Distributed hydrological forecasting method, device, computer device, and medium

Abstract

The present invention provides a distributed hydrological forecasting method, device, computer equipment, and medium that simulates runoff data by comprehensively taking into account the climate data and atmospheric boundary surface data of a target area, and corrects the runoff data obtained by the simulation with human activity data, thereby improving the accuracy of the simulated runoff data. [Solution] The method includes steps of acquiring multiple types of human activity data within a target area and first climate data and atmospheric boundary surface data for the target area in a climate scene mode; inputting the first climate data and atmospheric boundary surface data into a pre-constructed hydrological model to obtain first simulated runoff data corresponding to the climate scene mode; constructing a correction model corresponding to each human activity data based on each human activity data; and correcting the first simulated runoff data by each correction model to obtain second simulated runoff data corresponding to each human activity data.

Description

[Technical field]

[0001] The present invention relates to the technical field of hydrological information processing, and in particular to a distributed hydrological forecasting method, device, computer device, and medium. [Background technology]

[0002] Runoff is one of the exogenous forces that shape landforms and is involved in geochemical processes in the Earth's crust. It also influences soil development, plant growth, and the formation of lakes and swamps. Runoff is an important condition for the supply of water for industrial and agricultural use in a region, and is a limiting factor for the scale of regional development. Therefore, measurement, calculation, and forecasting of runoff are important tasks in the construction of hydraulic facilities.

[0003] Currently, when simulating and forecasting runoff in a changing environment in the future, the impact of climate change on runoff effect is often considered alone, or combined with the impact of a single or assumed land use type change scene based on historical trajectory on runoff effect. However, because runoff change is affected by multiple factors, the accuracy of runoff simulation decreases when only climate change or land use type change is considered. Summary of the Invention [Problem to be solved by the invention]

[0004] The present invention proposes a distributed hydrological forecasting method, apparatus, computer device, and medium to improve the accuracy of runoff forecasting. [Means for solving the problem]

[0005] In a first aspect, the present invention provides a method for producing a method for treating a cancer cell comprising the steps of: acquiring multiple types of human activity data within a target region, and first climate data and atmospheric boundary surface data of the target region in a climate scene mode; inputting the first climate data and the atmospheric interface data into a pre-constructed hydrological model to obtain first simulated runoff data corresponding to a climate scene mode; constructing a correction model corresponding to each human activity data based on each human activity data; and correcting the first simulated runoff data with each correction model to obtain second simulated runoff data corresponding to each human activity data.

[0006] In the conventional technology, when simulating and forecasting runoff data, only the impact of runoff effects due to climate change and changes in a single or hypothetical land use type are considered. However, since runoff data is affected by a comprehensive range of factors including climate data, atmospheric boundary surface data, and human activities, the above method simulates runoff data by comprehensively taking into account the climate data and atmospheric boundary surface data of the target area, and corrects the simulated runoff data with human activity data, thereby improving the accuracy of runoff forecast.

[0007] In any embodiment, the atmospheric interface data includes a Leaf Area Index value for each vegetation type, and obtaining the Leaf Area Index value for each vegetation type comprises: obtaining atmospheric carbon dioxide concentration and a geographic location of a region of interest; and inputting the first climate data, the atmospheric carbon dioxide concentration, and the geographic location into a pre-constructed Leaf Area Index prediction model to obtain Leaf Area Index values ​​for each vegetation type.

[0008] In any embodiment, the step of constructing a Leaf Area Index predictive model comprises: obtaining a first historical Leaf Area Index value for each vegetation type at a first temporal resolution for each grid within the region of interest; obtaining first historical climate data and historical atmospheric carbon dioxide concentrations for a region of interest; training the initial leaf area index prediction model based on the first historical climate data, the historical atmospheric carbon dioxide concentration, the geographic location, and the first historical leaf area index values ​​for each vegetation type to obtain a leaf area index prediction model.

[0009] In any embodiment, obtaining first historical Leaf Area Index values ​​for each vegetation type at a first temporal resolution for each grid in the region of interest comprises: obtaining a historical leaf area index sum value at a second temporal resolution for each grid within the region of interest; obtaining the percentage of land use types of each vegetation type in each grid; calculating a second historical Leaf Area Index value for each vegetation type in each grid based on the percentage of each land use type and each historical Leaf Area Index sum value; and calculating each second historical Leaf Area Index value for each day using cubic spline interpolation to obtain each first historical Leaf Area Index value at the first temporal resolution for each grid.

[0010] In an optional embodiment, obtaining first historical climate data for the region of interest comprises: acquiring first historical climate data corresponding to a plurality of climate components of a region of interest; Calculating a correlation coefficient between the first historical climate data corresponding to each climate element and the first historical leaf area index value; Screening the climate elements based on each correlation coefficient to obtain a screened climate element; determining first historical climate data for the target area from the first historical climate data corresponding to the screened climate elements.

[0011] In any embodiment, the atmospheric boundary surface data further comprises land use type data, and obtaining the land use type data comprises: obtaining land use change factors of a target area in a climate scene mode; The method includes a step of inputting the land use change factors into a previously constructed future land use simulation model to obtain land use type data.

[0012] In any embodiment, the future land use simulation model includes a system dynamics simulation model and a cellular automaton module based on an adaptive inertial competition mechanism, and the step of inputting the land use change factors into the pre-constructed future land use simulation model to obtain land use type data includes: inputting the land use change factors into a system dynamics simulation model to obtain land use type land demand; and predicting land use type data based on land use change factors, land use type land demand, and a cellular automaton module based on an adaptive inertial competition mechanism.

[0013] In any embodiment, the step of predicting the land use type data based on the land use change factors, the land use type land demand, and the cellular automaton module based on the adaptive inertial competition mechanism includes: inputting the land use change factors into an artificial neural network model to predict land use suitability probability; calculating neighborhood effects, adaptive inertia coefficients, and conversion costs corresponding to each land use type of the area of ​​interest; Calculating the occurrence probability of each land use type based on the land use suitability probability, the neighborhood effect, the adaptive inertia coefficient, and the conversion cost; and predicting land use type data based on each occurrence probability and the roulette selection mechanism.

[0014] In an optional embodiment, the step of acquiring the first climate data comprises: acquiring second climate data for the area of ​​interest in a climate scene mode and second historical climate data corresponding to the second climate data; calculating a deviation between the second climate data and the second historical climate data; acquiring third climate data for the area of ​​interest in a climate scene mode; and correcting the third climate data in response to the deviation to obtain the first climate data.

[0015] In an optional embodiment, the step of acquiring second climate data comprises: acquiring fourth climate data for the area of ​​interest in a climate scene mode; and interpolating the fourth climate data to a predetermined spatial resolution grid to obtain second climate data.

[0016] In any embodiment, the hydrological model is a variable infiltration capacity model, and constructing the variable infiltration capacity model comprises: obtaining training data for the variable infiltration capacity model, the training data including second historical climate data, historical atmospheric interface data, soil topography data, and the first historical runoff data; Building an initial variable infiltration capacity model based on the second historical climate data, the historical atmospheric interface data, and the soil topography data in the training data; inputting the second historical climate data, the historical atmospheric interface data, and the soil topography data in the training data into an initial variable infiltration capacity model to obtain a third simulated runoff data; and calibrating parameters in the initial variable permeability model based on the third simulation runoff data and the first historical runoff data in the training data to obtain a variable permeability model.

[0017] In any embodiment, the step of calibrating parameters in the initial variable permeability model based on the third simulation runoff data and the first historical runoff data in the training data to obtain the variable permeability model includes: Calculating a correlation between the third simulation runoff data and the first historical runoff data in the training data; calibrating parameters in the initial variable permeability model according to the degree of correlation to obtain a calibrated variable permeability model; acquiring test data including second historical climate data, historical atmospheric interface data, soil topography data, and the first historical runoff data; Calibrating parameters in the calibrated variable permeability model based on the test data to obtain a variable permeability model.

[0018] In any embodiment, the correlation measure includes at least one of a relative error, a Nash efficiency coefficient, and a Klein-Gupta efficiency coefficient.

[0019] In an optional embodiment, each human activity data is human activity data for a different time period, and obtaining each human activity data comprises: Obtaining second history leak data within a preset time period; determining a plurality of sudden change times of the second historical spiking data; Dividing a preset time period into a plurality of periods according to each sudden change time; determining human activity data corresponding to each time period.

[0020] In an optional embodiment, the correction models are long short-term memory network models, and constructing each long short-term memory network model comprises: Dividing the second history leaked data into third history leaked data of multiple periods according to each sudden change time; Constructing each long short-term memory network model based on each third historical runoff data and first simulated runoff data corresponding to each third historical runoff data.

[0021] In any embodiment, the sudden change time of the second historical spill data is determined by any of the following methods: MK sudden change test, sliding t-test, and Pettitt test.

[0022] In a second aspect, the present invention further comprises: an acquisition module for acquiring multiple types of human activity data within a target region, first climate data of the target region in a climate scene mode, and atmospheric boundary surface data; a simulation module for inputting the first climate data and the atmospheric interface data into a pre-constructed hydrological model to obtain first simulated runoff data corresponding to a climate scene mode; A construction module for constructing a correction model corresponding to each human activity data based on each human activity data; and a correction module for correcting the first simulated runoff data with each correction model to obtain second simulated runoff data corresponding to each human activity data.

[0023] In conventional technologies, when simulating and forecasting runoff data, only the impact of runoff effects due to climate change and changes in a single or hypothetical land use type are taken into consideration. However, since runoff data is affected by a comprehensive range of factors, including climate data, atmospheric boundary surface data, and human activities, the above-mentioned device simulates runoff data by comprehensively taking into account the climate data and atmospheric boundary surface data of the target area, and corrects the simulated runoff data with human activity data, thereby improving the accuracy of runoff forecasts.

[0024] In a third aspect, the present invention further provides a computing device comprising a memory and a processor, the memory and the processor being communicatively coupled to each other, the memory having computer instructions stored therein, the processor executing the computer instructions to perform the steps of the distributed hydrological forecasting method according to the first aspect or any of the embodiments of the first aspect.

[0025] In a fourth aspect, the present invention further provides a computer readable storage medium having stored thereon a computer program which, when executed by a processor, implements the steps of the method for distributed hydrological forecasting according to the first aspect or any of the embodiments of the first aspect. [Brief description of the drawings]

[0026] In order to more clearly describe the specific embodiments of the present invention or the technical solutions in the prior art, the following will briefly introduce the drawings required for the description of the specific embodiments or the prior art. However, the drawings in the following description are only some embodiments of the present invention, and it is obvious to those skilled in the art that they can obtain other drawings based on these drawings without any creative efforts. [Figure 1] 1 is a flow chart of a method for distributed hydrological forecasting in accordance with an exemplary embodiment. [Diagram 2] FIG. 2 is a schematic diagram of obtaining Leaf Area Index values ​​for each vegetation type in an example. [Diagram 3] FIG. 1 is a schematic diagram of a distributed hydrological forecasting device according to an exemplary embodiment. [Figure 4] FIG. 2 is a schematic diagram of a hardware configuration of a computer device according to an exemplary embodiment. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0027] The technical solutions of the present invention will be described below clearly and completely with reference to the drawings, but it is clear that the described embodiments are only a part of the embodiments of the present invention, and are not all of the embodiments. Based on the embodiments of the present invention, all other embodiments that a person skilled in the art can obtain without making creative efforts belong to the protection scope of the present invention.

[0028] Furthermore, the technical features according to each embodiment of the present invention described below may be combined with each other as long as they are not inconsistent with each other.

[0029] The present invention proposes a distributed hydrological forecasting method, apparatus, computer device, and medium to improve the accuracy of simulated runoff data.

[0030] 1 is a flow chart of a distributed hydrological forecasting method according to an exemplary embodiment. As shown in FIG. 1, the distributed hydrological forecasting method includes the following steps S101 to S104.

[0031] Step S101: Obtain multiple types of human activity data in a target area, and first climate data and atmospheric boundary surface data of the target area in a climate scene mode.

[0032] In any embodiment, the human activity data may include, but is not limited to, industrial water usage, agricultural water usage, domestic water usage, construction of water conservancy projects, reforestation, and the like, and there are no particular limitations herein.

[0033] In any embodiment, the weather data may include, but is not limited to, precipitation, temperature, humidity, wind speed, etc., as specifically limited herein.

[0034] In an optional embodiment, the atmospheric interface includes important elements such as land use type and vegetation growth, is an important link between ecological elements such as soil, atmosphere and moisture, and is affected by climate change and human activities, and affects runoff data through influencing hydrological processes such as infiltration, interception, evaporation, etc. In an embodiment of the present invention, the atmospheric interface data includes Leaf Area Index values ​​for each vegetation type, land use type data.

[0035] In any embodiment, the weather data and / or atmospheric boundary surface data may be different for different weather scene modes.

[0036] In an optional embodiment, the climate scene mode is obtained from the global climate mode data of the 6th Coupled Mode Comparison Program (CMIP6 project). The CMIP6 project has taken into account complex biogeophysical and chemical processes in the experimental design process, and has significantly improved the resolution of the atmosphere-ocean mode. According to the mode participation information provided by the CMIP6 project, a total of about 50 modes are involved in the CMIP6 mode comparison program, and each mode can provide global weather prediction data from the historical period to the year 2100 or beyond. The global climate modes in the CMIP6 project include several representative scenes, as shown in Table 1.

[0037] Table 1. Representative scenes in the CMIP6 project JPEG0007662881000002.jpg118170

[0038] Step S102: The first climate data and the atmospheric interface data are input into a pre-constructed hydrological model to obtain first simulation runoff data corresponding to a climate scene mode.

[0039] In any embodiment, the hydrological model may adopt a variable infiltration capacity model (VIC). The VIC model is a large-scale hydrological model. The VIC model adopts a water balance mode or an energy balance mode to perform simulation calculations, and the main processes include evaporation calculation, soil water content calculation, runoff calculation, and confluence calculation, and the effects of frozen soil and snowmelt may also be considered.

[0040] Step S103: Based on each human activity data, a correction model corresponding to each human activity data is constructed.

[0041] In an optional embodiment, the correction model is trained using the model simulation runoff data and the historical runoff data of the corresponding time period, and in the process of constructing the correction model, since the human activity data of different time periods have different impacts on the runoff data, multiple correction models are constructed based on the human activity data of different time periods, that is, different correction models are selected when the human activity data is different.

[0042] In any embodiment, the correction model may be a Long Short-Term Memory network model (LSTM).

[0043] Step S104: Correct the first simulation runoff data according to each correction model to obtain second simulation runoff data corresponding to each human activity data.

[0044] In conventional technologies, when simulating and forecasting runoff data, only the impact of runoff effects due to climate change and changes in a single or hypothetical land use type are considered. However, runoff data is affected by a comprehensive range of factors including climate data, atmospheric boundary surface data, and human activities. Therefore, the above method simulates runoff data by comprehensively considering the climate data and atmospheric boundary surface data of the target area, and uses a long-short-term memory network model that takes human activities into account to correct the simulated runoff data, thereby improving the accuracy of runoff forecasts.

[0045] In one example, the atmospheric interface data includes a Leaf Area Index (LAI) value for each vegetation type, and the LAI value represents a change in vegetation growth, such as when plants turn green or grow earlier phenologically. The LAI value is the most important vegetation parameter for hydrological simulation and is directly affected by climate change. The related technology obtains only the total LAI value for each grid in the target area, and does not obtain the LAI value for each vegetation type in each grid. Different vegetation types have different effects on runoff data, and obtaining the LAI value for each vegetation type is advantageous for more accurate runoff forecasting. Therefore, in the embodiment of the present invention, in the above step S101, the total LAI value is separated as follows to obtain the LAI value for each vegetation type.

[0046] First, obtain the atmospheric carbon dioxide concentration (CO2) and the geographic location of the target area. Exemplarily, the geographic location includes, but is not limited to, the internal longitude, latitude, and altitude.

[0047] Then, the first climate data, the atmospheric carbon dioxide concentration, and the geographic location are inputted into a pre-constructed leaf area index prediction model to obtain a leaf area index value for each vegetation type. Illustratively, the leaf area index prediction model may be obtained by a random forest model. Predicting the leaf area index value for each vegetation type for each grid by the first climate data, the atmospheric carbon dioxide concentration, the geographic location, and the random forest model may be expressed as follows: JPEG0007662881000003.jpg13170

[0048] Here, LAI sim represents the leaf area index value of each vegetation type obtained by simulation, and F RF represents the random forest model calibrated and tested using historical period data, M represents climatic data such as monthly precipitation and temperature, CO2 represents the atmospheric CO2 content for the corresponding year, and location represents the longitude, latitude, and altitude of the pre-defined grid center point.

[0049] In one example, a Leaf Area Index prediction model is constructed in the following steps:

[0050] Step a1: obtain a first historical LAI value of each vegetation type at a first time resolution of each grid in a target region. Illustratively, the vegetation type may be cultivated land, forest land, grassland, etc., and is not particularly limited here.

[0051] Step a2: Obtain first historical climate data and historical atmospheric carbon dioxide concentrations for the target area.

[0052] Step a3: Train an initial leaf area index prediction model based on the first historical climate data, the historical atmospheric carbon dioxide concentration, the geographic location, and the first historical leaf area index value of each vegetation type to obtain a leaf area index prediction model. Illustratively, the initial leaf area index prediction model may be a random forest model. Furthermore, after constructing the leaf area index prediction model, the model may be tested and adjusted using other historical data to further improve the prediction accuracy of the leaf area index value of each vegetation type.

[0053] In the above embodiment, the first historical climate data, the historical atmospheric carbon dioxide concentration, the geographical location, and the LAI value are used to create a combined climate-CO2-geographical location leaf area index prediction model, and the relationship between the climate data, CO2, and the geographical location and the LAI value of each vegetation type is clarified. Furthermore, the leaf area index prediction model may be used to retroactively track the historical leaf area index values ​​of each vegetation type.

[0054] In an optional embodiment, in the above step a1, a first historical Leaf Area Index value of each vegetation type at a first temporal resolution of each grid in the target region is obtained as follows:

[0055] First, obtain the total historical leaf area index value at the second temporal resolution for each grid in the target region.

[0056] Next, obtain the percentage of land use types for each vegetation type in each grid. Illustratively, the land use percentages may be determined from the land use data by the spatial analysis functions of ArcGIS software.

[0057] Further, a second historical leaf area index value for each vegetation type in each grid is calculated based on the percentage of each land use type and each total historical leaf area index value.

[0058] Finally, the cubic spline interpolation is used to calculate each second historical LAI value for each day to obtain each first historical LAI value at the first time resolution of each grid. Since the time resolution of the original LAI total value (which can be obtained by GIMMSLAI3g or GLASS LAI data) is 15 days / 8 days and the VIC hydrological model requires the input of LAI values ​​for each day, the cubic spline interpolation is used to obtain the daily LAI value of each vegetation type in the historical period, i.e., the first historical LAI value.

[0059] As shown in FIG. 2, the LAI value of the (1,1) grid is L11, and the total LAI value of the (1,1) grid is calculated by the method according to the embodiment of the present invention as follows: NThe LAI value for each vegetation type (e.g., cultivated land, forest land, grassland) is calculated as follows: JPEG0007662881000004.jpg47170

[0060] where X i is the leaf area index value of vegetation type i within a preset grid, M is the total leaf area index value, n = 1, 2, 3, ... N, representing N different types of vegetation, and Fraction i is the percentage of a preset grid that the i-th vegetation type occupies, n=N+1 represents bare soil, and the sum of the percentages of the N+1 land use types within one grid is 1. In the embodiment of the present invention, Fraction i A reasonable threshold is set for, for example, Fraction i If is less than 0.05, Fraction i is set to 0. This method can avoid outliers. The percentages of other land use types must be adjusted according to the percentage of the grid to ensure that the sum of the percentages of all land use types in a given grid is 1.

[0061] This allows the creation of an N-element linear equation for each grid. A system of N-element linear equations is constructed using the N-element linear equations created for all the preset grids within a preset neighborhood range (i.e., within a certain range around the preset simulation unit, for example, a 5×5 grid) and the N-element system of linear equations is solved by the least squares method to obtain the leaf area index value of each vegetation type in each preset simulation unit.

[0062] In one example, in step a2 above, the first historical climate data of the target area is obtained as follows.

[0063] First, obtain a first historical climate data corresponding to a plurality of climate elements of a target area. Exemplarily, the climate elements may be temperature, precipitation, humidity, etc.

[0064] Next, a correlation coefficient between the first historical climate data corresponding to each climate element and the first historical leaf area index value is calculated.

[0065] Furthermore, the climate elements are screened based on each correlation coefficient to obtain the screened climate elements.

[0066] Finally, the first historical climate data of the target area is determined from the first historical climate data corresponding to the climate elements after screening. Exemplarily, the first historical climate data may be obtained by screening the climate elements whose correlation coefficient is greater than a preset value.

[0067] In one example, the atmospheric boundary surface data further includes land use type data, and in the above step S101, the land use type data is obtained as follows.

[0068] First, obtain the land use change factors of the target area in the climate scene mode. For example, considering the actual topographical and geomorphic conditions of the watershed and factors such as urban development, information such as altitude, slope, population distribution, GDP distribution, and climate change are selected as land use change factors.

[0069] Then, the land use change factors are input into a pre-constructed future land-use simulation model (FLUS: Future Land-Use Simulation) to obtain land use type data.

[0070] In the embodiment of the present invention, after the construction of the future land use simulation model is completed, the Kappa coefficient should be used to verify the simulation effect of the future land use simulation model. The land use type data of the Yangtze River Basin in 2010 and 2015 are taken as the initial data, and the temperature, precipitation, DEM, slope length, gradient, GDP, population, distance from the city center, distance from the county center, distance from the railway, and distance from the road are taken as the land use change factors. The land use type data of the Yangtze River Basin in 2020 is simulated and compared with the actual value to calculate the Kappa coefficient. The calculation formula of the Kappa coefficient is as follows: JPEG0007662881000005.jpg22160

[0071] where P0 is the fraction of correctly simulated grids and P c is the proportion of correctly simulated grids in a random situation, and 1 is the proportion of correctly simulated grids in an ideal situation. In general, when Kappa<0.4, it indicates low similarity, when 0.4≦Kappa≦0.75, it indicates that the similarity is general, and when Kappa>0.75, it indicates that there is significant agreement, which indicates that the simulation effect is good.

[0072] In one example, the FLUS model includes a systems dynamics simulation model (SD) and a cellular automaton module based on an adaptive inertial competition mechanism, and obtains land use type data in the following steps:

[0073] Step b1: Input the land use change factors into the system dynamics simulation model to obtain the land use type land demand.

[0074] In the embodiment of the present invention, the socio-economic data and land use types of each urban area in the Yangtze River Basin in 2010 and 2015 are used to fit the empirical equation of the interaction between population, socio-economic and meteorological factors to construct an SD model for the Yangtze River Basin. Based on the changes in population, socio-economic and meteorological factors in the study area in 2030 under five different climate scene modes, a simulation is carried out to predict the land demand for each land use type under five different climate scene modes in 2030.

[0075] Step b2: Predict land use type data based on land use change factors, land use type land demand, and a cellular automaton module based on adaptive inertia competition mechanism.

[0076] In an optional embodiment, the land use type data is predicted as follows.

[0077] Firstly, the land use change factors are input into an artificial neural network model to predict the land use suitability probability.

[0078] Next, the neighborhood effect, adaptive inertia coefficient, and conversion cost corresponding to each land use type in the target area are calculated.

[0079] Furthermore, the occurrence probability of each land use type is calculated based on the land use suitability probability, the neighborhood effect, the adaptive inertia coefficient, and the conversion cost.

[0080] Finally, the land use type data is predicted based on each occurrence probability and the roulette selection mechanism.

[0081] In the embodiment of the present invention, a three-layer BP neural network model is used to obtain the land use suitability probability, the number of neurons in each layer of the neural network is set according to the selected spatial variables, random sampling is performed according to the land use change factors, and then an artificial neural network algorithm (ANN: Artificial Neural Network) is trained according to the samples to obtain the cell development probability. ANN is a multi-layer feedforward neural network, and the calculation process includes the stages of training and prediction, that is, as follows: JPEG0007662881000006.jpg39170

[0082] Here, net j (p,t) represents the signal received by the jth hidden layer grid p at time t, and X i (p,t) is the i-th variable corresponding to the i-th input layer grid p and time t, and w i,j is the weight between the input layer and the hidden layer, sp(p,k,t) represents the suitability probability that the kth land use type occurs at grid p and time t, and W j,k is the weight between the hidden layer and the output layer, and is Sigmoid(net j (p,t)) is the correlation function from the hidden layer to the output layer. For the output sp(p,k,t) from the BP-ANN, the sum of the land use suitability probability of each land use type at grid p and time t is always 1, that is, JPEG0007662881000007.jpg21167

[0083] In an embodiment of the present invention, land use type data (spatial distribution of land use types) for 2030 is predicted based on land use change factors, land use type land demand for a specific future year (2030), and a cellular automaton module (CA) based on an adaptive inertial competition mechanism.

[0084] JPEG0007662881000008.jpg44170

[0085] JPEG0007662881000009.jpg13154Denotes the total number of grid units that land use type k occupies in the N × N window at the last iteration time t-1, and w k is the variable weight between different land use types, because different land use types have different neighborhood effects. k can be obtained based on expert knowledge, model testing, or the default value of the FLUS model can be directly adopted.

[0086] In the cellular automaton module based on the adaptive inertia competition mechanism, the adaptive inertia coefficient is defined according to the difference between the land demand and the allocation amount of each land use type, and the inheritance of the current land use type in each grid unit is automatically adjusted. The core idea of ​​this adaptive inertia mechanism is that when the land demand of a certain land use type contradicts with its current amount and development trend, the inertia coefficient is automatically adjusted to slow down / accelerate it, reverse the trend, and shift the development trend of the land use type to its macro demand direction. For example, if more urban construction land is needed in the future plan, but the area of ​​urban construction land has decreased in the previous iteration process, the inertia coefficient is increased in the next iteration process to maintain the area of ​​urban construction land and promote the shift from land of other land use types to urban construction land. The land use adaptive inertia coefficient is defined as follows: JPEG0007662881000010.jpg60170

[0087] JPEG0007662881000011.jpg29170

[0088] Conversion cost SC c→kis another factor influencing the dynamic change of land use type, which represents the difficulty of transforming the current land use type c to the target land use type k. The value of the transformation cost varies in the range of [0,1], and a higher value indicates a more difficult transformation. The transformation cost can be determined by the analysis of the historical land use type data of the study area and the experience of experts, and the default parameters of the FLUS model can also be adopted. The transformation cost reflects the inherent attributes of the land use type and does not consider the influencing factors such as technological progress and human activities, so the transformation cost will not change in the model.

[0089] In an embodiment of the present invention, the overall probability of occurrence of land use types for each grid can be calculated based on land use type suitability probability, neighborhood effect, adaptive inertia coefficient, and conversion cost. JPEG0007662881000012.jpg15170

[0090] JPEG0007662881000013.jpg62170

[0091] The land use type with the highest occurrence probability is the type that this grid will preferentially convert, but land use types with a low occurrence probability also have a chance to convert. To achieve this, the FLUS model uses a roulette selection mechanism to determine which land use type occupies a grid unit. The land use type that the grid converts is proportional to the occurrence probability of that land use type. After calculating the occurrence probability of the land use type of each grid for a specific grid unit p at iteration time t, a roulette is constructed based on the occurrence probability of all land use types. Each part of the roulette is represented by a land use type. The area of ​​the sector is proportional to the occurrence probability of each land use type. Then, a uniformly distributed random number ranging from 0 to 1 is generated, and the land use type corresponding to the grid unit in this iteration is assigned according to the sector corresponding to the random number. Through this roulette selection mechanism, the land use type of this grid has a higher probability of being selected as the land use type with a higher occurrence probability.

[0092] In the embodiment of the present invention, in combination with the land demand for various land use types in 2030 under different climate scene modes simulated by the SD model, the current land use situation in 2015 is taken as the base year, and the weather, elevation (DEM), slope length, gradient, gross domestic product (GDP), population, distance from the city center, distance from the county center, distance from the railway, and distance from the road are taken as land use change factors, and the spatial distribution of land use types is obtained by cellular automaton module (CA) simulation based on the adaptive inertial competition mechanism.

[0093] In one example, the first climate data is grid point data corrected by station data. In the above step S101, the first climate data is acquired by the following steps.

[0094] Step c1: Obtain second climate data of a target area in a climate scene mode and second historical climate data corresponding to the second climate data.

[0095] In an optional embodiment, the second historical climate data is obtained by weather station data. Considering that the actual measured weather station data has problems such as unit differences and some measurement shortages, it is necessary to first convert the weather station data into a standard unit (e.g., the unified standard unit of precipitation is mm, the unified standard unit of maximum temperature is °C, the unified standard unit of minimum temperature is °C, and the unified standard unit of wind speed is m / s), and exclude weather stations with measurement shortage times exceeding 10% of the total time period. In an embodiment of the present invention, the second historical climate data includes precipitation data, maximum temperature, minimum temperature, and wind speed near the ground.

[0096] In an embodiment of the present invention, the second climate data is global climate mode data in the CMIP6 project after interpolation processing.

[0097] In the embodiment of the present invention, considering that the global climate mode data in the CMIP6 project is different grid point data from the weather station data, it is necessary to match the geographical positions of the CMIP6 global climate mode data and the second historical climate data (weather station data). The specific process of acquiring the second climate data is as follows: First, the fourth climate data of the target area in the climate scene mode, i.e., the CMIP6 global climate mode data, is acquired, and then the fourth climate data is interpolated to a preset spatial resolution grid to obtain the second climate data. Illustratively, each global climate mode data (fourth climate data) with different resolutions is uniformly downscaled and interpolated to a 1 / 8×1 / 8 grid to obtain the second climate data. At this time, the second historical climate data corresponding to the second climate data is similarly interpolated to the same spatial resolution grid.

[0098] Step c2: Calculate the deviation between the second climate data and the second historical climate data.

[0099] Step c3: Obtain third climate data of the target area in a different climate scene of the CMIP6 project. Similarly, the third climate data is climate data interpolated to the same pre-set spatial resolution grid as the second climate data. That is, the third climate data, the second climate data, and the second historical climate data are in the same spatial resolution grid.

[0100] Step c4: The third climate data is corrected according to the deviation to obtain the first climate data.

[0101] In an optional embodiment, the third climate data is corrected using an equidistant cumulative distribution function (EDCDF). The empirical distribution of the climate data for the future time period is corrected according to the relationship between the second climate data and the second historical climate data. The basic principle is that for a certain percentile, the adjustment function (Δx) does not change if the difference (deviation) between the historical simulation value and the observed value is assumed to be carried over to the future period. This method may be expressed as follows: JPEG0007662881000014.jpg13170

[0102] JPEG0007662881000015.jpg70170

[0103] In the embodiment of the present invention, data from 1961 to 2003 is set as a training period, and data from 2004 to 2100 is set as a calibration period (here, data from 2004 to 2014 is set as a validation period).

[0104] In one example, the hydrological model is a variable infiltration capacity model, and the variable infiltration capacity model is constructed as follows.

[0105] First, training data for the variable infiltration capacity model is obtained, including the second historical climate data, the historical atmospheric interface data, the soil topography data, and the first historical runoff data.

[0106] Then, an initial variable infiltration capacity model is constructed based on the second historical climate data, the historical atmospheric interface data, and the soil topography data in the training data, where the second historical climate data is the climate data similarly interpolated to the preset spatial resolution grid, the historical atmospheric interface data is the percentage of each land use type statistically calculated in the preset spatial resolution grid, and the soil topography data refers to the soil type with the largest percentage within the preset spatial resolution grid.

[0107] Then, the second historical climate data, the historical atmospheric interface data, and the soil topography data in the training data are input into the initial variable infiltration capacity model to obtain the third simulated runoff data.

[0108] Finally, the parameters in the initial variable permeability model are calibrated according to the third simulation runoff data and the first historical runoff data in the training data to obtain a variable permeability model. In an embodiment of the present invention, after calibrating the parameters in the initial variable permeability model, it is also necessary to test the calibrated variable permeability model with test data to verify the availability of the model.

[0109] In an embodiment of the present invention, the parameters in the initial variable infiltration capacity model include the shape parameter of the soil water storage capacity curve (b), the percentage of the maximum velocity when the base runoff grows nonlinearly (Ds), the maximum velocity of the base runoff (Dm), the percentage of the bottom soil moisture content to the maximum soil moisture content when the nonlinear base runoff occurs (Ws), and the depths of the three soil layers (d1, d2, d3).

[0110] In an optional embodiment, the parameters in the initial variable osmotic potential model are calibrated as follows:

[0111] First, a correlation between the third simulation runoff data and the first historical runoff data in the training data is calculated, the correlation comprising at least one of a relative error, a Nash efficiency coefficient, and a Kling-Gupta efficiency coefficient.

[0112] JPEG0007662881000016.jpg43170

[0113] The Nash-Sutcliffe Efficiency coefficient (NSE) represents the degree of matching between the simulated historical runoff data and the average of the actual historical runoff data within the corresponding time period. JPEG0007662881000017.jpg22166

[0114] JPEG0007662881000018.jpg26170

[0115] The Kling-Gupta efficiency (KGE) is calculated as follows: JPEG0007662881000019.jpg98170

[0116] where γ, α, and β are the simulated runoff data χ s and observed historical runoff data χ o The linear correlation coefficient, relative variability and deviation between s and μ o are the simulated outflow data χ s and observed historical runoff data χ o and σ s and σ o is the simulated outflow data χ s and observed historical runoff data χ o Cov so is the simulated outflow data χ s and the observed historical runoff data χ o The closer the KGE value is to 1, the better the simulation effect.

[0117] Then, according to the degree of correlation, the parameters in the initial variable permeability model are calibrated to obtain a variable permeability model. Illustratively, according to the degree of correlation, parameters can be manually adjusted by empirical calibration method (i.e., the meaning of parameters and their influence on model simulation results), and can also be adjusted automatically by using automatic parameter calibration method (e.g., SCE-UA parameter automatic calibration method).

[0118] Test data is then obtained, the test data including the second historical climate data, the historical atmospheric interface data, the soil topography data, and the first historical runoff data.

[0119] Finally, based on the test data, the parameters in the calibrated variable permeability model are calibrated to obtain the variable permeability model.

[0120] In an embodiment of the present invention, the first simulation runoff data obtained by the VIC model simulation may be expressed as follows: JPEG0007662881000020.jpg10170

[0121] Here, Q sim represents the first simulation runoff data obtained by simulation, and F VIC represents the simulation process of the VIC model, P represents the daily precipitation required for the simulation of the VIC model, and T max represents the maximum daily temperature required for the VIC model simulation, and T min represents the daily minimum temperature required for the VIC model simulation, Wind represents the daily wind speed required for the VIC model simulation, Landuse represents the land use type data for the corresponding time period, LAI represents the daily LAI value required for the VIC model simulation, Veg represents other vegetation parameters other than the LAI value, such as vegetation albedo. Since such data is not easily changed and is not easy to obtain, the default values ​​of the vegetation parameter set of the VIC model are adopted, Soil represents the soil parameters for the corresponding time period, Basin represents the watershed characteristic parameters for the corresponding time period, such as the watershed elevation, area, and river network state, and Parameters represents the parameters obtained by adjusting the parameters using historical runoff data and simulated runoff data. Here, P, T max , T min ,Wind, and LAI are daily data, and land use type data is entered annually since it generally does not undergo sudden changes from day to day, while feature data such as Veg, Soil, and Basin do not change easily and can be considered as fixed data.

[0122] In one example, considering that different human activity data have different impacts on the runoff data, different runoff data are used to train the correction model, resulting in different correction models, so different correction models are selected based on the human activity data to correct the first simulation runoff data.

[0123] In the above step S101, each piece of human activity data is human activity data within a different period of time, and the step of acquiring each piece of human activity data includes step d1 and step d2.

[0124] Step d1: Obtain second history outflow data within a preset time period.

[0125] Step d2: determine multiple sudden change times of the second history outflow data. Illustratively, the sudden change times of the second history outflow data may be determined by any of the MK sudden change test method, the sliding t test method, and the Pettitt test method. Also, the division points can be artificially set to divide the history outflow data into multiple segments. For example, by setting 1970, 1980, 1990, 2000, and 2010, the outflow sequence can be divided into before 1970, 1970 to 1979, 1980 to 1989, 1990 to 1999, 2000 to 2010, and after 2010.

[0126] In any embodiment, the specific procedure of the MK mutation assay is as follows.

[0127] Let the original time series be x1, x2, …x n Let s k is the i-th sample x i x j Define a cumulative number greater than (1≦j≦i) and define a statistic. JPEG0007662881000021.jpg48170

[0128] Under the assumption that the original sequences are random and independent, s k The average E(s k ) and variance V ar(s k ) are as follows: JPEG0007662881000022.jpg58170

[0129] In the above formula, s k When we standardize it, we get the following equation: JPEG0007662881000023.jpg35170

[0130] UF k is a statistical sequence, which forms a UF curve, and can be used to determine whether there is an obvious change trend through reliability testing. This method is applied to the reverse sequence to calculate another curve UB, and the intersection of the two curves within the confidence interval is determined as the sudden change point. With a significance level of α=0.05, the critical value of the statistics UF and UB is ±1.96. UF>0 indicates that the sequence is trending upwards, and conversely, if the sequence is trending downwards, and is greater than 1.96 or less than -1.96, it indicates that the upward or downward trend is significant.

[0131] In an optional embodiment, the sliding t-test method tests for sudden change points by analyzing whether there is a significant difference between the means of two subsequences in a sequence.

[0132] JPEG0007662881000024.jpg73170

[0133] The statistic follows a t-distribution with V=n1+n2-2 degrees of freedom.

[0134] In an optional embodiment, the Pettitt test determines the sudden change point by whether there is a significant difference in the cumulative distribution function before and after the sudden change point. In the Pettitt test, the original time series is divided into x1, x2, ... x N The statistic U constructed as t,N is expressed as follows: JPEG0007662881000025.jpg51170

[0135] statistic Ut,N The maximum value of K τ Then, the corresponding sample τ is a sudden change point, and the test formula for the significance level is: JPEG0007662881000026.jpg44170

[0136] Step d3: Divide the preset time period into a plurality of periods according to each sudden change time.

[0137] Step d4: Determine human activity data corresponding to each period.

[0138] In one example, the correction models are long short-term memory network models, and each long short-term memory network model is constructed in the following steps.

[0139] First, the second history outflow data is divided into third history outflow data of a plurality of periods according to each sudden change time.

[0140] Next, each long short-term memory network model is constructed based on each third historical outflow data and the first simulated outflow data corresponding to each third historical outflow data.

[0141] The future runoff simulation scenes jointly determined by future climate data, atmospheric interface data, and different human activity intensity data are listed in Table 2. Among them, the future climate scene (first climate data) is obtained from the global climate mode data in the CMIP6 project.

[0142] Table 2: Simulation scene settings for future runoff in this example JPEG0007662881000027.jpg198170

[0143] In one example, in step S104, the plurality of second simulation outflow data are obtained as follows. JPEG0007662881000028.jpg13170

[0144] Here, Q cor (t) represents the second simulation runoff data corresponding to each human activity data at time t, and Q sim (t) represents the first simulation runoff data at time t, and Q sim (t-1) represents the first simulation outflow data at time t-1, and N is the number of inputs to the long-short memory network model. JPEG0007662881000029.jpg19170

[0145] Based on the same inventive concept, an embodiment of the present invention further provides a distributed hydrological forecasting device, which includes an acquisition module 301, a simulation module 302, a construction module 303 and a correction module 304, as shown in Figure 3.

[0146] The acquisition module 301 is used for acquiring multiple types of human activity data in the target area, and the first climate data and atmospheric boundary surface data of the target area in the climate scene mode. For details, please refer to the description of step S101 in the above embodiment, and will not be described in detail here.

[0147] The simulation module 302 is used to input the first climate data and the atmospheric interface data into a pre-constructed hydrological model to obtain the first simulated runoff data corresponding to the climate scene mode. For details, please refer to the description of step S102 in the above embodiment, and will not be described in detail here.

[0148] The construction module 303 is used to construct a correction model corresponding to each human activity data according to each human activity data, for details, please refer to the description of step S103 in the above embodiment, and will not be described in detail here.

[0149] The correction module 304 is used to correct the first simulation runoff data according to each correction model to obtain second simulation runoff data corresponding to each human activity data. For details, please refer to the description of step S104 in the above embodiment, and will not be described in detail here.

[0150] In conventional technologies, when simulating and forecasting runoff data, only the impact of runoff effects due to climate change and changes in a single or hypothetical land use type are taken into consideration. However, since runoff data is affected by a comprehensive range of factors, including climate data, atmospheric boundary surface data, and human activities, the above-mentioned device simulates runoff data by comprehensively taking into account the climate data and atmospheric boundary surface data of the target area, and corrects the simulated runoff data with human activity data, thereby improving the accuracy of runoff forecasts.

[0151] In one example, the atmospheric interface data includes Leaf Area Index values ​​for each vegetation type, and the acquisition module 301 includes a first acquisition sub-module and a first prediction sub-module.

[0152] The first acquisition submodule is used for acquiring the atmospheric carbon dioxide concentration and the geographical location of the target area, for details, please refer to the description of the above embodiment, and will not be described in detail here.

[0153] The first prediction submodule is used to input the first climate data, the atmospheric carbon dioxide concentration, and the geographical location into a pre-constructed LAI prediction model to obtain LAI values ​​for each vegetation type. For details, please refer to the description of the above embodiment, and will not be described in detail here.

[0154] In one example, the first prediction submodule includes a first acquisition unit, a second acquisition unit, and a training unit.

[0155] The first acquisition unit is used to acquire a first historical LAI value of each vegetation type at a first time resolution of each grid in the target region. For details, please refer to the description of the above embodiment, and will not be described in detail here.

[0156] The second acquisition unit is used to acquire the first historical climate data and the historical atmospheric carbon dioxide concentration of the target area. For details, please refer to the description of the above embodiment, and will not be described in detail here.

[0157] The training unit is used to train the initial LAI prediction model according to the first historical climate data, the historical atmospheric carbon dioxide concentration, the geographical location, and the first historical LAI value of each vegetation type to obtain a LAI prediction model. For details, please refer to the description of the above embodiment, and will not be described in detail here.

[0158] In one example, the first acquisition unit includes a first acquisition subunit, a first acquisition subunit, a first calculation subunit, and a second calculation subunit.

[0159] The first acquisition sub-unit is used for acquiring the total value of the historical leaf area index at the second time resolution of each grid in the target region. For details, please refer to the description of the above embodiment, and will not be described in detail here.

[0160] The second acquisition sub-unit is used for acquiring the percentage of land use type of each vegetation type in each grid, the details of which are to be referred to the above embodiment, and will not be described in detail here.

[0161] The first calculation subunit is used to calculate a second historical LAI value of each vegetation type in the grid according to the percentage of each land use type and the total value of each historical LAI. For details, please refer to the description of the above embodiment, and will not be described in detail here.

[0162] The second calculation subunit is used to calculate each second historical LAI value for each day using cubic spline interpolation to obtain each first historical LAI value at the first time resolution of each grid. For details, please refer to the description of the above embodiment, and will not be described in detail here.

[0163] In one example, the second acquisition unit includes a third acquisition subunit, a third calculation subunit, a screening subunit, and a determination subunit.

[0164] The third acquisition sub-unit is used for acquiring first historical climate data corresponding to a plurality of climate elements of the target area, the details of which can be referred to the description of the above embodiment, and will not be described in detail here.

[0165] The third calculation subunit is used for calculating the correlation coefficient between the first historical climate data corresponding to each climate element and the first historical LAI value. For details, please refer to the description of the above embodiment, and will not be described in detail here.

[0166] The screening sub-unit is used to screen the climate elements according to the correlation coefficients, and obtain the climate elements after screening. For details, please refer to the description of the above embodiment, and will not be described in detail here.

[0167] The determining subunit is used for determining the first historical climate data of the target area from the first historical climate data corresponding to the climate elements after screening. For details, please refer to the description of the above embodiment, and will not be described in detail here.

[0168] In one example, the atmospheric interface data further includes land use type data, and the acquisition module 301 further includes a second acquisition sub-module and a first simulation sub-module.

[0169] The second acquisition sub-module is used for acquiring land use change factors of the target area in the climate scene mode. For details, please refer to the description of the above embodiment, and will not be described in detail here.

[0170] The first simulation submodule is used to input the land use change factors into the pre-constructed future land use simulation model to obtain land use type data. For details, please refer to the description of the above embodiment, and will not be described in detail here.

[0171] In one example, the future land use simulation model is a system dynamics simulation model,

[0172] a cellular automaton module based on an adaptive inertial competition mechanism, and the first simulation sub-module includes a first prediction unit and a second prediction unit.

[0173] The first prediction unit is used to input the land use change factors into the system dynamics simulation model to obtain the land demand for land use types. For details, please refer to the description of the above embodiment, and will not be described in detail here.

[0174] The second prediction unit is used to predict land use type data according to land use change factors, land use type land demand, and cellular automaton module based on adaptive inertia competition mechanism.For details, please refer to the above embodiment description, so it will not be described in detail here.

[0175] In one example, the second prediction unit includes a first prediction subunit, a fourth calculation subunit, a fifth calculation subunit, and a second prediction subunit.

[0176] The first prediction subunit is used for inputting the land use change factors into the artificial neural network model to predict the land use suitability probability. For details, please refer to the above embodiment, and will not be described in detail here.

[0177] The fourth calculation subunit is used for calculating the neighborhood effect, adaptive inertia coefficient and conversion cost corresponding to each land use type of the target area. For details, please refer to the description of the above embodiment, and will not be described in detail here.

[0178] The fifth calculation subunit is used for calculating the occurrence probability of each land use type according to the land use suitability probability, the neighborhood effect, the adaptive inertia coefficient and the conversion cost. For details, please refer to the description of the above embodiment, and will not be described in detail here.

[0179] The second prediction sub-unit is used to predict the land use type data according to the occurrence probability and the roulette selection mechanism, the details of which are referred to the above embodiment description, and will not be described in detail here.

[0180] In one example, the acquisition module 301 includes a third acquisition sub-module, a calculation sub-module, a fourth acquisition sub-module, and a correction sub-module.

[0181] The third acquisition submodule is used for acquiring the second climate data of the target area in the climate scene mode and the second historical climate data corresponding to the second climate data. For details, please refer to the description of the above embodiment, and will not be described in detail here.

[0182] The calculation submodule is used for calculating the deviation between the second climate data and the second historical climate data, the details of which can be referred to the description of the above embodiment, and will not be described in detail here.

[0183] The fourth acquisition sub-module is used for acquiring the third climate data of the target area in the climate scene mode, for details, please refer to the description of the above embodiment, and will not be described in detail here.

[0184] The correction submodule is used for correcting the third climate data according to the deviation to obtain the first climate data, for details, please refer to the description of the above embodiment, and will not be described in detail here.

[0185] In one example, the third acquisition sub-module includes a third acquisition unit and a fourth acquisition unit.

[0186] The third acquisition unit is used to acquire the fourth climate data of the target area in the climate scene mode, for details, please refer to the description of the above embodiment, and will not be described in detail here.

[0187] The fourth acquisition unit is used to obtain the second climate data by interpolating the fourth climate data into a preset spatial resolution grid. For details, please refer to the description of the above embodiment, and will not be described in detail here.

[0188] In one example, the hydrological model is a variable infiltration capacity model, and the simulation module 302 includes a fifth acquisition sub-module, a first construction sub-module, a second simulation sub-module, and a calibration sub-module.

[0189] The fifth acquisition submodule is used for acquiring training data for the variable infiltration capacity model, including the second historical climate data, the historical atmospheric interface data, the soil topography data, and the first historical runoff data. For details, please refer to the description of the above embodiment, and will not be described in detail here.

[0190] The first construction submodule is used to construct an initial variable infiltration capacity model according to the second historical climate data, the historical atmospheric interface data, and the soil topography data in the training data. For details, please refer to the description of the above embodiment, and will not be described in detail here.

[0191] The second simulation submodule is used to input the second historical climate data, historical atmospheric interface data, and soil topography data in the training data into the initial variable infiltration capacity model to obtain the third simulation runoff data. For details, please refer to the description of the above embodiment, and will not be described in detail here.

[0192] The calibration submodule is used to calibrate the parameters in the initial variable permeability model according to the third simulation runoff data and the first historical runoff data in the training data to obtain a variable permeability model. For details, please refer to the description of the above embodiment, and will not be described in detail here.

[0193] In one example, the calibration submodule includes a calculation unit, a calibration unit, a fifth acquisition unit, and a verification unit.

[0194] The calculation unit is used to calculate the correlation between the third simulation leakage data and the first historical leakage data in the training data. For details, please refer to the description of the above embodiment, and will not be described in detail here.

[0195] The calibration unit is used to calibrate the parameters in the initial variable permeability model according to the correlation degree, and obtain the calibrated variable permeability model.For details, please refer to the description of the above embodiment, and will not be described in detail here.

[0196] The fifth acquisition unit is used to acquire test data including the second historical climate data, the historical atmospheric boundary surface data, the soil topography data, and the first historical runoff data. For details, please refer to the description of the above embodiment, and will not be described in detail here.

[0197] The calibration unit is used to calibrate the parameters in the calibrated variable permeability model according to the test data, and obtain the variable permeability model. For details, please refer to the description of the above embodiment, and will not be described in detail here.

[0198] In one example, the correlation measure in the calculation unit includes at least one of a relative error, a Nash efficiency coefficient, and a Klein-Gupta efficiency coefficient.

[0199] In one example, each human activity data is human activity data in a different time period, and the acquiring module 301 further includes a sixth acquiring sub-module, a first determining sub-module, a first dividing sub-module, and a second determining sub-module.

[0200] The sixth acquiring sub-module is used for acquiring the second history leakage data within a preset time period.

[0201] The first determining sub-module is used for determining a plurality of sudden change times of the second historical spill data.

[0202] The first division sub-module is used for dividing the preset time period into a plurality of periods according to each sudden change time.

[0203] The second determining sub-module is used for determining human activity data corresponding to each period.

[0204] In one example, the correction model is a long short-term memory network model, and the construction module 304 includes a second division sub-module and a second construction sub-module.

[0205] The second dividing sub-module is used for dividing the second historical outflow data into third historical outflow data of multiple periods according to each sudden change time.

[0206] The second construction sub-module is used for constructing each long short-term memory network model based on each third historical data and the first simulation data corresponding to each third historical data.

[0207] In one example, in the determination submodule, the sudden change time of the second historical runoff data is determined by one of the MK sudden change test method, the sliding t-test method, and the Pettitt test method.

[0208] The specific limitations and beneficial effects of the above device may refer to the limitations of the above distributed hydrological forecasting method, and will not be described in detail here. Each of the above modules may be realized in whole or in part by software, hardware, and a combination thereof. Each of the above modules may be integrated in the processor of the computer device in hardware form, or may be independent, and may be stored in the memory of the computer device in software form to facilitate the processor to call and execute the operations corresponding to each of the above modules.

[0209] Fig. 4 is a schematic diagram of a hardware configuration of a computer device according to an exemplary embodiment. As shown in Fig. 4, the device includes one or more processors 410 and memory 420 including persistent memory, volatile memory, and a hard disk, and one processor 410 is illustrated in Fig. 4. The device may further include an input device 430 and an output device 440.

[0210] The processor 410, memory 420, input device 430, and output device 440 may be connected via a bus or other means, and FIG. 4 illustrates the connections via a bus.

[0211] The processor 410 may be a central processing unit (CPU). The processor 410 may be other general-purpose processors, chips such as digital signal processors (DSPs), application specific integrated circuits (ASICs), field programmable gate arrays (FPGAs) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, or combinations of the above chips. The general-purpose processor may be a microprocessor, any conventional processor, or the like.

[0212] The memory 420 is a non-transitory computer-readable storage medium, including persistent memory, volatile memory, and hard disk, and can be used to store non-transitory software programs, non-transitory computer-executable programs, and modules, such as program instructions / modules corresponding to the distributed hydrological forecasting method in the embodiments of the present application. The processor 410 executes the non-transitory software programs, instructions, and modules stored in the memory 420 to execute various functional applications and data processing of the server, i.e., to realize any of the distributed hydrological forecasting methods described above.

[0213] The memory 420 may include a program storage area capable of storing an operating system, an application program required for at least one function, and a data storage area capable of storing data to be used as required. Additionally, the memory 420 may include high-speed random access memory, and may further include non-transient memory, such as at least one magnetic disk memory device, flash memory device, or other non-transient solid-state memory device. In some embodiments, the memory 420 may optionally include memory located remotely relative to the processor 410, which may be connected to the data processing device via a network. Such networks include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0214] The input device 430 can receive input numeric or textual information and generate signal inputs for user settings and function control. The output device 440 can include a display device such as a display screen.

[0215] The one or more modules, when stored in memory 420 and executed by one or more processors 410, perform a method such as that shown in FIG.

[0216] The above product can execute the method according to the embodiment of the present invention, and includes the corresponding functional modules and beneficial effects of the execution method. For technical details not described in detail in this embodiment, please refer to the relevant description in the embodiment shown in FIG.

[0217] An embodiment of the present invention further provides a non-transitory computer storage medium having computer-executable instructions stored thereon, capable of performing any of the method embodiments described above, where the storage medium may be a magnetic disk, an optical disk, a read-only memory (ROM), a random access memory (RAM), a flash memory, a hard disk drive (HDD), a solid-state drive (SSD), etc. The storage medium may also include a combination of the above types of memory.

[0218] It should be noted that, in this specification, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply that such an actual relationship or sequence exists between these entities or operations. Furthermore, the terms "comprise", "contain" or any other variants thereof are intended to cover a non-exclusive inclusion, such that a process, method, article, or device that includes a set of elements includes not only these elements, but also other elements not expressly listed, or elements inherent to such process, method, article, or device. Unless further limited, an element defined by the phrase "comprises a ..." does not exclude the presence of other identical elements in the process, method, article, or device that includes the element.

[0219] The above are only specific embodiments of the present invention, which are intended to enable those skilled in the art to understand or practice the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Thus, the present invention is not intended to be limited to these embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features of the present application.

Claims

1. 1. A method for distributed hydrological forecasting, comprising: acquiring a plurality of types of human activity data within a target region, and a first climate data and an atmospheric boundary surface data of the target region in a climate scene mode; inputting the first climate data and the atmospheric interface data into a pre-constructed hydrological model to obtain first simulated runoff data corresponding to the climate scene mode; constructing a correction model corresponding to each of the human activity data based on each of the human activity data; correcting the first simulated runoff data by each of the correction models to obtain second simulated runoff data corresponding to each of the human activity data; The atmospheric interface data includes a Leaf Area Index value for each vegetation type, and obtaining the Leaf Area Index value for each vegetation type includes: obtaining an atmospheric carbon dioxide concentration and a geographic location of the region of interest; inputting the first climate data, the atmospheric carbon dioxide concentration, and the geographic location into a pre-constructed leaf area index prediction model to obtain a leaf area index value for each vegetation type; The step of constructing the leaf area index prediction model comprises: obtaining first historical Leaf Area Index values ​​for each vegetation type at a first temporal resolution for each grid within the region of interest; obtaining first historical climate data and historical atmospheric carbon dioxide concentrations for the region of interest; training an initial Leaf Area Index forecasting model based on the first historical climate data, the historical atmospheric carbon dioxide concentration, the geographic location, and a first historical Leaf Area Index value for each vegetation type to obtain the Leaf Area Index forecasting model; obtaining first historical Leaf Area Index values ​​for each vegetation type at a first temporal resolution for each grid within the region of interest, comprising: obtaining a historical Leaf Area Index sum value at a second temporal resolution for each grid within the region of interest; obtaining a percentage of land use types for each of said vegetation types in each grid; calculating a second historical Leaf Area Index value for each of the vegetation types in each grid based on the percentage of each of the land use types and each of the historical Leaf Area Index sum values; and calculating each of the second historical LAI values ​​on a daily basis using cubic spline interpolation to obtain each of the first historical LAI values ​​at a first temporal resolution for each of the grids; calculating a second historical LAI value for the vegetation type in each grid based on the percentage of each of the land use types and each of the historical LAI sum values, and for each grid, creating a system of equations using all predefined grids within a predefined neighborhood around the grid, and obtaining the first historical Leaf Area Index value at a first temporal resolution for the vegetation type in the grid based on the percentage of land use type and the total historical Leaf Area Index value for each of the predefined grids.

2. The step of acquiring first historical climate data for the region of interest comprises: acquiring first historical climate data corresponding to a plurality of climate elements of the target area; Calculating a correlation coefficient between first historical climate data corresponding to each of the climate elements and the first historical leaf area index value; Screening the climate elements based on each of the correlation coefficients to obtain screened climate elements; and determining first historical climate data for the region of interest from first historical climate data corresponding to the screened climate elements.

3. The atmospheric boundary surface data further includes land use type data, and the step of obtaining the land use type data includes: obtaining land use change factors of the target area in the climate scene mode; The method according to claim 1, further comprising the step of inputting the land use change factors into a previously constructed future land use simulation model to obtain the land use type data.

4. The future land use simulation model includes a system dynamics simulation model and a cellular automaton module based on an adaptive inertial competition mechanism, and the step of inputting the land use change factors into a previously constructed future land use simulation model to obtain the land use type data includes: inputting the land use change factors into the system dynamics simulation model to obtain land use type land demand; and forecasting the land use type data based on the land use change factors, the land use type land demand, and a cellular automaton module based on an adaptive inertial competition mechanism.

5. The step of predicting the land use type data based on the land use change factors, the land use type land demand, and a cellular automaton module based on an adaptive inertial competition mechanism includes: inputting the land use change factors into an artificial neural network model to predict land use suitability probability; calculating neighborhood effects, adaptive inertia coefficients, and conversion costs corresponding to each land use type of the target area; Calculating an occurrence probability of each land use type based on the land use suitability probability, the neighborhood effect, the adaptive inertia coefficient, and the conversion cost; and predicting the land use type data based on each of the occurrence probabilities and a roulette selection mechanism.

6. The step of acquiring the first climate data includes: acquiring second climate data for the area of ​​interest in the climate scene mode and second historical climate data corresponding to the second climate data; calculating a deviation between the second climate data and the second historical climate data; acquiring third climate data for the region of interest in the climate scene mode; and correcting the third climatic data in response to the deviation to obtain the first climatic data.

7. The step of acquiring second climate data includes: acquiring fourth climate data for the region of interest in the climate scene mode; and interpolating the fourth climate data to a predetermined spatial resolution grid to obtain the second climate data.

8. The hydrological model is a variable infiltration capacity model, and the step of constructing the variable infiltration capacity model includes: obtaining training data for the variable infiltration capacity model, the training data including second historical climate data, historical atmospheric interface data, soil topography data, and first historical runoff data; constructing an initial variable infiltration capacity model based on the second historical climate data, the historical atmospheric interface data, and the soil topography data in the training data; inputting the second historical climate data, the historical atmospheric interface data, and the soil topography data in the training data into the initial variable infiltration capacity model to obtain a third simulated runoff data; and calibrating parameters in the initial variable permeability model based on the third simulated runoff data and first historical runoff data in the training data to obtain the variable permeability model.

9. calibrating parameters in the initial variable permeability model based on the third simulation runoff data and the first historical runoff data in the training data to obtain the variable permeability model; Calculating a correlation between the third simulation runoff data and the first historical runoff data in the training data; calibrating parameters in the initial variable permeability model according to the degree of correlation to obtain a calibrated variable permeability model; acquiring test data including second historical climate data, historical atmospheric interface data, soil topography data, and the first historical runoff data; and calibrating parameters in a calibrated variable permeability model based on the test data to obtain the variable permeability model.

10. 10. The method of claim 9, wherein the correlation measure comprises at least one of a relative error, a Nash efficiency coefficient, and a Klein-Gupta efficiency coefficient.

11. Each of the human activity data is human activity data within a different time period, and the step of acquiring each of the human activity data includes: Obtaining second history outflow data within a preset time period; determining a plurality of sudden change times of the second historical spill data; Dividing the preset time period into a plurality of periods according to each of the sudden change times; and determining human activity data corresponding to each said period of time.

12. The correction model is a long short-term memory network model, and the step of constructing each of the long short-term memory network models includes: Dividing the second history data into a plurality of periods of third history data according to each of the sudden change times; 12. The method of claim 11, further comprising constructing each of the long short-term memory network models based on each of the third historical runoff data and first simulated runoff data corresponding to each of the third historical runoff data.

13. 12. The method of claim 11, wherein the sudden change time of the second historical spill data is determined by one of an MK sudden change test method, a sliding t-test method, and a Pettitt test method.

14. 1. A distributed hydrological forecasting system comprising: an acquisition module for acquiring multiple types of human activity data within a target region, and first climate data and atmospheric boundary surface data of the target region in a climate scene mode; a simulation module for inputting the first climate data and the atmospheric interface data into a pre-constructed hydrological model to obtain first simulated runoff data corresponding to the climate scene mode; a construction module for constructing a correction model corresponding to each of the human activity data based on each of the human activity data; a correction module for correcting the first simulated runoff data by each of the correction models to obtain second simulated runoff data corresponding to each of the human activity data; The atmospheric interface data includes a Leaf Area Index value for each vegetation type, and obtaining the Leaf Area Index value for each vegetation type includes: obtaining an atmospheric carbon dioxide concentration and a geographic location of the region of interest; inputting the first climate data, the atmospheric carbon dioxide concentration, and the geographic location into a pre-constructed leaf area index prediction model to obtain a leaf area index value for each vegetation type; The step of constructing the leaf area index prediction model comprises: obtaining first historical Leaf Area Index values ​​for each vegetation type at a first temporal resolution for each grid within the region of interest; obtaining first historical climate data and historical atmospheric carbon dioxide concentrations for the region of interest; training an initial Leaf Area Index forecasting model based on the first historical climate data, the historical atmospheric carbon dioxide concentration, the geographic location, and a first historical Leaf Area Index value for each vegetation type to obtain the Leaf Area Index forecasting model; obtaining first historical Leaf Area Index values ​​for each vegetation type at a first temporal resolution for each grid within the region of interest, comprising: obtaining a historical Leaf Area Index sum value at a second temporal resolution for each grid within the region of interest; obtaining a percentage of land use types for each of said vegetation types in each grid; calculating a second historical Leaf Area Index value for each of the vegetation types in each grid based on the percentage of each of the land use types and each of the historical Leaf Area Index sum values; and calculating each of the second historical LAI values ​​on a daily basis using cubic spline interpolation to obtain each of the first historical LAI values ​​at a first temporal resolution for each of the grids; calculating a second historical LAI value for the vegetation type in each grid based on the percentage of each of the land use types and each of the historical LAI sum values, and for each grid, creating a system of equations centered on the grid using all predefined grids within the predefined neighborhood, and obtaining the first historical Leaf Area Index value at a first temporal resolution for the vegetation type in the grid based on the percentage of land use type and the total historical Leaf Area Index value for each of the predefined grids.

15. A computer device comprising: a memory; and a processor, the memory and the processor being communicatively connected to each other; computer instructions stored in the memory; and the processor executing the computer instructions to perform steps of the distributed hydrological forecasting method according to any one of claims 1 to 13.

16. A computer-readable storage medium having a computer program stored thereon, A computer-readable storage medium, comprising a computer program which, when executed by a processor, implements the steps of the method for distributed hydrological forecasting according to any one of claims 1 to 13.

Images