River and underground water coupling simulation parameter generation method and system

Abstract

The invention relates to the technical field of coupling simulation, in particular to a river and underground water coupling simulation parameter generation method and system. The method comprises the following steps: collecting remote sensing and ground sensing monitoring data, and generating a multi-source space-time initial data set through time mark calibration, space resampling and signal-to-noise ratio weighted fusion; extracting features between the water level of the river and the water level of the underground water based on the data set, constructing a river-underground water space-time topological structure and identifying an interaction mode, and forming a space-time coupling feature map through significance screening and periodic enhancement; performing parameter inversion by adopting a hydrodynamic equation, and generating a physical inversion parameter set in combination with sensitivity analysis and local optimization; a parallelization hydrological simulation framework is constructed, error-driven optimization is executed, and dynamic optimization parameters are obtained; river and underground water exchange simulation is executed, and finally, coupling simulation parameter set updating is achieved through sliding window deviation evaluation and incremental parameter correction. Therefore, the precision and operability of a coupling simulation result are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of coupling simulation, and in particular to a method and system for generating river and groundwater coupling simulation parameters. Background Art

[0002] Coupled modeling dynamically simulates the exchange of water between surface water and groundwater, providing fundamental support for optimizing water resource allocation, managing groundwater overexploitation, and assessing water environmental quality. In actual simulations, the reliability of simulation results is closely linked to the scientific nature of parameter settings. River-groundwater coupled models involve a wide range of parameters, including hydraulic parameters (such as permeability and storage coefficients), riverbed seepage parameters (such as transmissivity and thickness), and spatiotemporal variables related to boundary and initial conditions. These parameters vary significantly across regions and timescales, exhibiting significant spatial heterogeneity and nonlinear characteristics. Currently, traditional parameter generation methods primarily include field measurements, empirical estimation, historical inversion, and manual calibration. Field measurements rely on extensive geological, hydrological, and drilling data, resulting in high acquisition costs, long acquisition cycles, and uneven data distribution, making them difficult to meet the needs of large-scale model construction. Empirical estimation methods, which infer parameters based on existing literature or empirical formulas, lack specificity and are prone to error accumulation. Summary of the Invention

[0003] Based on this, it is necessary for the present invention to provide a method and system for generating river and groundwater coupling simulation parameters to solve at least one of the above technical problems.

[0004] To achieve the above-mentioned purpose, a method for generating parameters for river and groundwater coupling simulation includes the following steps:

[0005] Step S1: Acquire the monitoring data of the study area, and perform spatiotemporal fusion of multi-source data based on the monitoring data of the study area to obtain a multi-source spatiotemporal initial dataset;

[0006] Step S2: Analyze the spatiotemporal topological structure of river-groundwater based on the multi-source spatiotemporal initial data set, and enhance the interactive time-varying characteristics of the spatiotemporal topological structure of river-groundwater to obtain a spatiotemporal coupling feature map;

[0007] Step S3: performing multi-dimensional parameter inversion on the spatiotemporal coupling characteristic map to obtain an inversion parameter set; performing physical parameter joint distribution correction on the inversion parameter set to obtain a physical inversion parameter set;

[0008] Step S4: performing dynamic optimization of spatiotemporal scale characteristics according to the physical inversion parameter set, and performing parallel simulation and global optimal solution selection on the dynamic optimization results to obtain a dynamic optimization parameter set;

[0009] Step S5: constructing a spatiotemporal multiscale hydrological model based on the dynamic optimization parameter set, and using the spatiotemporal multiscale hydrological model to perform multiscale water exchange simulation to obtain a hydrological simulation result set;

[0010] Step S6: Acquire real-time monitoring data of the study area, and minimize the model parameter deviation of the spatiotemporal multi-scale hydrological model based on the numerical deviation of the real-time monitoring data of the study area and the hydrological simulation result set to obtain a coupled simulation parameter set.

[0011] In coupled simulation, the present invention dynamically simulates the water exchange between surface water and groundwater, providing solid support for water resource allocation optimization, groundwater over-exploitation management, and water environment quality assessment. First, monitoring data from the study area is obtained and multi-source data is fused in space and time, effectively integrating data from different sources and ensuring the spatiotemporal consistency of the data. This fusion method can provide a comprehensive and accurate initial dataset, laying the foundation for subsequent analysis, thereby enhancing the reliability and accuracy of the data during the simulation process. By analyzing the spatiotemporal topological structure of river-groundwater, the spatial heterogeneity and time-varying characteristics of hydrological processes can be better understood, revealing the complex interaction patterns between surface water and groundwater. The generation of spatiotemporal coupling characteristic maps provides the model with more refined spatiotemporal characteristics, enhancing the ability to identify hydrological changes and enabling the simulation to dynamically adapt to changes at different time scales and within regions. The acquisition of the inversion parameter set, combined with the joint distribution correction of physical parameters, effectively improves the accuracy of the model parameters. Through multidimensional parameter inversion, not only can more influencing factors be considered, but the synergy between parameters can also be ensured, thereby improving the predictive ability and robustness of the model. The revision of the physical inversion parameter set further ensures the physical rationality of the simulation results, making the simulation output more consistent with the actual hydrological process. The combination of dynamic optimization of spatiotemporal scale characteristics and parallel simulation significantly improves the computational efficiency and optimization accuracy of the model. By selecting the global optimal solution, the parameter set that best suits the actual situation can be obtained, ensuring the high accuracy and operability of the simulation results. At the same time, the construction of a spatiotemporal multi-scale hydrological model enables the model to handle water exchange problems at multiple time scales and better adapt to complex hydrological environments. The introduction of real-time monitoring data and the optimization based on minimizing the numerical deviation between simulation results and real-time data not only improve the real-time response capability of the model, but also optimize the parameters through dynamic adjustment, ensuring that the simulation results can be corrected in real time as the actual situation changes. This process effectively reduces the model error through the optimization method of minimizing the deviation, so that the accuracy of the model is further improved. In summary, from data acquisition to model optimization, not only has the accuracy and efficiency of the simulation process been greatly improved, but through multi-dimensional parameter inversion, spatiotemporal scale optimization, and real-time data feedback, the hydrological model has become more adaptable to actual application needs, providing a scientific and reliable tool for water resources management and environmental assessment, and reducing the degree of error accumulation.

[0012] Optionally, this specification also provides a system for generating river and groundwater coupling simulation parameters, which is used to execute the method for generating river and groundwater coupling simulation parameters as described above. The system for generating river and groundwater coupling simulation parameters includes:

[0013] The spatiotemporal fusion module is used to obtain the monitoring data of the study area and perform spatiotemporal fusion of multi-source data based on the monitoring data of the study area to obtain the multi-source spatiotemporal initial data set;

[0014] The topological structure analysis module is used to analyze the spatiotemporal topological structure of rivers and groundwater based on the multi-source spatiotemporal initial data set, and enhance the interactive time-varying characteristics of the spatiotemporal topological structure of rivers and groundwater to obtain a spatiotemporal coupling feature map;

[0015] The parameter inversion module is used to perform multi-dimensional parameter inversion on the spatiotemporal coupling characteristic map to obtain an inversion parameter set; perform physical parameter joint distribution correction on the inversion parameter set to obtain a physical inversion parameter set;

[0016] The parameter optimization module is used to dynamically optimize the spatiotemporal scale characteristics according to the physical inversion parameter set, and to perform parallel simulation and global optimal solution selection on the dynamic optimization results to obtain the dynamic optimization parameter set;

[0017] The water exchange simulation module is used to construct a spatiotemporal multiscale hydrological model based on a dynamically optimized parameter set, and use the spatiotemporal multiscale hydrological model to perform multiscale water exchange simulations to obtain a set of hydrological simulation results;

[0018] The parameter deviation optimization module is used to obtain real-time monitoring data of the study area, and minimize the model parameter deviation of the spatiotemporal multi-scale hydrological model based on the numerical deviation of the real-time monitoring data of the study area and the hydrological simulation result set to obtain the coupled simulation parameter set.

[0019] The system for generating river and groundwater coupling simulation parameters of the present invention can implement any one of the methods for generating river and groundwater coupling simulation parameters of the present invention, and is used as a medium for combining operations and signal transmission between various modules to complete the method for generating river and groundwater coupling simulation parameters. The modules within the system cooperate with each other, thereby improving the accuracy and operability of the coupling simulation results and reducing the degree of error accumulation. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Other features, objects and advantages of the present invention will become more apparent upon reading the detailed description of non-limiting embodiments thereof made with reference to the following drawings:

[0021] Figure 1 Schematic diagram of the steps of the method for generating parameters of river and groundwater coupling simulation according to the present invention;

[0022] Figure 2 Detailed step flow diagram of step S1 in the present invention;

[0023] Figure 3 Detailed step flow diagram of step S4 in the present invention;

[0024] The purpose, features and advantages of the present invention will be further described with reference to the accompanying drawings and in conjunction with the embodiments. DETAILED DESCRIPTION

[0025] The following is a clear and complete description of the technical method of the present invention in conjunction with the accompanying drawings. It is obvious that the embodiments described are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without making any creative efforts are within the scope of protection of the present invention.

[0026] In addition, the accompanying drawings are merely schematic illustrations of the present invention and are not necessarily drawn to scale. Identical reference numerals in the figures denote identical or similar parts, and thus repetitive descriptions thereof will be omitted. Some of the block diagrams shown in the accompanying drawings are functional entities that do not necessarily correspond to physically or logically separate entities. These functional entities may be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor and / or microcontroller approaches.

[0027] It should be understood that although the terms "first," "second," and the like may be used herein to describe various elements, these elements should not be limited by these terms. These terms are used solely to distinguish one element from another. For example, a first element may be referred to as a second element, and similarly, a second element may be referred to as a first element, without departing from the scope of the exemplary embodiments. The term "and / or" as used herein includes any and all combinations of one or more of the listed associated items.

[0028] To achieve this, please refer to Figures 1 to 3 The present invention provides a method for generating parameters of river and groundwater coupling simulation, the method comprising the following steps:

[0029] Step S1: Acquire the monitoring data of the study area, and perform spatiotemporal fusion of multi-source data based on the monitoring data of the study area to obtain a multi-source spatiotemporal initial dataset;

[0030] In this embodiment, multi-source monitoring data of the study area are collected, which include remote sensing images, groundwater monitoring, meteorological data and flow monitoring data of the study area. Remote sensing images can be obtained through ground sensors, satellites or drone platforms. Groundwater level data comes from monitoring wells, with a measurement frequency of once a week and a spatial resolution of the local area of each monitoring point. Meteorological data and flow observation data are collected through meteorological stations and flow metering equipment, with a temporal resolution of hours and a spatial resolution of the coverage area of the local meteorological station. These data are paired and aligned according to time and space, and spatiotemporal fusion is performed based on weighted averaging and interpolation to form a multi-source spatiotemporal initial data set.

[0031] Step S2: Analyze the spatiotemporal topological structure of river-groundwater based on the multi-source spatiotemporal initial data set, and enhance the interactive time-varying characteristics of the spatiotemporal topological structure of river-groundwater to obtain a spatiotemporal coupling feature map;

[0032] In this embodiment, when analyzing the spatiotemporal topological structure of rivers and groundwater, a hydrogeological model is first constructed in the study area, and all hydrological elements (such as river basins, groundwater levels, soil permeability, etc.) are mapped into a high-dimensional spatiotemporal topological network. Through spatial pairing and time comparison, the spatial coupling relationship between different hydrological units is identified, such as the mutual influence between groundwater levels and river flow. A dynamic network model is used to describe these spatiotemporal relationships. In this spatiotemporal topological structure, the changes in river flow and groundwater levels are enhanced by time-varying characteristics, emphasizing their interaction patterns in different time periods. For example, the delayed response relationship between river flow changes and groundwater level changes will be extracted, and these features will be strengthened through the spatiotemporal convolutional network, so that the topological structure can dynamically reflect the hydrological exchange patterns in different time periods. Ultimately, the generated spatiotemporal coupling feature map maps the river-groundwater interaction characteristics into a three-dimensional matrix, in which each element represents the hydrological relationship at a specific time and spatial position.

[0033] Step S3: performing multi-dimensional parameter inversion on the spatiotemporal coupling characteristic map to obtain an inversion parameter set; performing physical parameter joint distribution correction on the inversion parameter set to obtain a physical inversion parameter set;

[0034] In this embodiment, multidimensional parameter inversion is performed based on the spatiotemporal coupling characteristic diagram. First, the characteristic vector of each spatiotemporal point is extracted from the spatiotemporal coupling characteristic diagram, and the inversion model is used for deduction to obtain a set of preliminary parameter sets. These parameters include physical quantities such as groundwater permeability, riverbed permeability, and flow transmission coefficient. The inversion result is obtained by the least squares method, but the inversion parameters obtained at this time may have physical consistency problems. In order to solve this problem, the inversion parameter set needs to be corrected by the joint distribution of physical parameters, and the constrained optimization method is adopted to make the inversion result conform to geological and hydrological laws. The specific method is to correct the joint distribution of each physical parameter through the Gaussian mixture model to ensure that the range of variation of each parameter is consistent with the geological conditions and known experience. The corrected inversion parameter set should conform to the actual physical laws and ensure the rationality of each physical parameter. The physical inversion parameter set finally obtained can be used as the input of the subsequent hydrological model.

[0035] Step S4: performing dynamic optimization of spatiotemporal scale characteristics according to the physical inversion parameter set, and performing parallel simulation and global optimal solution selection on the dynamic optimization results to obtain a dynamic optimization parameter set;

[0036] In this embodiment, according to the physical inversion parameter set, the spatiotemporal scale characteristics in the hydrological model are first dynamically optimized. The core of this step is to optimize the spatiotemporal response characteristics of the model by refining and adjusting the time scale (e.g., daily, monthly, seasonal) and spatial scale (e.g., local, regional). The method of dynamic time scale adjustment is adopted to adaptively adjust the hydrological processes in different periods, such as increasing the prediction weight of flow in the flood season, and focusing on the simulation of groundwater changes in the drought season. Through the parallel simulation method, the simulation tasks of different time scales are distributed to multiple computing nodes to speed up the calculation process. Each computing node performs the simulation according to its regional characteristics and parameter set, and finally merges the results and performs global optimization to select the optimal solution set. The goal of global optimization is to minimize the deviation between the hydrological simulation results and the actual observation values, and use genetic algorithms or particle swarm optimization methods to find the optimal parameter set to obtain the optimal parameter set after dynamic optimization. These parameter sets will be used to build a more accurate spatiotemporal hydrological model.

[0037] Step S5: constructing a spatiotemporal multiscale hydrological model based on the dynamic optimization parameter set, and using the spatiotemporal multiscale hydrological model to perform multiscale water exchange simulation to obtain a hydrological simulation result set;

[0038] In this embodiment, a spatiotemporal multi-scale hydrological model is constructed based on a dynamic optimization parameter set. During the construction process, the study area is first divided into multiple sub-areas, each of which represents a different hydrological unit, taking into account the heterogeneity of factors such as topography, soil, and vegetation. The hydrological processes (such as precipitation, evaporation, runoff, etc.) in each sub-area are modeled through dynamic multi-scale simulation to ensure that the hydrological exchange processes at different time scales and spatial scales can be accurately simulated. The model uses water flow transfer functions at different scales to consider the spatiotemporal characteristics of water exchange such as river-groundwater and precipitation-evaporation. When performing multi-scale water exchange simulation, the model combines boundary conditions and initial conditions at different scales to obtain the water distribution in each area through numerical solution methods (such as finite difference method or finite element method). Finally, a set of simulation result sets are obtained, which contain hydrological process data at various time scales (such as hours, days, and months) and can be used for further hydrological analysis and decision support.

[0039] Step S6: Acquire real-time monitoring data of the study area, and minimize the model parameter deviation of the spatiotemporal multi-scale hydrological model based on the numerical deviation of the real-time monitoring data of the study area and the hydrological simulation result set to obtain a coupled simulation parameter set.

[0040] In this embodiment, a spatiotemporal multiscale hydrological model is optimized in real time. First, real-time monitoring data is obtained for the study area, covering key hydrological elements such as groundwater level, flow rate, and precipitation. The real-time data is compared with the previously obtained hydrological simulation results, and the numerical deviation is calculated. Based on this deviation information, the model parameters are optimized to minimize the deviation. The goal is to adjust the model parameters so that the simulation results are closer to the observed data. To this end, an optimization method based on deviation minimization, such as the least squares method or gradient descent method, is used to fine-tune each parameter in the model to ensure that the error between the model output and the actual observation value is minimized. During the optimization process, dynamic weights are used to give different attention to data from different periods. For example, during flood periods, flow data is given a higher weight, while during drought periods, groundwater level data is given a higher weight. Ultimately, a set of coupled simulation parameters is obtained. These parameters will be used to update the spatiotemporal multiscale hydrological model so that it can more accurately reflect the hydrological characteristics of the study area in the future.

[0041] Optionally, step S1 specifically includes:

[0042] Step S11: Acquire monitoring data of the study area, wherein the monitoring data of the study area includes remote sensing data and ground sensing data;

[0043] In this embodiment, remote sensing image data of the study area is obtained through remote sensing satellites. This data contains information such as land use changes, vegetation cover, and meteorological characteristics in the area. The resolution of meteorological satellite images is 1 km, the resolution of land observations is 10 meters, and the temporal resolution is once a month. Ground sensors use multiple sensor devices installed in the study area to monitor groundwater levels, river levels, temperature, humidity, precipitation, and other data in real time. The frequency of data collection is once a day. The vertical resolution of groundwater level monitoring data is 0.01 meters, and the resolution of surface water level (such as river level) monitoring is 0.01 meters; the resolution of temperature monitoring is 0.1°C, the resolution of humidity monitoring is 1%, and the resolution of precipitation monitoring is 0.1 mm. All monitoring data are spatially aligned according to a unified coordinate system. The time range of all data sources is unified to the past three years, and the spatial coverage areas of remote sensing data and ground sensing data must match each other to ensure consistency of the monitoring areas. The acquisition of remote sensing data and ground data is carried out through standardized data interfaces to ensure the high quality and consistency of the data sources.

[0044] Step S12: calibrate the time stamps of the remote sensing data and the ground sensing data, and fill in the missing time periods to obtain a time-calibrated data set;

[0045] In this embodiment, the time stamps of the acquired remote sensing data and ground sensor data are calibrated. By comparing the timestamps of the remote sensing data and the ground sensor data, it is ensured that the time stamps of the two data sources are consistent. If it is found that there is data missing from the ground sensor (such as a weather station or groundwater monitoring well fails to collect data at certain time points), the interpolation method is used to fill the missing values. The specific method is to use linear interpolation or spline interpolation to fill the data missing in continuous time periods, and for non-continuous missing time periods, the weighted average method can be used to fill. When filling the missing data, the data of the adjacent time periods are used to estimate the missing time points to ensure smooth transition and temporal consistency of the data. The calibrated data set will retain the time series data and check the time synchronization to ensure the temporal consistency of the remote sensing data and the ground sensor data.

[0046] Step S13: performing spatial coordinate system 1 according to the time calibration data set, and performing spatial resampling on the remote sensing data with abnormal resolution in the result of spatial coordinate system 1 to obtain a time-space calibration data set;

[0047] In this embodiment, during the spatial coordinate system unification process, it is first ensured that both the remote sensing data and the ground sensor data use a unified coordinate reference system (e.g., WGS 84 or UTM). If the remote sensing data and the ground data use different coordinate systems, the data needs to be spatially converted. For remote sensing data, a geographic information system (GIS) tool is first used to project and convert its spatial coordinates so that the coordinate systems of the remote sensing image and the ground data are completely consistent. Next, the spatial resolution of the remote sensing image is checked. Since different satellite images may have different resolutions (e.g., some data are 1000-meter resolution and other data are 10-meter resolution), the image data with lower resolution needs to be spatially resampled. Resampling uses the nearest neighbor interpolation method, or the more advanced bilinear interpolation method, to unify data of different resolutions to the target resolution (e.g., 10 meters). This step ensures that all data are compared and fused at the same spatial scale and ensures consistency between different data sources.

[0048] Step S14: performing data cleaning and data consistency check on the spatiotemporal calibration dataset to obtain the dataset to be fused;

[0049] In this embodiment, the spatiotemporal calibration dataset is cleaned and checked for consistency, the main purpose of which is to eliminate invalid data and ensure data quality. First, the cleaning step removes outliers, such as noise data in remote sensing images caused by cloud cover, sensor failure or external interference. For ground sensor data, it will be checked whether there are outliers outside the expected range (for example, water level measurements exceed the maximum possible value or humidity measurements are negative). For valid data, the consistency of the data will be checked, for example, to ensure that the time tags of all remote sensing images and ground sensor data can be strictly matched, and to check the spatial consistency between different data sources, that is, to ensure that the data at the same location has no spatial offset or mismatch. If spatial or temporal deviations are found, corresponding corrections and adjustments will be made. Finally, the data after cleaning and consistency checking will constitute the data set to be fused, ensuring that the remote sensing data and ground sensor data at each moment can provide reliable monitoring information.

[0050] Step S15: Perform dynamic filtering and spatiotemporal fusion according to the data set to be fused to obtain a multi-source spatiotemporal initial data set.

[0051] In this embodiment, spatiotemporal fusion is performed based on the dataset to be fused. The purpose of spatiotemporal fusion is to synthesize data from different sources (remote sensing images and ground sensor data) into the same data framework according to spatiotemporal characteristics. To ensure the validity and accuracy of the data, a dynamic filtering method is used for spatiotemporal fusion. The specific operation is to perform spatial filtering and smoothing based on the data at each time point, and use a weighted average method to integrate the influence of remote sensing data and ground data. For example, at a certain time point, if the observation quality of remote sensing data is poor, the ground sensor data will be given a higher weight. Conversely, if the ground sensor data is missing due to a fault or coverage problem, the remote sensing data will bear more weight. During filtering, different spatial weighting parameters are set according to the reliability and accuracy of different data sources to ensure the effective fusion of each data type. Ultimately, the spatiotemporal fusion data after dynamic filtering forms a multi-source spatiotemporal initial dataset. This dataset can simultaneously reflect the spatiotemporal variation characteristics of the study area and provide basic data support for subsequent analysis.

[0052] Optionally, step S15 is specifically as follows:

[0053] Step S151: Estimating the data source signal strength and noise level of the data set to be fused, and calculating the data source signal-to-noise ratio based on the signal strength and noise level;

[0054] In this embodiment, the signal strength and noise level are estimated for each data source in the fused data set. The signal strength is evaluated by measuring the signal quality of each sensor when collecting data. For remote sensing data, the signal strength can be measured by the clarity or reflectivity value of the image, while for ground sensor data, the signal strength is estimated by the fluctuation range of the real-time data value collected by the sensor. The noise level is measured by analyzing the interference or error part in the data. For example, for remote sensing data, noise manifests as the influence of clouds, shadows or atmospheric refraction, while for ground sensor data, it is the error caused by equipment failure or environmental changes. The signal strength and noise level can be quantified by statistical analysis or detection methods based on preset thresholds. Subsequently, the formula is used Calculate the signal-to-noise ratio (SNR) of each data source, where P signal is the signal power, P noise is the noise power.

[0055] Step S152: normalizing the signal-to-noise ratio of the data source to obtain normalized signal-to-noise ratio data;

[0056] In this embodiment, the signal-to-noise ratio of each data source is normalized to facilitate uniform scaling and comparison. The normalization process can be achieved by subtracting the minimum signal-to-noise ratio value of all data sources from the signal-to-noise ratio value of each data source and dividing the result by the signal-to-noise ratio range. The specific formula is: Where SNRnormalized For each data source's signal-to-noise ratio, min(SNR) and max(SNR) represent the minimum and maximum values of the SNR across all data sources, respectively. Normalized SNR values fall within the range [0, 1], ensuring that SNRs from different data sources can be compared using the same criteria. This step eliminates differences in SNRs between data sources, enabling fair comparison during subsequent weighting.

[0057] Step S153: performing signal-to-noise ratio weighting on the data set to be fused according to the normalized signal-to-noise ratio data to obtain a weighted feature data set;

[0058] In this embodiment, the normalized signal-to-noise ratio data is used to assign weights to the data sets to be fused. According to the signal-to-noise ratio value of each data source, a weight can be assigned to each data source to ensure that the data sources with higher signal quality have a larger proportion in the fusion process. Specifically, for each data source, the weight W is assigned to it. source It can be allocated according to the normalized signal-to-noise ratio value, the formula is: W source =SNR normalized When assigning weights, remote sensing data or ground data with high signal-to-noise ratios are given higher weights, while data sources with low signal-to-noise ratios are given lower weights. For example, if the signal-to-noise ratio of the remote sensing data is 0.8 and the signal-to-noise ratio of the ground sensor data is 0.6, the remote sensing data will be weighted 0.8 and the ground sensor data will be weighted 0.6. These weights will influence the contribution of each data source in the subsequent filtering and fusion processes, ensuring that high-quality data sources have a greater impact on the final data synthesis.

[0059] Step S154: constructing a spatiotemporal filtering model based on the weighted feature dataset, and dynamically fusing the dataset to be fused time-step by time-step using the spatiotemporal filtering model to obtain a filtering result for each time step;

[0060] In this embodiment, a spatiotemporal filtering model is constructed based on a weighted feature data set. The purpose of the spatiotemporal filtering model is to dynamically process data fusion at different time steps. First, when constructing the filtering model, taking into account the temporal and spatial characteristics of the data source, a filtering method with spatiotemporal weighting, such as weighted mean filtering or Gaussian filtering, is adopted. The filtering parameters in the model (such as filter window size, weighting coefficient, etc.) can be adjusted according to the spatiotemporal characteristics of the data. For example, if the signal-to-noise ratio of remote sensing data is low within a certain period of time, the deficiency of remote sensing data can be compensated by increasing the weight of ground sensor data. The filtering process of each time step is completed by weighted fusion of the spatiotemporal information of all data sources to obtain the filtering result of that time step. During the dynamic fusion process, the filtering result of each time step will reflect the information of all data sources at the current moment, and according to the weight adjustment, the influence of higher quality data sources can be highlighted.

[0061] Step S155: Perform temporal integration on the filtering results of each time step to obtain a spatiotemporal fusion data set.

[0062] In this embodiment, the filtering results of each time step are time-series integrated to form a complete spatiotemporal fusion data set. During the time series integration process, the time series integration method of weighted average is adopted to integrate the filtering results of each time point into a continuous spatiotemporal data set. This data set contains the fusion data of all moments and can reflect the spatiotemporal change characteristics of the study area throughout the entire time period. The integrated data set will have a high spatiotemporal consistency and can fully integrate information from different data sources in the spatiotemporal domain to provide accurate hydrological, meteorological and other related feature data. The key to this step is to ensure the continuity of data in the time dimension and the consistency in the spatial dimension, so as to provide reliable basic data for subsequent analysis.

[0063] Optionally, the analysis of the spatiotemporal topological structure of rivers and groundwater in step S2 is specifically as follows:

[0064] Extract remote sensing image data and groundwater level data based on spatiotemporal fusion datasets;

[0065] In this embodiment, remote sensing image data and groundwater level data are extracted from a spatiotemporal fusion data set. Remote sensing image data is usually extracted using satellite images, and image data of a certain time period is specifically selected as the basis. For example, Landsat or Sentinel satellite images are used and cropped according to a specified area and time range. During extraction, it should be ensured that the spatial resolution of the image is suitable for the analysis task, and a resolution of 10 to 30 meters is usually selected to ensure accuracy. Groundwater level data is obtained from existing ground monitoring stations or hydrological databases, and can include regularly monitored groundwater level records. The data type is generally the depth value of each monitoring point. In order to improve the spatiotemporal consistency of the data, remote sensing images and groundwater level data should be synchronized according to the same time scale, and time calibration is usually used for matching to ensure that the data of the two at the same time can be correctly associated.

[0066] Image pixel classification is performed based on remote sensing image data to extract river basin boundaries. Combined with hydrological station monitoring information, river water level and flow characteristics within the study area are obtained. Continuous river basin spatial interpolation is performed based on river basin boundaries and river water level and flow characteristics to obtain continuous river flow spatial distribution data.

[0067] In this embodiment, remote sensing image data is used to classify image pixels. A supervised classification algorithm, such as a support vector machine (SVM) or random forest (RF), is used to classify different land features based on the spectral characteristics in the image, especially to clearly distinguish water areas. A pixel classification process based on sample spectral differences is used to divide the image into land feature types and clearly mark the water body area. During the water body boundary extraction process, the normalized difference water index (NDWI) is used as an auxiliary indicator to extract the river basin boundary, and a boundary layer is formed through vectorization processing. At the same time, the data of the hydrological monitoring stations located in the target basin are retrieved to extract the water level and flow data of each station during the corresponding remote sensing image shooting period. The station spacing is controlled at 2 to 5 kilometers, and the monitoring frequency is once every 4 hours. The station data is spatially aligned with the river vector contours extracted from the remote sensing image to form a joint spatial reference system of hydrological data and river basin boundaries. After obtaining the river basin boundary and the corresponding water level and flow monitoring data, the continuous distribution of the flow data is calculated based on the spatial interpolation method. Spatial interpolation was performed using the Kriging method, with an exponential variogram model, a search radius of 10 kilometers, a minimum number of interpolation points of 5, and a maximum number of 12. The interpolation results were output in raster format, with a spatial resolution consistent with the remote sensing imagery. This generated a continuous spatial distribution layer of river flow within the study area. This layer provided key input for subsequent coupled analyses with groundwater data.

[0068] Based on the groundwater level data, the groundwater spatial distribution trend is integrated to obtain continuous groundwater spatial distribution data;

[0069] In this embodiment, by analyzing the temporal and spatial variation trends of groundwater level data, the variation patterns of groundwater at each monitoring point are obtained. Groundwater level data may have seasonal variations, so when integrating, it is necessary to smooth and trend analyze the water level data of different time periods. Trend analysis methods (such as polynomial regression or time series decomposition) are used to identify the long-term variation trends of groundwater levels. Through these trend information, a spatial distribution model of groundwater levels can be constructed, and spatial interpolation methods such as IDW (inverse distance weighting) or Kriging interpolation are usually used to generate groundwater level estimates for each location. Ultimately, the groundwater spatial distribution data obtained will present the distribution status and trend of groundwater in each region, providing a basis for water resources management and hydrological models.

[0070] The spatial correlation of continuous river flow and groundwater distribution data was analyzed to obtain the regional hydrological spatial correlation.

[0071] In this embodiment, a spatial correlation analysis is performed on the continuous river flow spatial distribution data and groundwater level data using a geostatistical analysis method. Spatial correlation analysis usually uses semi-variance analysis or spatial autocorrelation analysis, such as Moran's I or geographically weighted regression (GWR). These methods can reveal the spatial correlation between river flow and groundwater level. In the specific operation, the continuous data points are divided into several spatial units, each unit representing the average value or other statistics of the river flow and groundwater level in a certain area. The correlation between these spatial units is then calculated to obtain the spatial dependence between the hydrological variables. If the spatial correlation value is high, it means that there is a strong spatial coupling relationship between the river flow and the groundwater level.

[0072] The temporal correlation of continuous river flow spatial distribution data and continuous groundwater spatial distribution data was analyzed to obtain the regional hydrological temporal correlation;

[0073] In this embodiment, a time correlation analysis is performed on the river flow data and the groundwater level data, which is usually achieved through a time series analysis method. Methods such as the Pearson correlation coefficient or cross-correlation analysis can be used to calculate the time synchronization of the two data in different time periods. Specifically, the river flow and groundwater level data of each monitoring point are compared over time to analyze their synchronization and lag relationship over time. For example, if the river flow and groundwater level show similar fluctuation trends in certain time periods, it indicates that there is a strong time correlation between the two. In this way, the dynamic change process of rivers and groundwater and their mutual influence can be revealed, further providing a quantitative basis for hydrological analysis.

[0074] The interaction regions are divided according to the regional hydrological spatial correlation and regional hydrological temporal correlation. If both the spatial correlation and the temporal correlation are greater than 0.8, the corresponding region is divided into normal correlation regional data; if both the spatial correlation and the temporal correlation are between [0.2, 0.8], the corresponding region is divided into normal correlation regional data.

[0075] In this embodiment, based on the hydrological spatial correlation and temporal correlation data obtained in the previous step, a correlation threshold is first set. If both the spatial correlation and the temporal correlation are greater than 0.8, the area is considered to be a normal correlation area, indicating that the rivers and groundwater in the area have highly synchronized hydrological behaviors. If the spatial correlation and the temporal correlation are between 0.2 and 0.8, the area is considered to be an intermittent correlation area, indicating that there is a certain degree of uncertainty and fluctuation in the hydrological behavior in the area. According to this division standard, the entire study area is divided into normal correlation areas and intermittent correlation areas. The data in each area can be further processed, and different analysis and simulation methods can be used for different areas.

[0076] The regional water flow path topology structure analysis was carried out based on the normal-related regional data and the intermittent-related regional data, and the spatiotemporal topology structure of the river-groundwater was obtained.

[0077] In this embodiment, a water flow path topology structure analysis is performed based on the data of the normal correlation area and the intermittent correlation area. First, for the normal correlation area, a water flow path analysis method (such as the D8 algorithm or the watershed analysis method) is used to model the river and groundwater flow paths, and the main water flow channels and groundwater recharge areas are identified. Then, for the intermittent correlation area, taking into account the uncertainty of the water flow behavior in the area, a more flexible modeling method (such as a random model or a fuzzy logic model) is used to construct the water flow path. Finally, the analysis results of the normal correlation area and the intermittent correlation area are combined to obtain a complete river-groundwater spatiotemporal topological structure. This structure describes the spatial and temporal relationship between rivers and groundwater, as well as their interaction patterns in different regions, providing support for further hydrological forecasting and resource management.

[0078] Of particular importance is the analysis of the regional water flow path topology:

[0079] The spatial distribution points of the normal-related area data and the intermittent-related area data are triangulated to generate the initial water flow path candidate network;

[0080] In this embodiment, the spatial data points of the normal correlation area and the intermittent correlation area are triangulated. It is assumed that the data points of the normal correlation area and the intermittent correlation area have been obtained through the spatial correlation analysis of the previous step, and include the distribution points of rivers and groundwater in these areas. For example, in the normal correlation area, there are multiple hydrological observation points (such as river flow monitoring points, groundwater level monitoring points, etc.), and the spatial distribution of these data points can be represented by a two-dimensional coordinate system. On this basis, the Delaunay triangulation algorithm is used to connect these spatial data points to form a triangular mesh. The edge of each triangle represents a possible candidate edge of the water flow path. The candidate water flow path network will constitute a preliminary water flow path candidate map, which will be probabilistically tracked and evolved in subsequent steps. The parameters of the triangulation can include the minimum side length and maximum side length of the triangle to ensure that the generated network can effectively cover the entire area.

[0081] Based on the initial water flow path candidate network, the water flow path probability tracking is performed to obtain the dominant water flow path set;

[0082] In this embodiment, the initial water flow path candidate network is used to perform probabilistic tracking of water flow paths. First, the preliminary water flow path candidate network is obtained by triangulating the spatial data points of the normal related area and the intermittent related area. Between these data points, the connecting line segments represent the potential water flow path candidate network. The starting point and end point of each path represent the flow monitoring points at the actual geographical location, and the hydrological data of these points (such as water flow, flow velocity, etc.) have been obtained through preliminary analysis. Each candidate path will be weighted according to the relevant hydrological characteristics. First, hydrological and geographical characteristics such as the length, flow velocity, slope, and riverbed roughness of the path will be taken into consideration. For example, a shorter path or a path with a larger slope may be given a higher weight, indicating that its water flow is stronger or flows more smoothly. At the same time, based on known hydrological data such as flow velocity and flow, the weight of the path will also be affected by these actual measurements. The calculation of the weight can be expressed as: Where W i is the weight of path i, L i is the length of path i, V i is the flow velocity of path i, θ i is the slope angle of path i. The shorter the path, the greater the flow rate, the more suitable the slope for water flow, and the higher the weight. The corresponding probability of each path is calculated based on its weight. When tracing the water flow path, the starting point (such as the source of the river or the groundwater recharge area) begins to propagate along the candidate path. By traversing all candidate paths, the path with the largest weight is selected, and the evolution of the water flow path is further refined based on the relative probabilities of these paths. As the path tracing proceeds, the probability of the path is adjusted at each step according to the hydrological characteristics of the point (such as flow rate, slope, etc.). By traversing all candidate paths, the weighted probability of each path is calculated, and finally the path with the largest cumulative probability is selected as the dominant water flow path. These paths will reflect the most likely water flow paths and are the dominant channels for water flow movement. The final water flow path will include the path sequence and the spatial position relationship of the path. This information can provide data support for subsequent hydrological models and water flow simulations.

[0083] Based on regional hydrological time correlation data, the dominant flow path set is dynamically evolved. Periods with a time correlation greater than 0.8 are marked as steady-state periods, and those with a correlation between [0.2, 0.8] are marked as transient periods. The spatial offsets of the paths corresponding to the steady-state and transient periods are extracted to obtain a spatiotemporal evolution path sequence.

[0084] In this embodiment, a temporal dynamic analysis of the dominant flow path set is performed based on regional hydrological time correlation data. For each path, it is classified according to its hydrological time correlation value in different time periods: if the time correlation is greater than 0.8, it means that the flow path in this period shows a relatively stable flow pattern, which is marked as a steady-state period; if the time correlation is between 0.2 and 0.8, it means that the flow characteristics of the flow path in this period have large fluctuations, which is marked as a transient period. The steady-state period represents the stable stage of the flow path, while the transient period represents the stage where the flow rate or flow velocity fluctuates more. Next, the spatial offset of the path is extracted for each period. For example, in the steady-state period, the flow path is relatively consistent and does not deviate significantly, while in the transient period, the offset of the flow path is larger. By recording the spatial changes of the flow path in each period, a spatiotemporal evolution path sequence is formed. These path sequences will show the evolution trend of the flow path at different time scales, facilitating further analysis of hydrological dynamics.

[0085] Based on the measured flow velocity in the spatiotemporal fusion dataset, the path connection weights of the spatiotemporal evolution path sequence are corrected to construct the spatiotemporal topological structure of river-groundwater.

[0086] In this embodiment, the spatiotemporal evolution path sequence undergoes spatiotemporal correction to match the actual observed data. By correcting the measured flow velocity data in the spatiotemporal fusion dataset, the flow path sequence is ensured to be consistent with the actual flow conditions. For example, the flow velocity values obtained from remote sensing or ground sensor data can be used to adjust the path connection weights. The path connection weights reflect the intensity of water flow along a specific section of the flow path and are generally related to factors such as flow velocity, riverbed type, and flow depth. The actual measured flow velocity data is compared with the flow velocity along the simulated path, and the difference is corrected to adjust the path connection weights. In this process, the least squares method can be used to optimize the flow velocity correction to ensure that the flow velocity of the path connection matches the actual measured data. The corrected path weights will be more accurate. After the flow paths are determined, further work is to combine these paths with the spatial and temporal evolution of the groundwater system to construct a spatiotemporal topological structure. The interaction between rivers and groundwater often varies significantly in time and space, so it is necessary to combine the spatial distribution of flow paths with the distribution of the groundwater system (such as groundwater level, flow direction, etc.). The spatial location of each flow path corresponds to specific groundwater level data. Groundwater level fluctuations at the start and end points of the flow path, as well as at nodes along the path, can affect the direction and velocity of the flow. Therefore, by first overlaying the flow paths with the groundwater data, we establish a spatial correlation between each flow path and the groundwater system, forming a preliminary spatiotemporal topological structure. The temporal evolution of flow paths is a crucial component in constructing the spatiotemporal topological structure. Temporal correlation analysis reveals that flow paths shift over time. While flow paths are relatively fixed during steady-state periods, they can vary significantly during transient periods. Therefore, it is necessary to separate steady-state and transient periods and extract the spatial shifts of flow paths for each time period. For example, during steady-state periods, the weights of path connections remain constant. However, during transient periods, due to drastic changes in water flow, path connections may shift, increasing the "fluidity" of the flow path. Path shift data for each time period can be represented as a vector, reflecting the temporal changes in water flow. The path connection weights of the spatiotemporally evolving path sequence must be calibrated to match the observed data. In actual operation, remote sensing data, ground sensor data, etc. provide the basis for correction. For example, by comparing the measured flow velocity data in each time period with the simulated path, the connection weight of the path is corrected. The connection weight of the path will be dynamically adjusted according to the actual measured water velocity, flow and water level data to ensure that the spatiotemporal topological structure is consistent with the real environment. This process is usually corrected by minimizing errors (such as flow velocity differences). In the dynamic evolution and correction process of the path, a spatiotemporal topological structure is gradually formed. This structure provides a dynamic spatiotemporal coupling relationship between rivers and groundwater by reflecting the interaction between water flow paths and groundwater systems. This spatiotemporal topological structure can not only describe the spatiotemporal evolution of water flow, but also provide a basis for subsequent hydrological models, resource scheduling, ecological protection, etc.The final topological structure consists of nodes (such as river and groundwater intersections) and edges (such as water flow paths). By correcting the weights of the edges, a stable and realistic spatiotemporal network is obtained.

[0087] Optionally, the interaction time-varying feature enhancement in step S2 is specifically as follows:

[0088] Identify river-groundwater interaction patterns based on river-groundwater spatiotemporal topology;

[0089] In this embodiment, after constructing the spatiotemporal topological structure of rivers and groundwater, it is necessary to conduct an in-depth analysis of the structure to identify the interaction patterns between rivers and groundwater. First, a river-groundwater interaction model is constructed using the nodes in the spatiotemporal topological structure (such as rivers and groundwater monitoring points) and the connection relationships between them. The connection weights between nodes represent the flow intensity or interaction intensity between water flow and groundwater. These weights are closely related to flow velocity, groundwater level, and other relevant hydrological data. For example, a rise in groundwater level in a certain area may lead to an increase in river flow, and vice versa. Then, by analyzing the spatial layout and temporal changes of water flow paths, typical hydrological patterns are identified, such as the "river recharge type" pattern (groundwater recharges water resources through the river system) and the "groundwater recharge type" pattern (river water infiltrates into the groundwater system). Through these identified patterns, we can further understand the river-groundwater interaction patterns in different hydrological regions.

[0090] The significance of time-varying characteristics of river-groundwater interaction patterns was evaluated, and the significance threshold was set at 0.05 to screen out highly significant features and obtain the time-varying features to be enhanced.

[0091] In this embodiment, after identifying the river-groundwater interaction pattern, it is necessary to evaluate the significance of changes in its time-varying characteristics to ensure that the focus is on the characteristics that have an important impact on the hydrological system. First, the time-varying characteristics are evaluated for significance through time series analysis methods (such as T-test or analysis of variance) to obtain the degree of change in each time period. For example, whether the river flow or groundwater level changes significantly in different seasons or years, and whether it is affected by climate change. In the statistical analysis, the significance level is set to 0.05. When the P value is less than 0.05, it means that the change in the feature is statistically significant. At this point, it can be considered that these time-varying features have a significant impact on the river-groundwater interaction pattern and need to be further processed. By screening out features with higher significance, a set of time-varying features to be enhanced can be obtained. These features represent important dynamic changes in the hydrological system.

[0092] Perform time-varying data interpolation and time-varying periodic enhancement on the enhanced time-varying features to obtain enhanced time-varying features;

[0093] In this embodiment, the screened time-varying features to be enhanced are interpolated and periodically enhanced to improve the continuity and periodicity of the data. First, a time series interpolation method (such as spline interpolation or Lagrange interpolation) is used to fill in the missing data points in the time series to ensure the continuity of the time-varying features. For example, the groundwater level data of a certain river basin may be missing during certain periods of each month. These missing values can be predicted through interpolation to ensure the integrity of the data. Then, periodic enhancement techniques (such as Fourier transform or wavelet transform) are used to extract the periodic characteristics of the time-varying data and enhance the seasonal fluctuations of rivers and groundwater systems. This step ensures that the laws of seasonal changes, interannual changes, etc. in the hydrological data can be effectively retained to enhance the accuracy of the time-varying features.

[0094] The enhanced time-varying characteristics are used to optimize the topological parameters of the river-groundwater spatiotemporal topological structure and obtain the optimized spatiotemporal topological structure.

[0095] In this embodiment, after obtaining the enhanced time-varying features, the next step is to use these features to optimize the spatiotemporal topological structure. First, the connection strength of each node and edge in the topological structure is adjusted through the influence of the time-varying features. For example, when the groundwater level changes significantly, it may cause the river flow in certain areas to increase, so it is necessary to adjust the flow weights of these areas to enhance the river-groundwater interaction in the area. Then, the optimized topological structure is verified and adjusted through multi-objective optimization (such as minimizing the error function) to ensure that the water flow path and groundwater flow pattern after structural optimization are closer to the actual situation. For example, when optimizing the spatiotemporal topological structure, the objective function of minimizing the flow velocity difference, flow difference or groundwater level difference can be used to ensure that the optimization results are consistent with the actual monitoring data.

[0096] Based on the optimized spatiotemporal topological structure, time-varying-spatial feature coupling is performed to obtain a spatiotemporal coupling feature map.

[0097] In this embodiment, the optimized spatiotemporal topological structure is used to couple time-varying and spatial characteristics to generate a spatiotemporal coupling characteristic map. By combining the optimized topological structure with time-varying characteristics (such as groundwater level changes, flow fluctuations, etc.), a comprehensive characteristic map reflecting the spatiotemporal evolution process of the river-groundwater system is constructed. In this process, the spatiotemporal coupling characteristic map generated by combining the coupling model of spatial characteristics and time-varying characteristics can show the mutual influence and evolution trend of the river and groundwater system at different time and space scales. This characteristic map can clearly reflect the hydrological characteristics of different regions, such as the dynamic changes in water levels, seasonal fluctuations in flow, etc.

[0098] Optionally, the multidimensional parameter inversion in step S3 is specifically as follows:

[0099] Extract the physical variables of river-groundwater interaction based on the spatiotemporal coupling characteristic map, and divide the physical variables of river-groundwater interaction into multidimensional parameter spaces;

[0100] In this embodiment, the spatial and temporal data in the spatiotemporal coupling characteristic map are used to extract the physical variables of the river-groundwater interaction. For example, flow, groundwater level, soil moisture, etc. are used as key physical variables. These physical variables are important indicators of the interaction between the river and the groundwater system. Then, based on the data of these physical variables, a multidimensional parameter space is defined. Each dimension represents a key physical variable, and the value range of each variable is limited according to the actual hydrological data and environmental conditions. For example, the value range of the groundwater level is between 0 and 20 meters, and the flow is defined according to the maximum flow and minimum flow of the river. Based on these conditions, all the extracted physical variables are spatially divided to form a multidimensional parameter space, which can effectively quantify and describe the interaction between the river and the groundwater system.

[0101] Based on the spatiotemporal coupling characteristic map and combined with the preset hydrological dynamic equation constraints, the parameter inversion objective function is constructed;

[0102] In this embodiment, a target function is constructed to perform hydrological parameter inversion in combination with the spatiotemporal coupling characteristic diagram. Hydrodynamic equations (such as the continuity equation of water flow and the governing equation of groundwater flow) are used as constraints to ensure the physical consistency of the model. For example, in areas where rivers and groundwater interact, changes in flow not only depend on time, but are also closely related to factors such as groundwater level and precipitation. When constructing the parameter inversion target function, the hydrodynamic equation is combined with the spatiotemporal coupling characteristic diagram. Specifically, the hydrodynamic equation is used to describe the water flow transmission and exchange process in the river-groundwater system. The hydrological equations used generally include: continuity equation, flow equation and material conservation equation. The variables extracted from the spatiotemporal coupling characteristic diagram (such as river flow, groundwater level, etc.) are used as inputs to the hydrodynamic equation to construct a comprehensive inversion target function. Assume that the parameter inversion target function is F(θ), which has the following form: where Y obs,i is the measured data of the i-th observation point (such as river flow or groundwater level); Y sim,i (θ) is the river flow or groundwater level simulated according to the hydrodynamic equation; is the observation error variance of the i-th data point; D couple,j (θ) is the physical variable at the jth time point in the spatiotemporal coupling characteristic map (e.g., the interaction between groundwater and river flow); is the measured data at the jth time point.

[0103] The parameter inversion objective function is used to perform global parameter sampling in the multidimensional parameter space to obtain the parameter posterior probability distribution;

[0104] In this embodiment, the global parameter sampling in the parameter inversion process adopts the Monte Carlo sampling method, which aims to explore possible solutions in the parameter space through random sampling. For hydrological models, the parameter space is very large, so sampling is an effective way to obtain the posterior probability distribution of the parameters. First, the sampling interval is set according to the theoretical range of each physical variable. For example, the range of river flow is [0,1000]m 3 / s, the range of groundwater level is [0,20] meters, and the range of soil moisture is [0,100]%. These ranges are determined based on the actual hydrological data and the preset conditions of the physical model. Define the number of samples and the sampling method: Set the number of sampling times to 500 times to ensure that sufficient sampling covers the entire parameter space. In each sampling, all parameter values are randomly sampled from their respective uniform distribution ranges. Specifically, the Latin Hypercube Sampling (LHS) method is adopted to ensure that the samples are evenly distributed in the parameter space. In Latin Hypercube sampling, the value range of each parameter is divided into N equal parts, and sampling ensures that all intervals of each dimension are fully explored by randomly selecting a value from each interval. For example, if the number of sampling times for river flow is 500, its value range is [0,1000]m 3 The flow rate per second is divided into 500 intervals, each with an independent sampling point to ensure uniform sampling. For each parameter combination obtained through sampling, the model calculates the corresponding simulation results and compares them with the measured data to calculate the objective function value F(θ). For example, if a parameter combination does not fit the measured flow data well during certain time periods, the objective function value will be large, and vice versa. After all 500 sampling runs, the objective function value for each parameter combination is obtained. These objective function values are used to construct a posterior probability distribution for each parameter, representing the likelihood of each parameter under the current data. By analyzing these posterior distributions, key parameters with significant influence on the hydrological model can be identified. For example, if the posterior distribution of groundwater level is concentrated between 5 and 10 meters across all sampling runs, this indicates that groundwater level fluctuations in the current area are small and the parameter uncertainty is low. On the other hand, the posterior distribution of flow may show greater fluctuations, reflecting greater uncertainty in the flow rate's impact on the system. Ultimately, a posterior distribution plot of each parameter is generated. These distributions can be used to observe the degree of parameter uncertainty and further determine which parameters have the greatest impact on the model. For example, if the posterior distribution of flow is relatively concentrated, it means that the flow has a greater certainty for the hydrological model, and further optimization of the flow can be given priority.

[0105] Perform parameter sensitivity analysis on the posterior probability distribution of parameters, set the parameters with the top 10% sensitivity index as the dominant parameters, and obtain the dimensionality reduction dominant parameter set;

[0106] In this embodiment, based on the posterior probability distribution of the parameters, a sensitivity analysis method (such as Sobol index analysis) is used to evaluate the contribution of each parameter to the objective function. Parameters with higher sensitivity indexes have a greater impact on the simulation results and are therefore selected as dominant parameters. The parameters ranked in the top 10% of the sensitivity index are set as the dominant parameter set after dimensionality reduction. These parameters have an important influence on the water flow and groundwater interaction of the system. Through this process, the dimension of the parameter space is reduced, making subsequent optimization more focused and efficient. For example, changes in flow, groundwater level, and soil moisture may be identified as dominant parameters, while other less important factors can be excluded, thereby simplifying the calculation and improving accuracy.

[0107] The number of iterations is set to 50, and the dimensionality reduction dominant parameter set is iteratively weighted locally optimized until the optimization target error is less than the set error threshold of 5%, and the preliminary inversion parameter set is obtained;

[0108] In this embodiment, the dominant parameter set after dimensionality reduction is locally optimized. The maximum number of iterations is set to 50 times, and the optimization goal is to make the model output error less than 5%. In each iteration, a weighted adjustment is performed on each dominant parameter to optimize its performance in the objective function. For example, when the error of flow is large, the weight of flow is increased and its value is adjusted. Through local optimization, the error is gradually reduced to ensure that the simulation results can more accurately reflect the dynamic behavior of the river-groundwater system. If the error of the optimization process reaches a preset threshold within 50 iterations, the optimization will stop and the preliminary inversion parameter set will be output.

[0109] The preliminary inversion parameter set is verified by hydrophysical constraints to obtain the inversion parameter set.

[0110] In this embodiment, the preliminary inversion parameter set is subjected to physical constraint verification to ensure that it conforms to hydrological laws. In this process, it is verified whether each parameter meets the physical constraints, such as whether the groundwater level is within a reasonable range and whether the flow rate conforms to the expected seasonal fluctuations. The performance of the model is evaluated by comparison with actual observation data. For example, if the flow rate value of the preliminary inversion is too high, the flow-related parameters need to be readjusted until the error is within the physical constraint range. After verification, a reliable inversion parameter set is finally obtained, which can be used for subsequent hydrological simulation and management decision-making.

[0111] Optionally, step S4 is specifically:

[0112] Step S41: performing time-space scale decomposition based on the physical inversion parameter set to obtain a time dynamic parameter subset and a space dynamic parameter subset;

[0113] In this embodiment, after the physical inversion parameter set is obtained, the parameter set is first decomposed into time and space scales. The goal of this step is to decompose the parameter set into a temporal dynamic parameter subset and a spatial dynamic parameter subset. Specifically, the temporal dynamic parameter subset includes all parameters related to temporal variation, such as river flow, groundwater level, and other time-varying characteristics; the spatial dynamic parameter subset includes parameters related to spatial variation, such as groundwater permeability and soil moisture in different geographical regions. For example, variables extracted from the spatiotemporal coupling feature map (such as flow and water level) can be decomposed according to their temporal variation characteristics. Assuming a time step of Δt, the parameter values at each spatiotemporal point are decomposed into multiple subsets according to time, each subset representing the dynamic behavior of a specific time period. The spatial dynamic parameter subset is divided according to geographical space, extracting and classifying parameters such as watersheds and groundwater levels within different regions. Through this decomposition, the changing trends of different parameters on the spatiotemporal scale can be better identified, providing refined temporal and spatial parameter subsets for subsequent simulations.

[0114] Step S42: using the spatiotemporal coupling characteristic map as a boundary condition, assigning independent simulation conditions to the temporal dynamic parameter subset and the spatial dynamic parameter subset, and integrating the independent simulation conditions in the form of a framework to obtain a parallelized hydrological simulation framework;

[0115] In this embodiment, after obtaining the spatiotemporal dynamic parameter subsets, independent simulation conditions are assigned to each parameter subset based on the spatiotemporal coupling characteristic map as the boundary conditions. For example, for the temporal dynamic parameter subsets, each subset corresponds to a specific time step, reflecting different time periods or seasonal changes. The spatial dynamic parameter subsets are assigned corresponding geographic information based on the division of geographical areas to ensure that each subset is consistent with the characteristics of the spatial area in which it is located. These independent simulation conditions are integrated to form a parallelized hydrological simulation framework. For example, assuming that the spatiotemporal coupling characteristic map is spatially divided into several areas, the hydrological processes in each area are simulated independently, but are interconnected through boundary conditions. These simulation conditions include: 1) time step: such as daily, weekly or monthly time resolution; 2) spatial scale: such as small watershed, large watershed or specific area.

[0116] Step S43: executing parallel simulation based on the parallelized hydrological simulation framework to generate hydrological simulation results, and performing error comparison between the hydrological simulation results and the multi-source spatiotemporal initial dataset to generate an error matrix;

[0117] In this embodiment, after the parallelized hydrological simulation framework is completed, parallel simulation is performed to generate hydrological simulation results. Each subset is simulated independently using its corresponding spatiotemporal dynamic parameters, and the hydrological process is deduced, such as changes in river flow, dynamics of groundwater levels, etc. In this way, the calculation speed can be greatly accelerated and multiple simulation results can be obtained. When performing the simulation, it is assumed that there are multiple regions (such as different hydrological regions), and the simulation process in each region may be subject to different boundary conditions, such as: 1) time conditions: such as daily or seasonal changes; 2) spatial conditions: such as groundwater recharge rate, flow rate, etc. These parallel simulation results will be compared with the multi-source spatiotemporal initial data set for error. It is assumed that the error matrix is the error value calculated by comparing the simulation results of each spatiotemporal point with the observed data (such as measured river flow, groundwater level, etc.). The error matrix can help evaluate the accuracy and reliability of each simulation process.

[0118] Step S44: Evaluate parameter performance according to the error matrix and select the global optimal solution to obtain a candidate parameter combination set;

[0119] In this embodiment, the performance of each parameter combination is evaluated based on the obtained error matrix. By calculating the error between each parameter combination and the actual observed data, the parameter set with the best performance is further screened out. For example, assuming that the error matrix shows that the simulation accuracy of certain parameter combinations in a specific area is higher than that of other combinations, then these combinations will be marked as candidate optimal solutions. During the evaluation process, an error threshold (such as 5%) is set to screen the parameter combinations. Only when the error of the simulation result is less than the set threshold, the parameter combination is considered to be a candidate optimal solution. Each element in the error matrix represents the deviation between the simulation result and the measured value. The total error of each combination is calculated, and the global optimal solution is obtained by weighted average.

[0120] Step S45: Perform hydrological process consistency verification on the candidate parameter combination set to obtain a dynamically optimized parameter set.

[0121] In this embodiment, after the candidate parameter combinations are screened out, the consistency verification of the hydrological process is performed. The purpose of this verification is to confirm whether the candidate parameter combination can simulate a reasonable hydrological process. By comparing the simulation results with the actual hydrological process (such as flow fluctuations, groundwater changes, etc.), check whether each candidate parameter combination conforms to the known hydrological laws and physical constraints. For example, for the water flow model of a specific area, its consistency is judged by comparing the simulated river flow with the trend of historical observation data. If certain candidate parameter combinations can maintain a consistent hydrological process and show stability in multiple time steps and spatial regions, then these combinations will be selected as the final dynamic optimization parameter set.

[0122] Optionally, step S5 is specifically:

[0123] Step S51: Using the spatiotemporal coupling characteristic map as the initial field and the dynamically optimized parameter set as the basic parameter set, a spatiotemporal multi-scale hydrological model is constructed;

[0124] In this example, the resulting spatiotemporal coupling characteristic maps are used as the initial fields. These characteristic maps contain key information such as river flow and groundwater levels, providing preliminary geographic spatial and temporal information for the construction of the hydrological model. Based on these initial conditions and combined with the dynamically optimized parameter set (i.e., the optimized parameter set obtained in the previous step), a spatiotemporal multiscale hydrological model is constructed. Based on hydrological and geological principles, appropriate physical process equations are defined and selected to simulate hydrological phenomena, including precipitation-evaporation, precipitation-runoff, and groundwater flow. Each process behaves differently at different spatial and temporal scales. For example, for small watersheds, local hydrological processes such as runoff and evaporation within the watershed need to be considered; at the scale of large watersheds, longer-term water resource management and scheduling issues may be more of a concern. Based on the dynamically optimized parameter set obtained in the previous step (such as permeability, evaporation, soil moisture, etc.), these parameters reflect the physical properties under different geographical and climatic conditions. These parameters are dynamically adjusted within the model to ensure their adaptability to different spatiotemporal scales. For example, groundwater recharge rates may differ between mountainous and plain areas, while evaporation rates also differ significantly between humid and arid regions. The spatiotemporal hydrological model is discretized into a computable numerical form. Appropriate boundary conditions (such as watershed boundaries, river water levels, and groundwater recharge) and initial conditions (such as initial water levels and initial flow rates) are set within the model. These boundary and initial conditions directly influence the accuracy of the model results. This model can dynamically simulate the interaction between rivers and groundwater at different time scales (such as daily, monthly, and annual) and spatial scales (such as small and large watersheds). For example, in the model, spatiotemporal coupling characteristic maps provide detailed spatial information for different watershed regions, such as river flow velocity and slope, while dynamically optimized parameter sets provide the model with precise physical parameters such as permeability and evaporation. Combining these conditions through spatial interpolation and temporal integration creates a hydrological model framework that adapts to multi-scale changes.

[0125] Step S52: performing a multi-time-scale water exchange simulation based on the spatiotemporal multi-scale hydrological model to obtain a multi-scale simulation result set;

[0126] In this embodiment, a multi-time-scale water exchange simulation is performed using the constructed spatiotemporal multi-scale hydrological model. This simulation aims to reveal the changes in hydrological processes at different time scales, such as daily, seasonal, and annual water exchange patterns. During the simulation process, the model takes into account the mutual influence between river flow and groundwater, and calculates the changes in water flow at each time step. For example, in the simulation, it is assumed that the simulation at the daily time scale will focus on the daily flow changes of the river and the daily recharge of groundwater. In the seasonal simulation, the model takes into account the impact of changes in seasonal precipitation on water exchange and simulates the contribution of each season to the hydrological process. Finally, through annual simulation, the overall trend of long-term water exchange is evaluated. Each type of simulation result will be compiled into a multi-scale simulation result set to provide data support for subsequent analysis.

[0127] Step S53: using the water level time series data in the river-groundwater spatiotemporal topological structure and the hydrodynamic equation constraints as the simulation verification benchmark, performing simulation verification on the multi-scale simulation result set to obtain simulation verification results;

[0128] In this embodiment, the simulation results are verified by the water level time series data in the river-groundwater spatiotemporal topological structure. The water level time series data represents the actual water level data of the river and groundwater at different time nodes, which is used to compare the predicted values in the multi-scale simulation result set. Furthermore, the hydrodynamic equations (for example, water flow dynamics, groundwater flow equations, etc.) serve as physical constraints to guide the simulation process and ensure that the simulation results conform to the actual hydrological laws. For example, in the simulation verification, it is assumed that the water level time series data provides the measured water level information at a specific time point (such as the water level change in a certain river section), while the hydrodynamic equations will limit the physical laws of water flow in the simulation process, such as the conservation of flow, groundwater recharge and leakage, etc. By comparing the deviation between the simulation results and the measured data, the accuracy of the simulation can be evaluated and appropriate adjustments can be made.

[0129] Step S54: performing a global sensitivity analysis based on the multi-scale simulation result set, quantifying the contribution of the parameters to the river-groundwater exchange rate, screening parameters with a contribution of ≥ 5%, and obtaining sensitivity screening results;

[0130] In this embodiment, a global sensitivity analysis is performed on each parameter in the model based on the obtained multi-scale simulation result set. The purpose of the sensitivity analysis is to evaluate the role of each parameter in the hydrological model and quantify their contribution to the river-groundwater exchange rate. Through this analysis, it is possible to identify which parameters have the greatest impact on the water exchange process. For example, assume that there are multiple key parameters in the model that affect water flow exchange, such as soil permeability, groundwater recharge rate, precipitation, etc. The sensitivity analysis will quantify the contribution of each parameter by comparing the simulation results of different parameter combinations. For those parameters with a contribution greater than or equal to 5%, these parameters will be considered the most important and will be included in the subsequent model optimization. Through this screening, we can focus on the parameters that have a greater impact on the hydrological process and improve the accuracy of the simulation results.

[0131] Step S55: Using the simulation verification results as the weight basis and combining them with the sensitivity screening results, a multi-criteria dynamic weighted integration is performed to obtain a hydrological simulation result set.

[0132] In this embodiment, the sensitivity parameters and simulation verification results screened out in the previous step are combined to obtain the final set of hydrological simulation results through a multi-criteria dynamic weighted integration method. First, the simulation verification results (for example, error matrix, comparison results of measured and simulated data) are used as weight benchmarks to evaluate the credibility of different simulation results. Then, in combination with the sensitivity screening results, appropriate weights are assigned to each parameter. Finally, the weighted parameters and simulation results are integrated to obtain the final set of hydrological simulation results. For example, in the weighted integration process, it is assumed that the simulation verification results show that the deviation between the simulation results and the measured data in certain time periods is small, while certain sensitivity parameters (such as groundwater permeability) have a greater impact on the hydrological process. In this case, groundwater permeability will be given a higher weight to ensure its importance in the final simulation results. Through this weighted integration, more accurate hydrological simulation results can be obtained, further optimizing the predictive ability of the hydrological model.

[0133] It is particularly important that step S55 is specifically as follows:

[0134] Extract the error index from the simulation verification results, perform initial weight distribution in inverse proportion to the error value, and obtain the initial weight benchmark;

[0135] In this embodiment, after each simulation run in a multi-scale hydrological model, a set of simulation results is output. These results typically include hydrological variables such as flow, groundwater level, and soil moisture. To assess the accuracy of the model, the error between the simulation results and the measured data needs to be calculated. Error is typically quantified using mean squared error (MSE), relative error (RE), or other error metrics (such as normalized root mean square error (NRMSE)). For example, for flow simulation error, the difference between the simulated flow and the measured flow can be calculated. The inverse proportional distribution of errors uses the error value as a weighting factor. Smaller errors are given higher weights, while larger errors are given lower weights. This process helps ensure that model predictions are more reliable in areas or time periods with smaller errors. For example, if the simulation error for a certain time period is small (e.g., MSE < 0.1), that time period is assigned a weight of 0.9; whereas, periods with larger errors (MSE > 0.5) are assigned a lower weight, such as 0.3. Ultimately, the error values for all simulation time periods are combined to obtain a set of initial weight benchmarks.

[0136] Based on the initial weight benchmark and combined with the sensitivity screening results, the corresponding time scale weights of highly sensitive parameters are strengthened to obtain dynamically adjusted weights;

[0137] In this embodiment, weight adjustment is performed based on the initial weight benchmark obtained in the previous step, combined with the highly sensitive parameters screened out from the global sensitivity analysis. In the hydrological model, highly sensitive parameters refer to those parameters that have a significant impact on the model output, such as precipitation, evaporation, soil permeability, etc. In order to enhance the stability and reliability of the model, appropriate weight increases are made for the relevant time scales of these highly sensitive parameters. For example, since precipitation is more sensitive to seasonal changes, the weight of this parameter needs to be increased on the time scales of winter and spring. Assuming that the sensitivity screening results show that precipitation has a greater impact on the water volume in the basin, the weight of the relevant time period of precipitation in the preliminary weight can be increased from 0.6 to 0.8, thereby strengthening the importance of precipitation in the simulation. This dynamic adjustment enables the model to more accurately capture the factors that have an important impact on the hydrological process and improve the reliability of the simulation results.

[0138] Construct a multi-criteria dynamic weight matrix based on the dynamic adjustment weights and initial weight benchmarks;

[0139] In this embodiment, a multi-criteria dynamic weight matrix is constructed based on the dynamic adjustment weights and initial weight benchmarks of the aforementioned steps. Each column of the matrix corresponds to the weight distribution of a hydrological process, such as flow, groundwater level, evaporation, etc., and each row corresponds to a time scale or spatial scale. For example, for the monthly time scale, the weight of flow is 0.5, the groundwater level is 0.3, and the soil moisture is 0.2; on the quarterly scale, the weight of flow may rise to 0.6, the groundwater level remains at 0.3, and the soil moisture may be reduced to 0.1. The construction of the multi-criteria weight matrix can be carried out by linear weighting or weighted averaging. The core function of this matrix is to flexibly allocate weights through different criteria (such as spatial scale, time scale, parameter type, etc.), so as to ensure that when the model is running, appropriate weighting can be performed according to the weight of each factor to ensure the accuracy and stability of the simulation results.

[0140] The multi-criteria dynamic weight matrix is used to perform weighted integration on the multi-scale simulation result set to obtain the hydrological simulation result set.

[0141] In this embodiment, the multi-scale simulation result set is weighted and integrated based on the multi-criteria dynamic weight matrix to generate the final hydrological simulation result set. Assume that in a certain simulation process, parameters such as flow, groundwater level and soil moisture are simulated on multiple time scales, and each time scale has a different weight assignment. For example, assume that on the monthly and quarterly time scales, the weights of flow, groundwater level and soil moisture are 0.5, 0.3, 0.2 (monthly scale) and 0.6, 0.3, 0.1 (quarterly scale) respectively. By multiplying the simulation results of these two time scales by the corresponding weights and then adding up the results, a weighted hydrological simulation result can be obtained. This process can be achieved by matrix multiplication, that is, multiplying the dynamic weight matrix by the simulation result matrix of multiple time scales to finally obtain a weighted hydrological simulation result set.

[0142] Optionally, step S6 specifically includes:

[0143] Step S61: acquiring real-time monitoring data of the study area, and performing spatiotemporal benchmark alignment on the real-time monitoring data of the study area to obtain a spatiotemporal calibrated real-time data set;

[0144] In the present embodiment, real-time monitoring data, including variables such as precipitation, flow, and groundwater level, are obtained from the study area. These data are usually from different sensors or monitoring sites, so it is necessary to align the spatiotemporal benchmarks, i.e., adjust the data to the same time and space benchmarks. In order to achieve this goal, interpolation methods can be used to resample data of different time and space resolutions so that they match the simulation period and spatial grid. For example, assuming that the real-time flow monitoring data has a time resolution of 30 minutes, and the simulation model uses hours as the time unit, it is necessary to align the flow data in time series, merge or interpolate the 30-minute data, and match the time step of the hourly data. Spatially, if the monitoring data come from multiple sites and have different spatial coordinates, these data need to be standardized to the unified grid used by the model by spatial interpolation (such as inverse distance weighted method). The final spatiotemporal calibration real-time data set will provide basic data for subsequent simulation verification.

[0145] Step S62: Taking the simulation period of the multi-time-scale water exchange simulation as the time window benchmark, a sliding window deviation calculation is performed on the spatiotemporal calibration real-time dataset and the hydrological simulation result set to obtain a multi-scale numerical deviation matrix;

[0146] In the present embodiment, the deviation between real-time data and model simulation results is calculated by the sliding window method. First, a time window (such as one week, one month, etc.) is set, and the window length is consistent with the simulation period of the multi-time scale water exchange simulation. The sliding window will cover multiple time periods in the simulation results and real-time data, gradually moving and calculating the deviation in each window. For each window, the error index (such as absolute error, mean square error, relative error, etc.) between the simulation results and the real-time data can be calculated. Assume that the simulation period is one month, and the time step of the real-time monitoring data and the simulation results is day. If each sliding window is one month, the error of 30 days in the window is calculated each time, and the sliding window moves backward in sequence, and calculates the error of each window. In this way, the multi-scale numerical deviation matrix obtained will contain the error value of each time window and can be used for subsequent optimization model parameters. This matrix not only reflects the difference between the simulation results and the real-time data, but also embodies the model error on different time scales.

[0147] Step S63: constructing a deviation minimization objective function based on the weight coefficients in the multi-criteria dynamic weighted integration and the multi-scale numerical deviation matrix;

[0148] In this embodiment, the multi-scale numerical deviation matrix obtained in the previous step is combined with the weight coefficients in the multi-criteria dynamic weighted integration to construct a deviation minimization objective function. Specifically, the weight coefficients are obtained through the multi-criteria weighted integration process and are used for error correction of different hydrological processes and time scales. The deviation matrix contains the error information within each time window. When constructing the deviation minimization objective function, the weighted average error can be used as the target, and the goal is to minimize the sum of weighted deviations within the entire simulation cycle. Assume that for each time window, the error of the simulation result can be expressed as E i , where i is the index of the time window and the weight coefficient is w i , then the objective function can be expressed as: Where N is the number of windows, E i is the error of the i-th window, w i is the corresponding weight coefficient. By minimizing this objective function, the model parameters can be optimized to make the simulation results as close to the real-time data as possible.

[0149] Step S64: using the deviation minimization objective function and the hydrological simulation result set, performing incremental gradient descent on the model parameters of the spatiotemporal multi-scale hydrological model to obtain an incremental adjustment parameter set;

[0150] In this embodiment, the parameters of the spatiotemporal multiscale hydrological model are optimized using an incremental gradient descent method. The model parameters are optimized based on the hydrological simulation result set using the aforementioned deviation minimization objective function. Specifically, the gradient of the objective function relative to each parameter is first calculated, which represents the contribution of each parameter to the total error. Then, using the gradient descent method, the model parameters are gradually updated, reducing the value of the objective function until the set convergence conditions are met. For example, assume that the parameters that need to be adjusted include precipitation coefficient, evaporation coefficient, and soil permeability coefficient. For each parameter, its gradient is calculated based on the deviation minimization result of the objective function, and then the parameter value is adjusted in the opposite direction of the gradient. The step size of each iteration is determined by the learning rate, which is set to 0.01. Through repeated iterations, a set of incrementally adjusted parameters is eventually obtained. These adjusted parameter sets will minimize the error between the model's simulation results and the actual data.

[0151] Step S65: re-verify the hydrophysical constraints of the incremental adjustment parameter set to obtain a coupled simulation parameter set.

[0152] In this embodiment, after obtaining the incrementally adjusted parameter set, these parameters need to be re-verified under hydrological physical constraints to ensure that they meet the actual limitations of physical and hydrological processes. Hydrological physical constraints include some actual boundary conditions and physical limitations, such as the groundwater level cannot be lower than a certain threshold, the river flow cannot exceed the maximum discharge capacity of the basin, etc. When verifying the incrementally adjusted parameter set, the physical model can be used to check whether each parameter is within a reasonable range. For example, assuming that a certain adjusted groundwater level coefficient causes the groundwater level to be lower than the known minimum water level limit, the parameter needs to be readjusted to meet the physical constraints. In addition, combined with hydrophysical theory, hydrological processes such as river flow and evaporation can be further verified to ensure that all parameters are within a physically reasonable range. Through this verification process, the coupled simulation parameter set is finally obtained to ensure that the hydrological model can accurately reflect the actual hydrological process.

[0153] Optionally, this specification also provides a system for generating river and groundwater coupling simulation parameters, which is used to execute the method for generating river and groundwater coupling simulation parameters as described above. The system for generating river and groundwater coupling simulation parameters includes:

[0154] The spatiotemporal fusion module is used to obtain the monitoring data of the study area and perform spatiotemporal fusion of multi-source data based on the monitoring data of the study area to obtain the multi-source spatiotemporal initial data set;

[0155] The topological structure analysis module is used to analyze the spatiotemporal topological structure of rivers and groundwater based on the multi-source spatiotemporal initial data set, and enhance the interactive time-varying characteristics of the spatiotemporal topological structure of rivers and groundwater to obtain a spatiotemporal coupling feature map;

[0156] The parameter inversion module is used to perform multi-dimensional parameter inversion on the spatiotemporal coupling characteristic map to obtain an inversion parameter set; perform physical parameter joint distribution correction on the inversion parameter set to obtain a physical inversion parameter set;

[0157] The parameter optimization module is used to dynamically optimize the spatiotemporal scale characteristics according to the physical inversion parameter set, and to perform parallel simulation and global optimal solution selection on the dynamic optimization results to obtain the dynamic optimization parameter set;

[0158] The water exchange simulation module is used to construct a spatiotemporal multiscale hydrological model based on a dynamically optimized parameter set, and use the spatiotemporal multiscale hydrological model to perform multiscale water exchange simulations to obtain a set of hydrological simulation results;

[0159] The parameter deviation optimization module is used to obtain real-time monitoring data of the study area, and minimize the model parameter deviation of the spatiotemporal multi-scale hydrological model based on the numerical deviation of the real-time monitoring data of the study area and the hydrological simulation result set to obtain the coupled simulation parameter set.

[0160] The present invention is therefore intended to be illustrative and non-restrictive in all respects, with the scope of the invention being defined by the appended claims rather than the foregoing description, and all changes that come within the meaning and range of equivalents of the application documents are intended to be embraced therein.

[0161] The foregoing description is intended only to provide specific embodiments of the present invention, which will enable those skilled in the art to understand and implement the present invention. Various modifications to these embodiments will be readily 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. Therefore, the present invention is not intended to be limited to the embodiments shown herein, but is to be construed in the widest possible manner consistent with the principles and novel features disclosed herein.

Claims

1. A method for generating parameters for river and groundwater coupling simulation, characterized in that: The following steps are involved: Step S1: Acquire the monitoring data of the study area, and perform spatiotemporal fusion of multi-source data based on the monitoring data of the study area to obtain a multi-source spatiotemporal initial dataset; Step S2: Analyze the spatiotemporal topological structure of river-groundwater based on the multi-source spatiotemporal initial data set, and enhance the interactive time-varying characteristics of the spatiotemporal topological structure of river-groundwater to obtain a spatiotemporal coupling feature map; Step S3: performing multi-dimensional parameter inversion on the spatiotemporal coupling characteristic map to obtain an inversion parameter set; performing physical parameter joint distribution correction on the inversion parameter set to obtain a physical inversion parameter set; Step S4: performing dynamic optimization of spatiotemporal scale characteristics according to the physical inversion parameter set, and performing parallel simulation and global optimal solution selection on the dynamic optimization results to obtain a dynamic optimization parameter set; Step S5: constructing a spatiotemporal multiscale hydrological model based on the dynamic optimization parameter set, and using the spatiotemporal multiscale hydrological model to perform multiscale water exchange simulation to obtain a hydrological simulation result set; Step S6: Acquire real-time monitoring data of the study area, and optimize the model parameter deviation of the spatiotemporal multi-scale hydrological model by minimizing the model parameter deviation according to the numerical deviation of the real-time monitoring data of the study area and the hydrological simulation result set to obtain a coupled simulation parameter set.

2. The method for generating river and groundwater coupling simulation parameters according to claim 1, characterized in that: Step S1 is specifically as follows: Step S11: Acquire monitoring data of the study area, wherein the monitoring data of the study area includes remote sensing data and ground sensing data; Step S12: calibrate the time stamps of the remote sensing data and the ground sensing data, and fill in the missing time periods to obtain a time-calibrated data set; Step S13: performing spatial coordinate system 1 according to the time calibration data set, and performing spatial resampling on the remote sensing data with abnormal resolution in the result of spatial coordinate system 1 to obtain a time-space calibration data set; Step S14: performing data cleaning and data consistency check on the spatiotemporal calibration dataset to obtain the dataset to be fused; Step S15: Perform dynamic filtering and spatiotemporal fusion according to the data set to be fused to obtain a multi-source spatiotemporal initial data set.

3. The method for generating river and groundwater coupling simulation parameters according to claim 2, characterized in that: Step S15 is specifically as follows: Step S151: Estimating the data source signal strength and noise level of the data set to be fused, and calculating the data source signal-to-noise ratio based on the signal strength and noise level; Step S152: normalizing the signal-to-noise ratio of the data source to obtain normalized signal-to-noise ratio data; Step S153: performing signal-to-noise ratio weighting on the data set to be fused according to the normalized signal-to-noise ratio data to obtain a weighted feature data set; Step S154: constructing a spatiotemporal filtering model based on the weighted feature dataset, and dynamically fusing the dataset to be fused time-step by time-step using the spatiotemporal filtering model to obtain a filtering result for each time step; Step S155: Perform temporal integration on the filtering results of each time step to obtain a spatiotemporal fusion data set.

4. The method for generating river and groundwater coupling simulation parameters according to claim 1, characterized in that: The analysis of river-groundwater spatiotemporal topological structure in step S2 is as follows: Extract remote sensing image data and groundwater level data based on spatiotemporal fusion datasets; Image pixel classification is performed based on remote sensing image data to extract river basin boundaries. Combined with hydrological station monitoring information, river water level and flow characteristics within the study area are obtained. Continuous river basin spatial interpolation is performed based on river basin boundaries and river water level and flow characteristics to obtain continuous river flow spatial distribution data. Based on the groundwater level data, the groundwater spatial distribution trend is integrated to obtain continuous groundwater spatial distribution data; The spatial correlation analysis of continuous river flow spatial distribution data and continuous groundwater spatial distribution data was performed to obtain the regional hydrological spatial correlation; The temporal correlation of continuous river flow spatial distribution data and continuous groundwater spatial distribution data was analyzed to obtain the regional hydrological temporal correlation; The interaction regions are divided according to the regional hydrological spatial correlation and regional hydrological temporal correlation. If both the spatial correlation and the temporal correlation are greater than 0.8, the corresponding region is divided into normal correlation regional data; if both the spatial correlation and the temporal correlation are between [0.2, 0.8], the corresponding region is divided into normal correlation regional data. The regional water flow path topology structure analysis was carried out based on the normal-related regional data and the intermittent-related regional data, and the spatiotemporal topology structure of the river-groundwater was obtained.

5. The method for generating river and groundwater coupling simulation parameters according to claim 1, characterized in that: The interaction time-varying feature enhancement in step S2 is specifically as follows: Identify river-groundwater interaction patterns based on river-groundwater spatiotemporal topology; The significance of time-varying characteristics of river-groundwater interaction patterns was evaluated, and the significance threshold was set at 0.05 to screen out highly significant features and obtain the time-varying features to be enhanced. Perform time-varying data interpolation and time-varying periodic enhancement on the enhanced time-varying features to obtain enhanced time-varying features; The enhanced time-varying characteristics are used to optimize the topological parameters of the river-groundwater spatiotemporal topological structure and obtain the optimized spatiotemporal topological structure. Based on the optimized spatiotemporal topological structure, time-varying-spatial feature coupling is performed to obtain a spatiotemporal coupling feature map.

6. The method for generating river and groundwater coupling simulation parameters according to claim 1, characterized in that: The multi-dimensional parameter inversion in step S3 is specifically as follows: Extract the physical variables of river-groundwater interaction based on the spatiotemporal coupling characteristic map, and divide the physical variables of river-groundwater interaction into multidimensional parameter spaces; Based on the spatiotemporal coupling characteristic map and combined with the preset hydrodynamic equation constraints, the parameter inversion objective function is constructed; The parameter inversion objective function is used to perform global parameter sampling in the multidimensional parameter space to obtain the parameter posterior probability distribution; Perform parameter sensitivity analysis on the posterior probability distribution of parameters, set the parameters with the top 10% sensitivity index as the dominant parameters, and obtain the dimensionality reduction dominant parameter set; The number of iterations is set to 50, and the dimensionality reduction dominant parameter set is iteratively weighted locally optimized until the optimization target error is less than the set error threshold of 5%, and the preliminary inversion parameter set is obtained; The preliminary inversion parameter set is verified by hydrophysical constraints to obtain the inversion parameter set.

7. The method for generating river and groundwater coupling simulation parameters according to claim 1, characterized in that: Step S4 is specifically as follows: Step S41: performing time-space scale decomposition based on the physical inversion parameter set to obtain a time dynamic parameter subset and a space dynamic parameter subset; Step S42: using the spatiotemporal coupling characteristic map as a boundary condition, assigning independent simulation conditions to the temporal dynamic parameter subset and the spatial dynamic parameter subset, and integrating the independent simulation conditions in the form of a framework to obtain a parallelized hydrological simulation framework; Step S43: executing parallel simulation based on the parallelized hydrological simulation framework to generate hydrological simulation results, and performing error comparison between the hydrological simulation results and the multi-source spatiotemporal initial dataset to generate an error matrix; Step S44: Evaluate parameter performance according to the error matrix and select the global optimal solution to obtain a candidate parameter combination set; Step S45: Perform hydrological process consistency verification on the candidate parameter combination set to obtain a dynamically optimized parameter set.

8. The method for generating river and groundwater coupling simulation parameters according to claim 1, characterized in that: Step S5 is specifically as follows: Step S51: Using the spatiotemporal coupling characteristic map as the initial field and the dynamically optimized parameter set as the basic parameter set, a spatiotemporal multi-scale hydrological model is constructed; Step S52: performing a multi-time-scale water exchange simulation based on the spatiotemporal multi-scale hydrological model to obtain a multi-scale simulation result set; Step S53: using the water level time series data in the river-groundwater spatiotemporal topological structure and the hydrodynamic equation constraints as the simulation verification benchmark, performing simulation verification on the multi-scale simulation result set to obtain simulation verification results; Step S54: performing a global sensitivity analysis based on the multi-scale simulation result set, quantifying the contribution of the parameters to the river-groundwater exchange rate, screening parameters with a contribution of ≥ 5%, and obtaining sensitivity screening results; Step S55: Using the simulation verification results as the weight basis and combining them with the sensitivity screening results, a multi-criteria dynamic weighted integration is performed to obtain a hydrological simulation result set.

9. The method for generating river and groundwater coupling simulation parameters according to claim 1, characterized in that: Step S6 is specifically as follows: Step S61: acquiring real-time monitoring data of the study area, and performing spatiotemporal benchmark alignment on the real-time monitoring data of the study area to obtain a spatiotemporal calibrated real-time data set; Step S62: Taking the simulation period of the multi-time-scale water exchange simulation as the time window benchmark, a sliding window deviation calculation is performed on the spatiotemporal calibration real-time dataset and the hydrological simulation result set to obtain a multi-scale numerical deviation matrix; Step S63: constructing a deviation minimization objective function based on the weight coefficients in the multi-criteria dynamic weighted integration and the multi-scale numerical deviation matrix; Step S64: using the deviation minimization objective function and the hydrological simulation result set, performing incremental gradient descent on the model parameters of the spatiotemporal multi-scale hydrological model to obtain an incremental adjustment parameter set; Step S65: re-verify the hydrophysical constraints of the incremental adjustment parameter set to obtain a coupled simulation parameter set.

10. A system for generating river and groundwater coupling simulation parameters, characterized in that: The method for generating river and groundwater coupling simulation parameters according to claim 1 is used to generate river and groundwater coupling simulation parameters, and the system comprises: The spatiotemporal fusion module is used to obtain the monitoring data of the study area and perform spatiotemporal fusion of multi-source data based on the monitoring data of the study area to obtain the multi-source spatiotemporal initial data set; The topological structure analysis module is used to analyze the spatiotemporal topological structure of rivers and groundwater based on the multi-source spatiotemporal initial data set, and enhance the interactive time-varying characteristics of the spatiotemporal topological structure of rivers and groundwater to obtain a spatiotemporal coupling feature map; The parameter inversion module is used to perform multi-dimensional parameter inversion on the spatiotemporal coupling characteristic map to obtain an inversion parameter set; perform physical parameter joint distribution correction on the inversion parameter set to obtain a physical inversion parameter set; The parameter optimization module is used to dynamically optimize the spatiotemporal scale characteristics according to the physical inversion parameter set, and to perform parallel simulation and global optimal solution selection on the dynamic optimization results to obtain the dynamic optimization parameter set; The water exchange simulation module is used to construct a spatiotemporal multiscale hydrological model based on a dynamically optimized parameter set, and use the spatiotemporal multiscale hydrological model to perform multiscale water exchange simulations to obtain a set of hydrological simulation results; The parameter deviation optimization module is used to obtain real-time monitoring data of the study area, and minimize the model parameter deviation of the spatiotemporal multi-scale hydrological model based on the numerical deviation of the real-time monitoring data of the study area and the hydrological simulation result set to obtain the coupled simulation parameter set.

Images