm intensity data and terrain catchment data of each sub-region, and obtain the irrigation district project layout plan for each sub-region; Step S4: Construct an evaluation index system, calculate the index values of the water network irrigation district project layout plans of each sub-region respectively, input them into the pre-constructed economic evaluation model, calculate the corresponding comprehensive benefit scores, and the one with the highest score is the project scale layout plan for the rural and pastoral water network irrigation district.

[0007] According to one aspect of the present application, the step S1 is further as follows: Step S11: Collect rural and pastoral data, including: rural and pastoral digital elevation data, meteorological data, rainfall station observation data, crop planting structure data, livestock breeding scale data; Step S12: Grid the study area based on the rural and pastoral terrain data, extract the rural and pastoral digital elevation data and rainfall station observation data, and construct a rural and pastoral terrain feature matrix; Step S13: Calculate the terrain similarity between each grid and each rainfall station using the Euclidean distance improved by cosine similarity, optimize the inverse distance weighting formula based on the terrain similarity to obtain a mixed weight function, and perform weighted averaging on each grid using the mixed weight function to obtain the corresponding rainfall; Step S14: Collect rural and pastoral historical data to calculate the historical potential evapotranspiration value of the rural and pastoral areas, and train the pre-constructed long short-term memory model together with the historical actual evapotranspiration value of the rural and pastoral areas. Extract the rural and pastoral data as the input of the trained long short-term memory model, and calculate the historical actual evapotranspiration value of the rural and pastoral areas; Step S15: Calculate the historical available water volume of the rural and pastoral areas based on the historical rainfall and historical potential evapotranspiration values of the rural and pastoral areas, perform a water and soil balance calculation on the rural and pastoral areas based on the rural and pastoral data to obtain the historical water demand of the rural and pastoral areas, count the historical water demand of each grid in the rural and pastoral areas, extract several drought centers and divide the rural and pastoral areas into several sub-regions.

[0008] According to one aspect of the present application, the step S12 is further as follows: Step S12a: Extract the rural and pastoral terrain data, set the grid resolution to grid the rural and pastoral areas to obtain the rural and pastoral terrain grid; Step S12b: Based on the rural and pastoral terrain grid resolution, use the inverse distance weighting method to convert the rainfall station data into a spatial distribution grid; Step S12c: Integrate the rural and pastoral terrain grid and the rainfall spatial distribution grid to obtain a rural and pastoral terrain feature matrix.

[0009] According to one aspect of the present application, the step S15 is further as follows: Step S15a: Extract the historical rainfall and historical potential evapotranspiration values in the rural and pastoral areas, and calculate the historical available water volume in the pastoral areas; Step S15b: Extract the crop types, irrigation methods, soil characteristics, crop water requirement coefficients and crop growth cycles in the rural and pastoral areas, calculate the crop water requirements at each time point for each grid in the rural and pastoral areas, and summarize the domestic and ecological water requirements to obtain the historical water requirements in the rural and pastoral areas; Step S15c: Extract the water requirement data at each time point for each grid in the rural and pastoral areas. Taking the grid as a unit, map the colors of the water requirements to draw a drought distribution map. The colors corresponding to the water requirements from small to large are from light to dark. Overlay the distribution maps at each time point, calculate the total sum and average value of the water requirements for each grid, extract the grids where both the total sum and average value of the water requirements are greater than the threshold, and divide the connected grids into the same area to obtain several drought centers; Step S15d: Based on the spatial distribution of the drought centers, use Thiessen polygons to divide the rural and pastoral areas into several sub - regions, and each sub - region contains one drought center.

[0010] According to one aspect of the present application, the step S2 is further as follows: Step S21: Collect the detection data of the hydrological station and use the cross - section flow calculation method to collect the incoming water volume data at different frequencies; Step S22: Collect hydrological data, dynamic changes in water levels, detect water quality parameters, water quality indicators and water volume parameters, and draw a water layer distribution map; Step S23: Calculate the water quality preservation rate based on the water volume data of the water source and the water requirements. Based on the elevation difference between the water source level and the water intake point in the irrigation area, determine the feasibility of water intake, and screen to obtain the water intake points; Step S24: Based on the spatial distribution of the drought centers, construct surface water source engineering plans and groundwater source engineering plans respectively, and combine them to obtain the water source engineering layout plans for each sub - region.

[0011] According to one aspect of the present application, the step S24 is further as follows: Step S24a: Collect the groundwater level monitoring data and river - lake water level data of each sub - region, combine with the water requirements, draw a water level contour map, extract the coordinates of the water level contours, and calculate the slope of each contour line; Step S24b: Set an initial buffer distance, generate a circular initial buffer zone with the drought center as the center, correct the initial buffer zone based on the density of the water level contours, set a slope threshold, and divide the buffer zone into a core buffer zone and a secondary buffer zone based on the calculated slope of the contour lines; Step S24c: Construct surface water source engineering plans and groundwater source engineering plans for each sub - region for the core buffer zone and the secondary buffer zone respectively, input them into the pre - constructed water resources joint dispatching model, and solve the model to obtain the water source engineering layout plans for each sub - region.

[0012] According to one aspect of the present application, step S3 is further as follows: Step S31: Collect the terrain slope, soil permeability coefficient, crop planting area distribution, and the position of the water outlet of the water source project in the irrigation area; Step S32: Select an irrigation system based on the terrain slope, and calculate the pipe diameter, dripper spacing, channel cross-section size, and pump station lift; Step S33: Construct a drainage project based on the rainfall intensity data and terrain catchment data of each sub-region to obtain the irrigation area project layout plan for each sub-region.

[0013] According to one aspect of the present application, step S4 is further as follows: Step S41: Construct an evaluation index system, including economic value index, social benefit index, and ecological benefit index; Step S42: Calculate the index values of the water network irrigation area project layout plans for each sub-region respectively; Step S43: Construct an economic evaluation model, extract the index values of the water network irrigation area project layout plans for each sub-region as the model input, calculate the comprehensive benefit scores of the water network irrigation area project layout plans for each sub-region, and the one with the highest score is the project scale layout plan for the rural and pastoral water network irrigation area.

[0014] According to one aspect of the present application, step S42 is further as follows: Step S42a: Calculate the economic benefit index values using the dynamic analysis method, including: net present value, internal rate of return, and investment payback period; Step S42b: Calculate the social benefit indicators using the analytic hierarchy process, including: quality of life, employment opportunities, and social stability; Step S42c: Calculate the ecological benefit indicators using the ecological environment quality index method.

[0015] According to one aspect of the present application, step S43 is further as follows: Step S43a: Construct a multi-level comprehensive evaluation model, including an objective layer, a criterion layer, and an index layer. The objective layer is the comprehensive benefit of the project scale layout plan for the rural and pastoral water network irrigation area, the criterion layer includes economic benefits, social benefits, and ecological benefits, and the index layer is further subdivided according to the criterion layer; Step S43b: Score the importance of the indicators pairwise based on the 1-9 scale method, construct a judgment matrix, calculate the eigenvector and the maximum eigenvalue of the judgment matrix, conduct a consistency test, determine the weights of each indicator, and configure the multi-level comprehensive evaluation model; Step S43c: Extract the index values of the engineering layout plan for the water network irrigation area in each sub-region as the input of the multi-level comprehensive evaluation model, calculate the comprehensive benefit score of the engineering layout plan for the water network irrigation area in each sub-region, and the one with the highest score is the engineering scale layout plan for the water network irrigation area in the rural and pastoral areas.

[0016] According to another aspect of the present application, there is provided a system for determining the engineering scale layout of a water network irrigation area in rural and pastoral areas, including: At least one processor; and A memory communicatively connected to at least one of the processors; wherein, The memory stores instructions executable by the processor, and the instructions are used to be executed by the processor to implement the method for determining the engineering scale layout of the water network irrigation area in rural and pastoral areas according to any one of the above technical solutions.

[0017] Beneficial effects: By adopting the method for determining the engineering scale layout of the water network irrigation area in rural and pastoral areas, relevant data in rural and pastoral areas are collected and analyzed through scientific methods, a reasonable engineering layout plan is constructed, and the optimal engineering scale layout plan for the water network irrigation area in rural and pastoral areas is determined through a comprehensive evaluation index system and an economic evaluation model. Brief Description of the Drawings

[0018] Figure 1 is a flowchart of the present invention.

[0019] Figure 2 is a flowchart of step S1 of the present invention.

[0020] Figure 3 is a flowchart of step S2 of the present invention.

[0021] Figure 4 is a flowchart of step S3 of the present invention.

[0022] Figure 5 is a flowchart of step S4 of the present invention. Detailed Embodiments

[0023] As Figure 1 shown, the following technical solutions are proposed. According to one aspect of the present application, there is provided a method for determining the engineering scale layout of a water network irrigation area in rural and pastoral areas, characterized by including the following steps: Step S1: Collect rural and pastoral area data, grid the rural and pastoral areas, construct a terrain feature matrix, calculate the rainfall of each grid, calculate the evapotranspiration value using a pre-constructed long short-term memory model, perform water and soil balance calculations, extract several drought centers, and divide the rural and pastoral areas into several sub-regions; Step S2: Calculate the water inflow of each sub-region, draw the water layer distribution map, calculate the water quality preservation rate based on the water source volume data and water demand, determine the water intake points, and construct the water source project layout plan for each sub-region based on the spatial distribution of the drought centers; Step S3: Extract the water outlet of the water source project in each sub-region, combine with the irrigation area terrain data to determine the irrigation system, and construct the drainage project based on the rainstorm intensity data and terrain water collection data in each sub-region to obtain the irrigation area project layout plan for each sub-region; Step S4: Construct an evaluation index system, calculate the index values of the water network irrigation area project layout plans in each sub-region respectively, input them into the pre-constructed economic evaluation model, calculate the corresponding comprehensive benefit scores, and the one with the highest score is the water network irrigation area project scale layout plan for the rural and pastoral areas.

[0024] First, collect the digital elevation, meteorological, rainfall station observation, crop planting structure, and livestock breeding scale data of the rural and pastoral areas, grid the rural and pastoral areas, construct a terrain feature matrix, calculate the terrain similarity between each grid and the rainfall station using the Euclidean distance improved by cosine similarity, optimize the inverse distance weight formula to obtain a mixed weight function, so as to accurately calculate the rainfall of each grid; at the same time, construct a long short-term memory model, calculate the historical potential evapotranspiration value in combination with historical data, train the model together with the historical true evapotranspiration value, input the current data to obtain the true evapotranspiration value, calculate the available water volume based on the rainfall and evapotranspiration values, conduct a soil and water balance calculation in combination with other rural and pastoral area data, obtain the water demand and water shortage, identify the drought centers, and divide the rural and pastoral areas based on the drought centers to obtain several sub-regions; Next, collect the hydrological station detection data, use the cross-section flow calculation method to obtain the water inflow data at different frequencies, at the same time collect the hydrological, water level dynamics, and water quality and water volume parameters, draw the water layer distribution map, calculate the water quality preservation rate based on the water source volume and water demand, determine the water intake points in combination with the elevation difference between the water source level and the irrigation area water intake point, construct the surface water and groundwater water source project plans respectively, and solve through the water resources joint dispatching model to obtain the water source project layout plan for each sub-region, realizing the rational allocation and planning of water resources; Then, collect the irrigation area terrain slope and soil permeability coefficient data, select a suitable irrigation system according to the terrain slope and calculate the relevant parameters, and then construct the drainage project in combination with the rainstorm intensity and terrain water collection data to form the irrigation area project layout plan for each sub-region, ensuring the scientific rationality of irrigation and drainage; Finally, construct an evaluation index system covering economic, social, and ecological benefits, calculate the index values using dynamic analysis and hierarchical analysis methods, construct a multi-level comprehensive evaluation model to determine the index weights, input the index values of each plan to calculate the comprehensive benefit scores, and select the plan with the highest score as the water network irrigation area project scale layout plan for the rural and pastoral areas.

[0025] Such as Figure 2As shown, according to one aspect of the present application, the step S1 is further as follows: Step S11, collect data of rural and pastoral areas, including: digital elevation data of rural and pastoral areas, meteorological data, rainfall station observation data, crop planting structure data, livestock breeding scale data; Step S12, grid the research area based on the terrain data of rural and pastoral areas, extract the digital elevation data of rural and pastoral areas and the rainfall station observation data, and construct a terrain feature matrix of rural and pastoral areas; Step S13, calculate the terrain similarity between each grid and each rainfall station using the Euclidean distance improved by cosine similarity, optimize the inverse distance weighting formula based on the terrain similarity to obtain a hybrid weighting function, and perform weighted averaging on each grid using the hybrid weighting function to obtain the corresponding rainfall; In a certain embodiment, the step S13 is specifically as follows: Step S13a, extract multi-dimensional terrain features of each grid and rainfall station based on the digital elevation data, construct terrain feature vectors, calculate the similarity of directions between vectors using cosine similarity, and calculate the spatial distance between vectors using the Euclidean distance, and fuse the similarity and spatial distance to obtain the terrain similarity between vectors, that is, the terrain similarity between each grid and each rainfall station; Preprocess the digital elevation data of the research area, extract slope, aspect, terrain wetness index, terrain position index, and construct a terrain feature matrix: F = [f1, f2,..., fn] T ; where, fi represents the i-th terrain feature layer; Calculate the terrain feature vectors of the interpolation point p and each rainfall station s: Vp = [s p , a p , tpi p , twi p ; Vs = [s s , a s , tpi s , twi s ; where, s represents slope, a represents aspect, tpi represents terrain position index, and twi represents terrain wetness index; Construct a comprehensive similarity index using cosine similarity combined with Euclidean distance: S = α·cosθ + (1 - α)·exp(-β·dmaxd); where, θ is the angle between feature vectors, d is the Euclidean distance, and α and β are weight coefficients; Step S13b, fuse the terrain similarity with the inverse distance weighting formula to obtain an optimized inverse distance weighting formula, and use an adaptive particle swarm algorithm to dynamically determine the optimization parameters to obtain a hybrid weighting function; Adjusting the traditional inverse distance weight based on terrain similarity: Wi = d i -pm ·exp(-γ·(1 - Si)); or Wi = (1 / d i pm ) × exp(-γ × (1 - Si)); Where, di is the distance, pm is the distance power parameter, Si is the terrain similarity, and γ is the terrain influence coefficient; Dynamically determine the optimal parameters p and γ through the adaptive particle swarm optimization algorithm; For each grid cell: Pgrid = ∑ i=1 n Wi·Pi; where, Pi is the observed value of the ith rain gauge station, and n is the number of rain gauge stations affecting this grid; Step S13c, use a hybrid weight function to perform weighted averaging on each grid to obtain the corresponding rainfall.

[0026] In this embodiment, integrating the multi-dimensional terrain feature similarity into the areal rainfall calculation breaks through the limitation of traditional methods that only rely on distance factors; using an adaptive optimization algorithm to dynamically determine the model parameters improves the calculation accuracy and adaptability; in the sparse station area and the areal rainfall calculation accuracy is greatly improved compared with traditional methods, especially suitable for complex terrain areas.

[0027] Step S14, collect historical data in the rural and pastoral areas to calculate the historical potential evapotranspiration value in the rural and pastoral areas, and train the pre-constructed long short-term memory model together with the historical actual evapotranspiration value in the rural and pastoral areas, extract the rural and pastoral area data as the input of the trained long short-term memory model, and calculate the actual evapotranspiration value in the rural and pastoral areas; In a certain embodiment, the specific steps of step S14 are: Step S14a, construct a long short-term memory model; Step S14b, collect historical data and historical actual evapotranspiration values in the rural and pastoral areas, and calculate the historical potential evapotranspiration value in the rural and pastoral areas based on the historical data in the rural and pastoral areas; Step S14c, use the historical potential evapotranspiration value and historical actual evapotranspiration value in the rural and pastoral areas as the training set to train the long short-term memory model and optimize the model parameters; Step S14d, extract the rural and pastoral area data and input it into the trained long short-term memory model to calculate the actual evapotranspiration value in the rural and pastoral areas.

[0028] In this embodiment, by constructing a multi-input feature space, combining the LSTM time series model with the physical constraint loss function, the learning and characterization of the evaporation paradox phenomenon are realized, and the fitting accuracy of the relationship between potential evapotranspiration and actual evapotranspiration is improved; First, integrate meteorological data, soil moisture data, vegetation indices, and measured evapotranspiration data, calculate potential evapotranspiration, and extract characteristic indicators related to radiation trends, wind speed trends, vapor pressure deficit, and the evaporation paradox. At the same time, encode daily sequence and seasonal time characteristics. Subsequently, build a long short-term memory network model. Aiming at the characteristic of the contradiction between potential evapotranspiration and actual evapotranspiration trends in the evaporation paradox, design a loss function with double constraints: on the one hand, use the mean square error to ensure the fitting degree between the predicted value and the measured value; on the other hand, construct a paradox penalty term. When radiation increases but actual evapotranspiration decreases, or radiation decreases but actual evapotranspiration increases, the prediction error of the model is weighted and amplified. Through hyperparameter adjustment, the balance optimization of the two constraints is achieved. In the model training session, divide the data set into a training set and a test set according to an 8:2 ratio, adopt an early stopping mechanism to prevent overfitting, and construct time series samples through a rolling time window for model training. Finally, based on the trained model, input the preprocessed multi-source data and output hourly or daily actual evapotranspiration data. This method transforms the evaporation paradox phenomenon into computable constraint conditions through in-depth data feature mining and innovative model structure design. Compared with traditional empirical models, the prediction accuracy of actual evapotranspiration in areas with significant evaporation paradox is improved, effectively solving the problem of insufficient interpretability of traditional methods for complex hydrological phenomena.

[0029] In this embodiment, a special constraint loss function is designed for the evaporation paradox phenomenon to solve the technical problem that traditional models cannot effectively handle the contradiction between potential evapotranspiration and actual evapotranspiration trends. Specifically, a composite loss function including a paradox penalty term is constructed: L total =L mse +λ×L paradox ; where, L mse is the mean square error loss function, λ is the paradox penalty weight coefficient, and L paradox is the paradox penalty term.

[0030] The specific calculation method of the paradox penalty term Lparadox is: Lparadox = Σ[wi × max(0, sign(ΔRi) × sign(ΔETactual_i)+1)]; where, w i is the weight at the i-th time step, ΔR i is the radiation change amount, ΔET actual_i is the actual evapotranspiration change amount, sign() is the sign function. When radiation increases but actual evapotranspiration decreases, sign(ΔR i ) × sign(ΔET actual_i ) = -1, making the penalty term 2×w i ; when the change trends of the two are the same, the penalty term is 0.

[0031] It should be noted that the value range of the weight coefficient λ is 0.1 - 0.5, preferably 0.3. The weight w_i is determined according to the intensity of the evaporation paradox phenomenon, and the calculation formula is: w i =|ΔR i | / (|ΔR i |+|ΔET potential_i| ); Through the above paradox penalty mechanism, the model focuses on correcting the prediction results that violate physical laws during the training process, improving the prediction accuracy in the significant area of the evaporation paradox.

[0032] In some embodiments, a segmented penalty strategy is optionally adopted. When |ΔR i | is greater than the preset threshold, the penalty weight is doubled to further strengthen the constraint effect on the strong evaporation paradox phenomenon.

[0033] For example, in a test in a certain arid area, when the solar radiation increases by 15% and the actual evapotranspiration decreases by 8%, the value of the paradox penalty term is 0.46, significantly improving the model's ability to identify and correct such abnormal situations.

[0034] Step S15: Calculate the historical available water volume in the rural and pastoral areas based on the historical rainfall and historical potential evapotranspiration values in the rural and pastoral areas, conduct a water and soil balance calculation for the rural and pastoral areas based on the rural and pastoral area data to obtain the historical water demand in the rural and pastoral areas, count the historical water demand of each grid in the rural and pastoral areas, extract several drought centers, and divide the rural and pastoral areas into several sub-regions.

[0035] The water and soil balance calculation method integrating the evaporation paradox provided by the present invention creatively combines machine learning technologies such as geographically weighted regression and long short-term memory network with traditional hydrological calculation models, improves the deficiencies of traditional methods in rainfall calculation, evapotranspiration processing, and water and soil balance assessment, optimizes the evapotranspiration calculation model by introducing the evaporation paradox as a constraint condition, and combines national land standards and crop characteristics in rural and pastoral areas to achieve accurate calculation of rainfall - evapotranspiration - water and soil balance driven by multi-source data, with significant improvements in calculation accuracy and practicality compared to the prior art.

[0036] According to one aspect of the present application, the step S12 is further as follows: Step S12a: Extract the topographic data of the rural and pastoral areas, set the grid resolution to grid the rural and pastoral areas, and obtain the topographic grid of the rural and pastoral areas; Step S12b: Based on the topographic grid resolution of the rural and pastoral areas, use the inverse distance weighting method to convert the rainfall station data into a spatial distribution grid; Step S12c: Integrate the topographic grid of the rural and pastoral areas and the rainfall spatial distribution grid to obtain the topographic feature matrix of the rural and pastoral areas.

[0037] In a certain embodiment, specifically: Extract the topographic data of rural and pastoral area A from the geographic information system database, including contour line and elevation point information. Set the grid resolution to 100 m × 100 m, gridify the rural and pastoral area, and obtain 50,000 topographic grids, each of which contains corresponding topographic information; There are 10 rainfall stations distributed in this rural and pastoral area, which record the rainfall data of the past year. Based on the set grid resolution of 100 m, the inverse distance weighting method is used to interpolate the observed data of these 10 rainfall stations into a spatial distribution grid. The inverse distance weighting method will assign different weights according to the distance between the rainfall station and the interpolation point to estimate the rainfall of each grid. The closer the distance, the greater the weight; Integrate the topographic grid of the rural and pastoral area obtained in step S12a and the rainfall spatial distribution grid obtained in step S12b to obtain the topographic feature matrix of the rural and pastoral area.

[0038] Specifically for the selected inverse distance weighting method, the Kriging interpolation method can also be selected. First, calculate the variogram of the rainfall data, analyze its spatial variation characteristics, and determine appropriate semi-variogram models and parameters, such as nugget effect, range, and sill value. Then, based on these parameters and the set grid resolution, interpolate the rainfall station data into a spatial distribution grid, and a more spatially correlated rainfall spatial distribution can be obtained.

[0039] According to one aspect of the present application, step S15 is further as follows: Step S15a: Extract the historical rainfall and historical potential evapotranspiration values of the rural and pastoral area, and calculate the historical available water volume of the pastoral area; Step S15b: Extract the crop types, irrigation methods, soil characteristics, crop water requirement coefficients, and crop growth cycles of the rural and pastoral area, calculate the crop water requirements at each time point for each grid in the rural and pastoral area, and summarize the domestic and ecological water requirements to obtain the historical water requirements of the rural and pastoral area; Step S15c: Extract the water requirement data at each time point for each grid in the rural and pastoral area. Taking the grid as a unit, map the water requirement to colors to draw a drought distribution map. The colors corresponding to the water requirements from small to large change from light to deep. Superimpose the distribution maps at each time point, calculate the total and average water requirements for each grid, extract the grids where both the total and average water requirements are greater than the threshold, and divide the connected grids into the same area to obtain several drought centers; Preferably, it can also be: extracting the water demand data of each grid in the rural and pastoral areas at each time point, mapping the water demand with colors in units of grids to draw a drought distribution map, where the colors corresponding to the water demand from small to large are from light to deep, overlaying the distribution maps at each time point to calculate the total water demand and average value of each grid, using an adaptive threshold determination method based on statistical analysis to calculate the threshold, extracting the grids where both the total water demand and the average value are greater than the threshold, and dividing the connected grids into the same area using a dual connectivity judgment criterion based on spatial adjacency relationship and water demand similarity to obtain several drought centers.

[0040] Furthermore, calculate the mean and standard deviation of the water demand of all grids, calculate the skewness and kurtosis of the water demand in the study area, determine the adjustment coefficient according to the distribution form. When the absolute value of the skewness is less than or equal to 0.5 and the absolute value of the difference between the kurtosis and 3 is less than or equal to 0.5, the adjustment coefficient takes the value of 1.5. When the skewness is greater than 0.5, the adjustment coefficient takes the value of 1.2. When the skewness is less than -0.5, the adjustment coefficient takes the value of 1.8. Adjust the adjustment coefficient according to the kurtosis, and use a sliding window smoothing process to obtain the final threshold.

[0041] Furthermore, judge that two grids are adjacent within the 8-neighborhood range and the distance between the grid center points is less than or equal to the square root of 2 times the grid side length, calculate the relative error of the difference in water demand between the two grids. When the relative error is less than the similarity threshold, it is determined to be connected. The similarity threshold is determined according to the coefficient of variation of water demand for all adjacent grid pairs in the study area. Use the depth-first search method to identify the connected areas, and set a minimum area constraint for the regions to avoid forming too small drought centers.

[0042] Step S15d: Based on the spatial distribution of the drought centers, use Thiessen polygons to divide the rural and pastoral areas into several sub-regions, and each sub-region contains one drought center.

[0043] In a certain embodiment, specifically: The area of rural and pastoral area A is 500 square kilometers. If the grid resolution is set to 100 meters × 100 meters, there are a total of 50,000 grids. Extract the historical rainfall data and historical potential evapotranspiration values of this rural and pastoral area in the past 10 years from the local meteorological department and water resources monitoring database. For example, in January 2014, the average rainfall in this rural and pastoral area was 10 mm, and the average potential evapotranspiration was 15 mm; in February 2014, the average rainfall was 8 mm, and the average potential evapotranspiration was 12 mm, obtaining 120 months of relevant data. Convert rainfall and potential evapotranspiration data into raster data. Through the raster calculator, calculate the available water volume for each month pixel by pixel, where the available water volume = rainfall - potential evapotranspiration. For example, in January 2014, the rainfall in a certain grid was 12 mm and the potential evapotranspiration was 16 mm, so the available water volume of this grid was -4 mm. Obtain the raster dataset of historical available water volume in pastoral and agricultural area A over the past 10 years; Extract the crop types, irrigation methods, soil characteristics, crop water requirement coefficients, and crop growth cycles in the pastoral and agricultural area, specifically: The proportion of wheat planting area is 30%, with 15,000 grids; the proportion of corn planting area is 25%, with 12,500 grids; the proportion of forage planting area is 45%, with 22,500 grids; Drip irrigation covers 60% of the farmland, with 30,000 grids; sprinkler irrigation covers 30%, with 15,000 grids; flood irrigation covers 10%, with 5,000 grids; The sandy soil area accounts for 20%, with 10,000 grids; the loam area accounts for 50%, with 25,000 grids; the clay area accounts for 30%, with 15,000 grids; The water requirement coefficient of wheat in the early growth stage is 0.3, in the middle stage is 0.6, and in the late stage is 0.4; the water requirement coefficient of corn in the early growth stage is 0.2, in the middle stage is 0.7, and in the late stage is 0.3; the average water requirement coefficient of forage during the growth season is 0.5; The growth cycle of wheat is 5 months, the growth cycle of corn is 4 months, and forage grows throughout the year.

[0044] For each grid, calculate the crop water requirement at each time point according to its crop type, growth cycle, water requirement coefficient, and irrigation method, combined with meteorological data; Take a loam grid planted with wheat as an example. In the middle growth stage, the rainfall is 20 mm and the potential evapotranspiration is 30 mm. Referring to the local meteorological data and crop water requirement coefficient, the crop water requirement of this grid is (30 - 20) × 0.6 = 6 mm; The population of this pastoral and agricultural area is about 10,000 people, and the per capita domestic water quota per month is 3 m 3 . Converting it to each grid, the domestic water requirement of each grid is approximately 10,000 × 3 ÷ 50,000 = 0.6 m 3 . Converting it to depth is approximately 0.6 × 1000 ÷ (100 × 100) = 0.06 mm; The ecological water requirement quota of this pastoral and agricultural area is 0.5 mm per month. Summarize the total water requirement at each time point for each grid, and then obtain the historical water requirement data of the pastoral and agricultural area; Extract the water demand data at each time point for each grid. Taking the grid as a unit, map the water demand to colors to draw a drought distribution map. Set the colors to change from light blue to dark blue corresponding to the water demand from small to large. Superimpose the drought distribution maps of each month, and calculate the total water demand and average value of each grid over 10 years; The average water demand of all grids over 10 years is 500 mm. Set the threshold to 1.5 times the average value, that is, 750 mm. Extract the grids where both the total water demand and the average value are greater than this threshold. The total water demand over 10 years is 800 mm, and the average value is 80 mm. Divide the connected grids into the same area to determine several drought centers.

[0045] In step S15c of this embodiment, the threshold is determined by an adaptive method based on statistical analysis to solve the technical problem that a fixed threshold cannot adapt to the differences in hydrological characteristics of different regions. Specifically, the threshold T = μ + k×σ; where μ is the mean of the water demand of all grids, σ is the standard deviation, and k is an adjustment coefficient.

[0046] The method for determining the adjustment coefficient k includes the following steps: First, calculate the distribution characteristic parameters of the water demand in the study area, including skewness S and kurtosis K; Then, determine the value of k according to the distribution form: when |S| ≤ 0.5 and |K - 3| ≤ 0.5, the data is close to a normal distribution, and k takes the value of 1.5; when S > 0.5, the data is right-skewed, and k takes the value of 1.2; when S < -0.5, the data is left-skewed, and k takes the value of 1.8; when |K - 3| > 0.5, adjust the value of k according to the kurtosis. When the kurtosis is high, k increases by 0.2, and when the kurtosis is low, k decreases by 0.2.

[0047] It should be noted that to improve the stability of the threshold, sliding window smoothing processing is adopted: T final = 0.4×T current + 0.3×T previous + 0.3×T next ; where T current is the currently calculated threshold, and T previous and T next are the thresholds of adjacent time periods respectively.

[0048] In some embodiments, optionally, the quantile method is used to determine the threshold. Take the 75th percentile as the reference threshold. For extremely arid regions, it can be adjusted to the 85th percentile.

[0049] For example, in a 500-square-kilometer research area, the average water demand of the grid is 45 mm, the standard deviation is 12 mm, and the skewness is 0.3. Then the value of k is 1.5, and the calculated threshold T = 45 + 1.5×12 = 63 mm. After smoothing by the sliding window, the final threshold is 61.5 mm.

[0050] In step S15c, the connected grids are divided into the same area to obtain several drought centers. The implementation process is as follows: In this embodiment, the connectivity judgment adopts a dual standard based on spatial adjacency relationship and water demand similarity to solve the unreasonable clustering problem that may be caused by simple spatial adjacency. Specifically, two grids are determined to be connected if they simultaneously meet the following conditions: Spatial adjacency condition: The two grids are adjacent within the 8-neighborhood range, that is, the distance d between the grid center points ≤ sqrt(2)×grid_size, where grid_size is the side length of the grid; Water demand similarity condition: The relative error of the difference in water demand between the two grids is less than the threshold. The calculation formula is: |W i -W j | / max(W i ,W j )≤ε; Among them, W i and W j are the water demands of the two grids respectively, and ε is the similarity threshold.

[0051] The determination method of the similarity threshold ε is as follows: First, calculate the coefficient of variation of water demand CV = σ neighbor / μ neighbor ; Then determine the value of ε according to the coefficient of variation: When CV ≤ 0.2, ε = 0.15; when 0.2 < CV ≤ 0.4, ε = 0.25; when CV > 0.4, ε = 0.35.

[0052] The connected region recognition algorithm adopts the depth-first search (DFS) method: starting from the seed grid that meets the threshold condition, recursively search for all adjacent grids that meet the connection conditions until there are no new connected grids, forming a connected region.

[0053] It should be noted that to avoid forming too small drought centers, a minimum area constraint is set; The number of grids included in the connected region is not less than: N min =max(9,0.01×N total ), where N total is the total number of grids in the research area. Small areas that do not meet the area constraint will be merged into the largest adjacent connected region.

[0054] In some embodiments, a correction method based on topographic connectivity is optionally adopted. When two high water demand grids are separated by topographic barriers (such as ridges, rivers), they are not determined to be connected even if the spatial and similarity conditions are met.

[0055] For example, in a study area with a grid resolution of 100m×100m, the water demands of two adjacent grids are 68mm and 72mm respectively, and the relative error is 5.9%, which is less than the threshold of 15%. The spatial distance is 100m, meeting the connection conditions, and they are divided into the same arid center area.

[0056] Through the above dual judgment criteria, the accuracy and rationality of arid center identification are ensured, and the problem of incorrect clustering of grids with significant differences in hydrological characteristics but close proximity in space is avoided.

[0057] According to one aspect of the present application, step S15 can also be: Construct a time-reversal prediction model to calculate the hydrological state constraint conditions for future periods, and obtain the historical available water volume in the pastoral and agricultural areas through backpropagation; Combine the time-reversal constraint factor to calculate the predictive crop water demand at each time point for each grid in the pastoral and agricultural areas, and summarize the future water demand change trends of life and ecology to obtain the predictive water demand in the pastoral and agricultural areas with time-reversal weight fusion; Calculate the dynamic prediction threshold based on time-reversal constraints, map the predictive water demand risk to colors to draw a drought risk distribution map, calculate the predictive drought intensity for each grid, and extract the grids with a predictive drought intensity greater than the dynamic threshold; Use the backpropagation connectivity algorithm to divide the pastoral and agricultural areas into several predictive drought impact sub-regions, and each sub-region contains a predictive drought center.

[0058] Furthermore, the backpropagation connectivity algorithm includes: Calculate the backpropagation gradient, set spatial gradient constraints, time synchronization constraints, and influence intensity constraints, identify the backpropagation seed points, construct a propagation tree structure in the reverse gradient direction with the seed points as the root nodes, determine the clustering boundary based on the attenuation characteristics of time-reversal intensity, adopt a competitive allocation mechanism to avoid clustering overlap, and adopt multi-time scale clustering verification to ensure the stability of the clustering results.

[0059] Furthermore, the calculation of the predictive water demand in the pastoral and agricultural areas with time-reversal weight fusion includes: The anticipatory water demand is calculated by multiplying the basic water demand by the future constraint correction factor and the time reversal correction factor. The future constraint correction factor is determined by the water demand trend factor, and the time reversal correction factor is calculated based on the time reversal operator. The domestic water demand takes into account population growth and changes in water use habits. The ecological water demand incorporates the response mechanism of the ecosystem to drought. The multi-scenario weighted average method is used to improve the accuracy of the anticipatory water demand.

[0060] Specifically, it includes the following steps: Step S151: Extract the historical rainfall and historical potential evapotranspiration values in the pastoral and agricultural areas, construct a time reversal prediction model, calculate the hydrological state constraint conditions in the future period, and obtain the historical available water volume in the pastoral area for backpropagation; In this embodiment, the time reversal prediction adopts a hydrological state deduction method based on causal inversion to solve the technical problem that traditional historical statistics cannot predict the future drought development trend. Specifically, a time reversal operator T_rev is constructed: Ψ(x,y,t) = ∫[t to t+ΔT] K_rev(τ-t) × S_future(x,y,τ) × w(τ) dτ; Where, K_rev(τ-t) is the time reversal kernel function, S_future(x,y,τ) is the predicted hydrological state at the future time τ, w(τ) is the time weight function, and ΔT is the prediction time window.

[0061] Time reversal kernel function: K_rev(τ-t) = exp(-α×(τ-t)²) × cos(β×(τ-t)); Where, α is the attenuation coefficient, and its value range is 0.01 - 0.05, preferably 0.03; β is the oscillation frequency parameter, and its value range is 0.1 - 0.5, preferably 0.25.

[0062] It should be noted that the future hydrological state S_future is obtained by predicting through an improved long short-term memory network. The prediction time window is set to 30 - 90 days, preferably 60 days. The time weight function w(τ) adopts an exponential decay form: w(τ) = exp(-γ×(τ-t) / ΔT) Where, γ is the weight decay factor, and its value is 1.5 - 2.5, preferably 2.0.

[0063] For example, for the calculation of the hydrological state of a certain grid in April, considering the decreasing trend of the predicted rainfall and the increasing trend of temperature within the next 60 days, the time reversal operator Ψ = 0.75, indicating a high future drought risk at this location.

[0064] Step S152: Extract the crop types, irrigation methods, soil characteristics, crop water requirement coefficients, and crop growth cycles in rural and pastoral areas. Combine the time-reversal constraint factors to calculate the predictive crop water requirements at each time point for each grid in rural and pastoral areas, summarize the future water requirement change trends of life and ecology, and obtain the predictive water requirements of rural and pastoral areas that incorporate the time-reversal weight. In this embodiment, the calculation of predictive water requirements uses a dynamic weight allocation method based on future constraints to solve the technical problem that traditional water requirement calculations only consider the current state and cannot reflect the changes in future water requirement pressure.

[0065] Specifically, the predictive water requirement W_predictive(x,y,t) = W_base(x,y,t) × [1 + λ_future × F_trend(x,y,t)] × R_rev(x,y,t); where W_base(x,y,t) is the basic water requirement, λ_future is the future constraint weight coefficient, F_trend(x,y,t) is the water requirement trend factor, and R_rev(x,y,t) is the time-reversal correction factor.

[0066] The basic water requirement W_base(x,y,t) = ET0(t) × K_c(crop,stage) × K_s(soil) × A_effective(x,y); where ET0 is the reference evapotranspiration, K_c is the crop coefficient, K_s is the soil coefficient, and A_effective is the effective irrigation area.

[0067] The water requirement trend factor F_trend(x,y,t) = Σ[i=1 to n] w_stage_i × K_c_future(t+i×Δt) / K_c_current(t); where w_stage_i is the weight of the i-th future growth stage, K_c_future is the predicted future crop coefficient, and n is the number of predictive stages, usually taking 3 - 5 stages.

[0068] The time-reversal correction factor R_rev is calculated based on the time-reversal operator in Step S151: R_rev(x,y,t) = 1 + β_rev × [Ψ(x,y,t) - Ψ_mean] / Ψ_std; where β_rev is the inversion correction coefficient, with a value range of 0.1 - 0.4, preferably 0.25; Ψ_mean and Ψ_std are the mean and standard deviation of the regional time-reversal operator, respectively.

[0069] The future constraint weight coefficient λ_future is dynamically adjusted according to the prediction confidence and time distance: λ_future = λ_max × confidence_score × exp(-decay_rate × t_distance); Where, λ_max is the maximum weight coefficient, with a value range of 0.3 - 0.7, preferably 0.5; confidence_score is the prediction confidence; decay_rate is the time decay rate, with a value range of 0.02 - 0.05; t_distance is the time distance.

[0070] The predictive calculation of domestic water demand takes into account population growth and changes in water use habits: W_domestic_predictive = W_domestic_base × [1 + r_population × t_future + δ_habit × Ψ(x,y,t)]; Where, r_population is the population growth rate, and δ_habit is the coefficient of change in water use habits.

[0071] The predictive calculation of ecological water demand incorporates the response mechanism of the ecosystem to drought: W_ecological_predictive = W_ecological_base × [1 + ε_stress × max(0, Ψ(x,y,t) - Ψ_threshold)]; Where, ε_stress is the ecological stress coefficient, and Ψ_threshold is the ecological stress threshold.

[0072] It should be noted that to improve the accuracy of predictive water demand, a multi-scenario weighted average method is adopted: W_final = Σ[k=1 to m] p_scenario_k × W_predictive_k; Where, m is the number of scenarios, usually set to 3 (optimistic, baseline, pessimistic), and p_scenario_k is the probability weight of the k-th scenario.

[0073] In some embodiments, an optional rolling update mechanism is adopted to update the time inversion weight and predictive water demand every 7 - 15 days according to the latest observation data and prediction results.

[0074] For example, the basic water requirement of a certain wheat planting grid in mid-April is 8.5 mm / week. Considering the increasing trend of water demand during the future growth peak period, F_trend = 1.35; the time reversal operator Ψ of this grid is 1.2, which is higher than the regional average value of 0.8, and R_rev = 1.1; the final predictable water requirement is 8.5 × [1 + 0.5 × 1.35] × 1.1 = 14.6 mm / week, which is 72% higher than the basic water requirement.

[0075] Step S153: Extract the water demand data of each grid in the rural and pastoral areas at each time point, calculate the dynamic prediction threshold based on the time reversal constraint, draw the drought risk distribution map by mapping the predictable water demand risk to colors, with the colors ranging from light to dark corresponding to the risk from small to large, superimpose the time reversal weight, calculate the predictable drought intensity of each grid, extract the grids with predictable drought intensity greater than the dynamic threshold, and divide the grids that meet the reverse connectivity condition into the same area to obtain several predictable drought centers. In this embodiment, the dynamic threshold T_dynamic adopts an adaptive calculation method based on the time reversal constraint to solve the technical problem that the traditional static threshold cannot reflect the dynamic evolution trend of drought.

[0076] Specifically, the calculation formula for the dynamic threshold is: T_dynamic(x,y,t) = T_base × [1 + λ_rev × Ψ(x,y,t)] × F_trend(t); where, T_base is the basic threshold, λ_rev is the time reversal weight coefficient, Ψ(x,y,t) is the value of the time reversal operator calculated in step S15a, and F_trend(t) is the trend correction factor.

[0077] Basic threshold: T_base = Q 75 + k_adaptive × IQR; where, Q 75 is the 75th percentile of the water demand, IQR is the interquartile range, and k_adaptive is the adaptive adjustment coefficient.

[0078] The adaptive adjustment coefficient k_adaptive is determined according to the regional hydrological variability: k_adaptive = 0.5 × [1 + tanh(CV_hydro - 0.3)]; where, CV_hydro is the regional hydrological variation coefficient. When CV_hydro ≤ 0.3, k_adaptive ≈ 0.5; when CV_hydro ≥ 0.6, k_adaptive ≈ 1.0.

[0079] The time reversal weight coefficient λ_rev ranges from 0.2 to 0.8 and is dynamically adjusted according to the prediction confidence: λ_rev = 0.2 + 0.6 × confidence_score; Among them, confidence_score is the confidence score of the future state prediction, ranging from 0 to 1.

[0080] The trend correction factor F_trend(t) takes into account seasonal and interannual variations: F_trend(t) = 1 + A_seasonal × sin(2π×t / 365) + A_interannual × sin(2π×t / 1095); Among them, A_seasonal is the seasonal amplitude coefficient, which ranges from 0.1 to 0.3; A_interannual is the interannual amplitude coefficient, which ranges from 0.05 to 0.15.

[0081] It should be noted that the calculation of the predictive drought intensity I_predictive integrates the current state and future constraints: I_predictive(x,y,t) = α_current × W_current(x,y,t) + α_future × Ψ(x,y,t); Among them, α_current and α_future are the weight coefficients of the current and future states, satisfying α_current + α_future = 1, and preferably α_current = 0.4 and α_future = 0.6.

[0082] In some embodiments, a multi-level threshold strategy may be optionally adopted to set warning thresholds, severe thresholds, and extreme thresholds, which correspond to different levels of predictive drought centers, respectively.

[0083] For example, in the calculation in May in a certain agricultural and pastoral area, the base threshold T_base = 52 mm, the time reversal operator Ψ = 1.2, λ_rev = 0.5, and F_trend = 1.1, then the dynamic threshold T_dynamic = 52 × [1 + 0.5 × 1.2] × 1.1 = 86.3 mm.

[0084] Step S154: Based on the spatiotemporal distribution characteristics of the predictive drought centers, the back propagation connectivity algorithm is used to divide the agricultural and pastoral areas into several predictive drought-affected sub-regions, each of which contains a dominant predictive drought center.

[0085] In this embodiment, the reverse connectivity judgment adopts a propagation path recognition method based on the time-reversal gradient field to solve the technical problem that traditional spatial adjacency cannot reflect the dynamic expansion of the drought influence range. Specifically, the calculation formula of the backward propagation gradient G_backward is as follows: G_backward(x,y,t) = -grad [Ψ(x,y,t+Δt) - Ψ(x,y,t)]; Where, grad is the gradient operator, Δt is the time step, with a value range of 1 - 7 days, preferably 3 days.

[0086] The reverse connectivity conditions include the following three constraints: Spatial gradient constraint: The consistency of the reverse gradient directions between two grids satisfies: cos(θ_gradient) = G_i · G_j / (|G_i| × |G_j|) ≥ θ_threshold; Where, θ_threshold is the angle threshold, with a value range from cos(45°) to cos(30°), preferably cos(35°) ≈ 0.819.

[0087] Time synchronization constraint: The phase difference of the drought development time between two grids is less than the allowable range: |Phase_i(t) - Phase_j(t)| ≤ Δφ_max; where, Phase is the time phase of the drought development, and Δφ_max is the maximum phase difference, with a value range from π / 6 to π / 4, preferably π / 5.

[0088] Influence intensity constraint: The relative difference of the predictive drought intensity of the connected grids satisfies: |I_predictive_i -I_predictive_j| / max(I_predictive_i, I_predictive_j) ≤ ε_intensity; where, ε_intensity is the intensity similarity threshold, with a value range of 0.15 - 0.35, preferably 0.25.

[0089] The predictive clustering algorithm adopts an improved backpropagation clustering method: Step 1: Identify the backpropagation seed points: Starting from the grid with the highest predictive drought intensity, calculate its backpropagation influence domain: Seed_influence(x0,y0) = {(x,y) | G_backward(x,y) → (x0,y0)}; Step 2: Construct the backpropagation tree: Taking the seed point as the root node, construct the propagation tree structure according to the reverse gradient direction: Tree_node(x,y) = {parent: (x0,y0), children: {(xi,yi)}, weight: I_predictive(x,y)}; Step 3, dynamic boundary determination: Determine the clustering boundary based on the decay characteristics of time-reversed intensity. Boundary_condition: I_predictive(x,y) ≥ I_seed × exp(-β_decay × d_propagation); Where, I_seed is the predictive drought intensity of the seed point, β_decay is the decay coefficient, with a value range of 0.1 - 0.3, and d_propagation is the propagation distance.

[0090] It should be noted that to avoid clustering overlap, a competitive allocation mechanism is adopted: when a grid satisfies the connectivity conditions of multiple clusters simultaneously, it is assigned to the cluster center with the maximum backpropagation intensity.

[0091] In some embodiments, multi-time-scale clustering verification is optionally adopted to ensure the stability of the clustering results under different time windows. The verification method is: Stability_score = Σ[Overlap(Cluster_t, Cluster_t+Δt)] / |Cluster_t|; when the stability score is lower than 0.7, adjust the clustering parameters and recalculate.

[0092] For example, in a 500-square-kilometer research area, 3 predictive drought centers are identified. The backpropagation influence domain of the main center covers 180 square kilometers, and the secondary centers cover 120 and 80 square kilometers respectively. The time phase differences of the three centers are 5 days, 8 days, and 12 days respectively, which satisfy the spatio-temporal evolution law of predictive drought development.

[0093] Through the above method for identifying predictive drought centers by time reversal, the technical leap from traditional historical statistical identification to future-constrained prediction is achieved, improving the forward-looking and accuracy of drought center identification, and providing a scientific basis for the advanced allocation of water resources in rural and pastoral areas.

[0094] As Figure 3 shown, according to one aspect of the present application, step S2 is further as follows: Step S21, collect hydrological station detection data. Use the cross-sectional flow calculation method to collect the incoming water volume data at different frequencies. The incoming water volume data is the basis for the water source planning of the irrigation area. The incoming water volume at different frequencies can reflect the supply stability and the abundance and drought changes of the water source. At a suitable section of the hydrological station, a current meter device is used to measure the flow velocities at different positions of the section, calculate the average flow velocity, and at the same time measure the area of the section to calculate the flow rate. The measurement is repeated multiple times and the average value is taken to obtain the incoming water volume data at different frequencies.

[0095] Step S22: Collect hydrological data and the dynamic changes of water levels, detect water quality parameters, water quality indicators and water volume parameters, and draw a water layer distribution map. Step S23: Calculate the water quality preservation rate based on the water source volume data and the water demand, determine the feasibility of water intake based on the elevation difference between the water source water level and the water intake point of the irrigation area, and screen to obtain the water intake points. Step S24: Based on the spatial distribution of the drought centers, construct surface water source engineering plans and groundwater source engineering plans respectively, and combine them to obtain the water source engineering layout plans for each sub-region.

[0096] According to one aspect of the present application, the step S24 is further as follows: Step S24a: Collect the groundwater level monitoring data and river-lake water level data of each sub-region, combine with the water demand, draw a water level contour map, extract the coordinates of the water level contours, and calculate the slope of each contour. Step S24b: Set an initial buffer distance, generate a circular initial buffer zone with the drought center as the center, correct the initial buffer zone based on the density of the water level contours, set a slope threshold, and divide the buffer zone into a core buffer zone and a secondary buffer zone based on the calculated contour slopes. Step S24c: Construct surface water source engineering plans and groundwater source engineering plans for each sub-region for the core buffer zone and the secondary buffer zone respectively, input the pre-constructed water resources joint operation model, and solve the model to obtain the water source engineering layout plans for each sub-region.

[0097] In a certain embodiment, specifically: The area of rural and pastoral area B is about 500 square kilometers. Based on the determined drought centers, it is divided into 5 sub-regions. Through the monitoring network of the water conservancy department, collect the groundwater level monitoring data of each sub-region in the past 3 years. In sub-region 1, on January 1, 2021, the groundwater level at monitoring point M was 15 meters, and the groundwater level at monitoring point N was 16 meters; at the same time, collect the river-lake water level data. For a certain river in sub-region 2, the water level on January 1, 2021 was 8 meters, and the water level on February 1, 2021 was 7.8 meters. Combine with the water demand data of each sub-region calculated above. The annual water demand of sub-region 1 is 8 million cubic meters, and the annual water demand of sub-region 2 is 6 million cubic meters. Interpolate the groundwater level data and river and lake water level data respectively to generate the groundwater level raster data and the river and lake water level raster data, and draw the groundwater level contour map and the river and lake water level contour map respectively; Calculate the slope of each contour line based on the coordinate data. For example, for a certain contour line, select two adjacent points (x1, y1) and (x2, y2), and the slope k = (y2 - y1) ÷ (x2 - x1). For example, the coordinates of two adjacent points of a certain groundwater level contour line are (100, 15) and (200, 14), and its slope k = (14 - 15) ÷ (200 - 100) = -0.01; Set the initial buffer distance to 500 meters, and take each drought center as the center of the circle to generate a circular initial buffer zone; Set the slope threshold to 0.02, and based on the calculated slope of the water level contour line, correct the initial buffer zone. If the absolute value of the contour line slope is greater than the threshold, it means that the water level changes violently. If it is less than the threshold, the water level changes relatively gently. Divide the buffer zone into a core buffer zone and a secondary buffer zone. Divide the area where the absolute value of the contour line slope is greater than the threshold into the core buffer zone, and the area less than the threshold into the secondary buffer zone; For the core buffer zone, considering the large variation in water resources, in the surface water source project plan, plan to build a reservoir or pond project with strong regulation capacity; in the groundwater source project plan, arrange deep wells; for the secondary buffer zone, the surface water source project plan adopts small diversion canals and pumping station facilities; the groundwater source project plan arranges shallow wells; Input the constructed surface water and groundwater source project plans for each sub-region into the pre-constructed water resources joint operation model for solution to obtain the water source project layout plans for each sub-region. In sub-region 1, it is finally determined to build 1 medium-sized reservoir in the core buffer zone and arrange 3 deep wells; build 5 small diversion canals in the secondary buffer zone and arrange 10 shallow wells.

[0098] In another embodiment, the step S24 is further as follows: Step S24a: Collect the groundwater level monitoring data and river and lake water level data of each sub-region, combine with the water demand, draw the water level contour map, extract the water level contour coordinates, and calculate the slope of each contour line; Step S24b: Take the drought center as the center of the circle, set the initial buffer distance to generate a circular initial buffer zone, use the spatial distribution of the water level contour line, the water demand distribution, etc. as input parameters, and optimize the initial buffer zone with the project construction cost, water resource utilization efficiency, and ecological impact as the optimization objectives to obtain regions with different functions and priorities. Specifically, divide the area close to the water source and with low construction cost into the priority construction area, and divide the area with complex water resource conditions and high construction cost into the cautious construction area; Step S24c: Construct surface water source project plans and groundwater source project plans respectively according to the optimized divided areas.

[0099] As Figure 4 shown, according to one aspect of the present application, the step S3 is further as follows: Step S31: Collect the terrain slope, soil permeability coefficient, crop planting area distribution and the position of the water outlet of the water source project in the irrigation area; Step S32: Select an irrigation system based on the terrain slope, and calculate the pipe diameter, dripper spacing, channel cross-section size and pump station lift; For areas with a terrain slope less than 15°, select drip irrigation and pipe irrigation systems, and calculate the pipe diameter and dripper spacing according to the irrigation flow rate and pressure requirements based on relevant formulas; for areas with a terrain slope greater than 15°, select pump stations and open channel irrigation systems, and calculate the channel cross-section size and pump station lift according to the terrain height difference and flow rate using hydraulic formulas, thereby determining the irrigation systems suitable for different terrain slopes and accurate engineering parameters.

[0100] Step S33: Construct a drainage project based on the rainstorm intensity data and terrain water collection data of each sub-region to obtain the irrigation area project layout plan for each sub-region.

[0101] Collect the rainstorm intensity data and terrain water collection data of each sub-region, use the drainage project design principles and calculation formulas to determine the direction, cross-section size and slope parameters of the drainage channels, design facilities such as drainage pump stations, construct a drainage project, and combine it with the irrigation project to form a complete irrigation area project layout plan.

[0102] As Figure 5 shown, according to one aspect of the present application, the step S4 is further as follows: Step S41: Construct an evaluation index system, including economic value indicators, social benefit indicators and ecological benefit indicators; Combined with the characteristics of the rural and pastoral water network irrigation areas, select representative indicators, take the net present value, internal rate of return, and investment payback period as economic value indicators, quality of life, employment opportunities, and social stability as social benefit indicators, and the ecological environment quality index as an ecological benefit indicator to construct an evaluation index system.

[0103] Step S42: Calculate the index values of the water network irrigation area project layout plans for each sub-region respectively; Step S43: Construct an economic evaluation model, extract the index values of the water network irrigation area project layout plans for each sub-region as model inputs, calculate the comprehensive benefit scores of the water network irrigation area project layout plans for each sub-region, and the one with the highest score is the project scale layout plan for the rural and pastoral water network irrigation areas.

[0104] According to one aspect of the present application, the step S42 is further as follows: Step S42a: Calculate the economic benefit index values by using the dynamic analysis method, including: net present value, internal rate of return, and payback period; Step S42b: Calculate the social benefit indicators by using the analytic hierarchy process, including: quality of life, employment opportunities, and social stability; Step S42c: Calculate the ecological benefit indicators by using the ecological environment quality index method.

[0105] According to one aspect of the present application, the step S43 is further as follows: Step S43a: Construct a multi-level comprehensive evaluation model, including an objective layer, a criterion layer, and an index layer. The objective layer is the comprehensive benefit of the engineering scale layout plan of the rural and pastoral water network irrigation area. The criterion layer includes economic benefits, social benefits, and ecological benefits. The index layer is further subdivided according to the criterion layer; Step S43b: Based on the 1-9 scale method, compare and score the importance of the indicators pairwise, construct a judgment matrix, calculate the eigenvector and the maximum eigenvalue of the judgment matrix, conduct a consistency test, determine the weights of each indicator, and configure the multi-level comprehensive evaluation model; Step S43c: Extract the indicator values of the engineering layout plan of the water network irrigation area in each sub-region as the input of the multi-level comprehensive evaluation model, calculate the comprehensive benefit scores of the engineering layout plans of the water network irrigation area in each sub-region, and the one with the highest score is the engineering scale layout plan of the rural and pastoral water network irrigation area.

[0106] In a certain embodiment, specifically: Invite 5 experts in the fields of water conservancy engineering, economics, sociology, and ecology to compare and score the importance of the indicators pairwise based on the 1-9 scale method. For example, for the comparison of the importance of economic benefits and social benefits, 3 experts think the score of economic benefits is 3, and 2 experts think they are equally important with a score of 1. The comprehensive average value is 2.2; Construct the judgment matrix as follows: Economic benefits: 1, 2.2, 3; Social benefits: 1 / 2.2, 1, 2; Ecological benefits: 1 / 3, 1 / 2, 1; Calculate the eigenvector and the maximum eigenvalue of the judgment matrix, and obtain the weights of economic benefits, social benefits, and ecological benefits as 0.5, 0.3, and 0.2 respectively; Score and calculate each indicator in the indicator layer to obtain the specific weights: Indicator layer of economic benefits: weight of net present value 0.4, weight of internal rate of return 0.3, weight of payback period option 0.3; Social benefit index layer: Quality of life weight 0.4, Employment opportunity weight 0.3, Social stability weight 0.3; Ecological benefit index layer: Vegetation coverage rate weight 0.2, Soil moisture content weight 0.2, Water quality category weight 0.3, Biodiversity index weight 0.3; After consistency test, the consistency ratio of all judgment matrices is less than 0.1, meeting the consistency requirement, and determining the above weight configuration multi-level comprehensive evaluation model; Substitute the index values of each sub-region into the model to calculate the comprehensive benefit score: Sub-region 1: Economic benefit score = (Net present value score × 0.4 + Internal rate of return score × 0.3 + Investment payback period score × 0.3) × 0.5 = ((486.7 / 510) × 10 × 0.4 + 12 / 13 × 10 × 0.3 + (6.5 / 7) × 10 × 0.3) × 0.5 ≈ 4.43; Social benefit score = 8.2 × 0.3 = 2.46; Ecological benefit score = 4.4 × 0.2 = 0.88; Comprehensive benefit score = 4.43 + 2.46 + 0.88 = 7.77; Sub-region 2: Economic benefit score = ((320 / 510) × 10 × 0.4 + 10 / 13 × 10 × 0.3 + (8 / 7) × 10 × 0.3) × 0.5 ≈ 3.47; Social benefit score = 7.8 × 0.3 = 2.34; Ecological benefit score = 3.8 × 0.2 = 0.76; Comprehensive benefit score = 3.47 + 2.34 + 0.76 = 6.57; Sub-region 3: Economic benefit score = ((510 / 510) × 10 × 0.4 + 13 / 13 × 10 × 0.3 + (6.5 / 7) × 10 × 0.3) × 0.5 ≈ 4.76; Social benefit score = 8.5 × 0.3 = 2.55; Ecological benefit score = 4.7 × 0.2 = 0.94; Comprehensive benefit score = 4.76 + 2.55 + 0.94 = 8.25; After comparison, the comprehensive benefit score of sub-region 3 is the highest. Therefore, the water network irrigation area project layout plan of sub-region 3 is determined as the project scale layout plan of the rural and pastoral water network irrigation area.

[0107] By constructing an economic evaluation model and quantitatively scoring the comprehensive benefits of each plan, the optimal plan can be scientifically and objectively selected to achieve the optimal allocation of resources; In this embodiment, the multi-level comprehensive evaluation model can comprehensively consider the influences of multiple levels and multiple indicators. The 1-9 scale method and judgment matrix calculation can scientifically determine the index weights, ensuring the rationality and reliability of the evaluation results.

[0108] The present invention fully considers the influence of terrain and climate factors on water resources in rural and pastoral areas. First, it collects data on rural and pastoral areas, processes the rural and pastoral areas through grid division, constructs a terrain feature matrix, calculates the evapotranspiration value using a long short-term memory model, conducts soil and water balance calculations in combination with rainfall, accurately extracts the drought center, and then divides sub-regions; Comprehensively considering the sources, storage, distribution, and discharge of water resources, it ensures that the project layout can adapt to different water resource conditions and demands in rural and pastoral areas. It calculates the incoming water volume of each sub-region, draws a water layer distribution map, determines the water intake points based on the water quality retention rate, arranges water source projects based on the drought center, determines the irrigation system in combination with the terrain of the irrigation area, and constructs a drainage project based on the rainfall intensity and terrain water collection data to form a complete irrigation area project layout plan; Constructs an evaluation index system covering multiple aspects, calculates the index values of the water network irrigation area project layout plan for each sub-region, and then obtains the comprehensive benefit score through an economic evaluation model. The final project scale layout plan is selected through multi-dimensional evaluation.

[0109] According to another aspect of the present application, there is provided a system for determining the project scale layout of a water network irrigation area in rural and pastoral areas, which is characterized by including: At least one processor; and A memory communicatively connected to at least one of the processors; wherein, The memory stores instructions executable by the processor, and the instructions are used to be executed by the processor to implement the method for determining the project scale layout of a water network irrigation area in rural and pastoral areas as described in any one of the above.

[0110] The preferred embodiments of the present invention have been described in detail above. However, the present invention is not limited to the specific details in the above embodiments. Within the scope of the technical concept of the present invention, various equivalent transformations can be made to the technical solutions of the present invention, and these equivalent transformations all fall within the protection scope of the present invention.

Claims

1. A method for determining the scale and layout of water network irrigation projects in agricultural and pastoral areas, characterized in that: The steps include: Step S1: Collect agricultural and pastoral area data, grid the agricultural and pastoral areas, construct a terrain feature matrix, calculate the rainfall of each grid, use a pre-built long-short-term memory model to calculate the evapotranspiration value, perform soil and water balance calculations, extract several drought centers, and divide the agricultural and pastoral areas into several sub-areas; Step S2: Calculate the water inflow of each sub-region, draw a water layer distribution map, calculate the water quality rate based on the water source water volume data and water demand, determine the water intake, and construct a water source project layout plan for each sub-region based on the spatial distribution of the drought center; Step S3: Extract the water outlet of each sub-region's water source project, combine it with the irrigation area's topographic data, determine the irrigation system, construct a drainage project based on the rainstorm intensity data and topographic water catchment data of each sub-region, and obtain the irrigation project layout plan for each sub-region; Step S4: Construct an evaluation index system, calculate the index values of the water network irrigation project layout plan for each sub-region, input them into the pre-built economic evaluation model, calculate the corresponding comprehensive benefit score, and the one with the highest score is the scale layout plan of the water network irrigation project in the agricultural and pastoral areas.

2. The method for determining the scale layout of a water network irrigation project in an agricultural and pastoral area according to claim 1, characterized in that: The step S1 is further as follows: Step S11, collecting agricultural and pastoral area data, including: agricultural and pastoral area digital elevation data, meteorological data, rainfall station observation data, crop planting structure data, and livestock breeding scale data; Step S12: gridding the study area based on the agricultural and pastoral area terrain data, extracting agricultural and pastoral area digital elevation data and rainfall station observation data, and constructing an agricultural and pastoral area terrain feature matrix; Step S13: Calculate the terrain similarity between each grid and each rainfall station using the Euclidean distance improved by cosine similarity, and optimize the inverse distance weight formula based on the terrain similarity to obtain a hybrid weight function. Use the hybrid weight function to perform weighted averaging on each grid to obtain the corresponding rainfall. Step S14: Collect historical data of agricultural and pastoral areas to calculate historical potential evapotranspiration values of agricultural and pastoral areas, and train a pre-built long-short-term memory model together with the historical actual evapotranspiration values of agricultural and pastoral areas. Extract the agricultural and pastoral area data as input to the trained long-short-term memory model to calculate the actual evapotranspiration values of agricultural and pastoral areas. Step S15: Calculate the historical available water volume in the agricultural and pastoral areas based on the historical rainfall and historical potential evapotranspiration values in the agricultural and pastoral areas, perform soil and water balance calculations on the agricultural and pastoral areas based on the agricultural and pastoral area data, obtain the historical water demand in the agricultural and pastoral areas, calculate the historical water demand of each grid in the agricultural and pastoral areas, extract several drought centers, and divide the agricultural and pastoral areas into several sub-areas.

3. The method for determining the scale layout of a water network irrigation project in an agricultural and pastoral area according to claim 2, characterized in that: The step S12 is further as follows: Step S12a, extracting the agricultural and pastoral area terrain data, setting the grid resolution to grid the agricultural and pastoral area, and obtaining the agricultural and pastoral area terrain grid; Step S12b: Based on the grid resolution of the agricultural and pastoral terrain, the rainfall station data is converted into a spatial distribution grid using the inverse distance weighted method; Step S12c: Integrate the agricultural and pastoral area terrain grid and the rainfall spatial distribution grid to obtain the agricultural and pastoral area terrain feature matrix.

4. The method for determining the scale layout of a water network irrigation project in an agricultural and pastoral area according to claim 2, wherein: The step S15 is further as follows: Step S15a, extracting historical rainfall and historical potential evapotranspiration values in the agricultural and pastoral areas, and calculating the historical available water volume in the pastoral areas; Step S15b: extract the crop type, irrigation method, soil characteristics, crop water requirement coefficient and crop growth cycle in the agricultural and pastoral areas, calculate the crop water requirement at each time point in each grid in the agricultural and pastoral areas, summarize the living and ecological water requirements, and obtain the historical water requirements of the agricultural and pastoral areas; Step S15c: Extract water demand data for each grid at each time point in the agricultural and pastoral areas, map water demand to color and draw a drought distribution map using the grid as a unit, with the color corresponding to the water demand from light to dark from small to large. Superimpose the distribution maps at each time point, calculate the total and average water demand of each grid, extract grids whose total and average water demand are both greater than a threshold, and divide the connected grids into the same area to obtain several drought centers; Step S15d: Based on the spatial distribution of drought centers, the agricultural and pastoral areas are divided into several sub-areas using Thiessen polygons, each of which contains a drought center.

5. The method for determining the scale layout of a water network irrigation project in an agricultural and pastoral area according to claim 1, wherein: The step S2 is further as follows: Step S21: Collecting hydrological station detection data and using cross-sectional flow calculation method to collect water flow data at different frequencies; Step S22: collecting hydrological data and water level dynamic changes, detecting water quality parameters, water quality indicators and water quantity parameters, and drawing a water layer distribution map; Step S23: Calculate the water quality rate based on the water source water quantity data and the water demand, determine the feasibility of water extraction based on the elevation difference between the water source water level and the water intake point in the irrigation area, and select the water intake; Step S24: construct surface water source engineering plans and groundwater source engineering plans based on the spatial distribution of drought centers, and combine them to obtain water source engineering layout plans for each sub-region.

6. The method for determining the scale layout of a water network irrigation project in an agricultural and pastoral area according to claim 5, characterized in that: The step S24 is further as follows: Step S24a: Collect groundwater level monitoring data and river and lake water level data for each sub-region, draw a water level contour map based on water demand, extract the coordinates of the water level contour lines, and calculate the slope of each contour line; Step S24b: Set an initial buffer distance, generate a circular initial buffer zone with the drought center as the center, modify the initial buffer zone based on the density of water level contour lines, set a slope threshold, and divide the buffer zone into a core buffer zone and a secondary buffer zone based on the calculated contour line slopes; Step S24c: construct surface water source engineering schemes and groundwater source engineering schemes for each sub-region for the core buffer zone and the secondary buffer zone, input the pre-constructed water resources joint scheduling model, and solve the model to obtain the water source engineering layout scheme for each sub-region.

7. The method for determining the scale layout of a water network irrigation project in an agricultural and pastoral area according to claim 1, wherein: The step S3 is further as follows: Step S31: collecting the terrain slope of the irrigation area, the soil permeability coefficient, the distribution of crop planting areas, and the location of the water outlet of the water source project; Step S32: Select an irrigation system based on the terrain slope, and calculate the pipe diameter, dripper spacing, channel cross-sectional dimensions, and pump station head; Step S33: construct drainage projects based on the rainstorm intensity data and topographic water catchment data of each sub-region, and obtain irrigation project layout plans for each sub-region.

8. The method for determining the scale layout of a water network irrigation project in an agricultural and pastoral area according to claim 1, wherein: The step S4 is further as follows: Step S41: constructing an evaluation index system, including economic value index, social benefit index and ecological benefit index; Step S42: Calculate the index value of the water network irrigation project layout plan for each sub-region respectively; Step S43: construct an economic evaluation model, extract the index values of the water network irrigation project layout plan of each sub-region as the model input, calculate the comprehensive benefit score of the water network irrigation project layout plan of each sub-region, and the one with the highest score is the scale layout plan of the water network irrigation project in the agricultural and pastoral areas.

9. The method for determining the scale layout of a water network irrigation project in an agricultural and pastoral area according to claim 8, characterized in that: The step S42 is further as follows: Step S42a: Calculate economic benefit indicators using a dynamic analysis method, including net present value, internal rate of return, and payback period; Step S42b: Calculate social benefit indicators using the analytic hierarchy process, including: quality of life, employment opportunities, and social stability; Step S42c: Calculate the ecological benefit index using the ecological environment quality index method.

10. The method for determining the scale layout of a water network irrigation project in an agricultural and pastoral area according to claim 8, wherein: The step S43 is further as follows: Step S43a: Construct a multi-level comprehensive evaluation model, including a target layer, a criterion layer, and an indicator layer. The target layer is the comprehensive benefits of the scale layout plan of the agricultural and pastoral water network irrigation project. The criterion layer includes economic benefits, social benefits, and ecological benefits. The indicator layer is further subdivided according to the criterion layer. Step S43b: perform pairwise comparison and scoring of the importance of indicators based on the 1-9 scale method, construct a judgment matrix, calculate the eigenvector and maximum eigenroot of the judgment matrix, perform consistency test, determine the weight of each indicator and configure a multi-level comprehensive evaluation model; Step S43c: extract the index values of the water network irrigation project layout plan of each sub-region as the input of the multi-level comprehensive evaluation model, calculate the comprehensive benefit score of the water network irrigation project layout plan of each sub-region, and the one with the highest score is the scale layout plan of the water network irrigation project in the agricultural and pastoral areas.

11. The system for determining the scale and layout of irrigation projects in agricultural and pastoral areas is characterized by: include: at least one processor; as well as a memory communicatively connected to at least one of the processors; wherein, The memory stores instructions that can be executed by the processor, and the instructions are used to be executed by the processor to implement the method for determining the scale layout of water network irrigation projects in agricultural and pastoral areas as described in any one of claims 1 to 10.

Images