Basin runoff simulation method and device, electronic equipment and medium

By coupling high-resolution satellite remote sensing imagery with a distributed hydrological model and employing various supervised classification algorithms and accuracy evaluation techniques, the problem of insufficient accuracy in runoff simulation under complex underlying surface conditions by the distributed hydrological model was solved. This enabled high spatiotemporal resolution runoff process simulation, improving the reliability and applicability of the simulation results.

CN121837693APending Publication Date: 2026-04-10SHANGHAI INVESTIGATION DESIGN & RES INST CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202511646717.8
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2025-11-11
Publication Date
2026-04-10

AI Technical Summary

Technical Problem

Existing distributed hydrological models face the problem of uncertainty and dynamic changes in underlying surface parameters when simulating runoff under complex underlying surface conditions, resulting in insufficient simulation accuracy and difficulty in meeting the needs of refined hydrological forecasting.

Method used

By acquiring high-resolution satellite remote sensing images, land cover classification is performed using various supervised classification algorithms (maximum likelihood method, neural network method, and support vector machine method). Combined with confusion matrix accuracy assessment, the underlying surface classification results that meet the preset criteria are selected and coupled with a distributed hydrological model to construct a coupled model for runoff simulation.

Benefits of technology

It enables a refined characterization of the spatial heterogeneity of the underlying surface of the watershed, improves the spatiotemporal accuracy of runoff simulation, enhances the model's ability to respond to different hydrological events, and improves the reliability and applicability of simulation results.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN121837693A_ABST
    Figure CN121837693A_ABST
Patent Text Reader

Abstract

The invention provides a drainage basin runoff simulation method and device, electronic equipment and a medium. The method comprises the following steps: acquiring a satellite remote sensing image of a target drainage basin; based on a plurality of supervised classification algorithms, land cover classification is performed on the satellite remote sensing image, and a plurality of underlying surface classification results are correspondingly obtained; based on the confusion matrix, performing precision evaluation on each underlying surface classification result, and taking the underlying surface classification result meeting a preset precision standard as a candidate underlying surface classification result; coupling the candidate underlying surface classification result with the distributed hydrological model to obtain a coupled distributed hydrological model; and based on the coupled distributed hydrological model, simulating a runoff process of the target drainage basin to obtain simulated runoff data. By coupling the high-resolution satellite remote sensing image and the distributed hydrological model, fine characterization of the spatial heterogeneity of the underlying surface of the drainage basin is realized, high-temporal-spatial-resolution runoff process simulation is realized on the basis, and the temporal-spatial precision of runoff simulation is effectively improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This application belongs to the field of hydrology and geographic information science and technology, and relates to a watershed runoff simulation method, device, electronic equipment and medium. Background Technology

[0002] Runoff is a crucial link in the terrestrial water cycle, and its formation and confluence are directly influenced by a variety of factors, including precipitation input, topographic features, soil properties, vegetation cover, and human activities. Accurate simulation and prediction of runoff at the watershed outlet section are of great significance for optimal water resource allocation, flood control and disaster reduction decision support, water environment quality assessment, and water infrastructure planning and design.

[0003] In recent years, distributed hydrological models have become an important tool for simulating watershed hydrological processes due to their ability to fully consider the spatial heterogeneity of underlying surface elements. However, existing distributed hydrological models still face many challenges in practical applications, among which the uncertainty of underlying surface parameters is one of the main bottlenecks affecting simulation accuracy. The underlying surface data used in traditional modeling processes, such as land use / cover, vegetation indices, and impervious surface ratios, are mostly derived from historical survey maps or low-resolution global datasets. These data generally suffer from low spatial resolution, long update cycles, inconsistent classification systems, and difficulty in reflecting the drastic changes in the underlying surface caused by rapid urbanization. Furthermore, the assumptions about static underlying surface parameters cannot reflect dynamic changes such as seasonal vegetation growth, agricultural crop rotation, and construction land expansion, leading to increased simulation biases in the model's response to different hydrological events and limiting its effectiveness in refined hydrological forecasting. Summary of the Invention

[0004] This application provides a watershed runoff simulation method, apparatus, electronic equipment, and medium to enhance the runoff simulation capability of distributed hydrological models under complex underlying surface conditions.

[0005] Firstly, this application provides a watershed runoff simulation method, comprising:

[0006] Acquire satellite remote sensing images of the target watershed;

[0007] Based on multiple supervised classification algorithms, land cover classification is performed on the satellite remote sensing images, resulting in multiple underlying surface classification results; the supervised classification algorithms include maximum likelihood method, neural network method and support vector machine method.

[0008] Based on the confusion matrix, the accuracy of each underlying surface classification result is evaluated, and the underlying surface classification result that meets the preset accuracy standard is taken as the candidate underlying surface classification result.

[0009] The candidate underlying surface classification results are coupled with the distributed hydrological model to obtain the coupled distributed hydrological model.

[0010] Based on the coupled distributed hydrological model, the runoff process of the target watershed is simulated to obtain simulated runoff data.

[0011] One implementation of the first aspect also includes:

[0012] Using a preset global land cover data as a reference dataset, several regions of interest are selected for training the supervised classification algorithm;

[0013] Each region of interest is divided into multiple typical land cover categories; the typical land cover categories include water bodies, bare land, cultivated land, grassland, woodland, urban land and snow cover, wherein the water bodies include rivers and lakes;

[0014] The separability between the regions of interest was verified using Jeffries-Matusita distance and transformation separation degree as quantitative indicators.

[0015] The regions of interest that have passed the separability verification are used as training samples and input into the maximum likelihood method, the neural network method, and the support vector machine method for supervised classification, respectively.

[0016] One implementation of the first aspect also includes:

[0017] Based on primary and / or secondary analysis methods, the classification results of each underlying surface are post-processed to eliminate noise patches generated during the land cover classification process.

[0018] In one implementation of the first aspect, the steps for constructing the confusion matrix include:

[0019] Based on the land cover sample data that is spatiotemporally matched with the satellite remote sensing image, the true land cover categories of the target watershed are determined;

[0020] The actual land cover categories are compared with the identified land cover categories corresponding to the underlying surface classification results using pixel-level comparison to form the corresponding confusion matrix; wherein:

[0021] The confusion matrix is ​​a C×C square matrix, where C represents the total number of the real feature categories or the total number of the identified feature categories, and the real feature categories are the same as the total number of the identified feature categories;

[0022] The rows of the confusion matrix represent the true land cover categories, the columns represent the identified land cover categories, and the elements... This represents the number of pixels where the actual land cover category is class k and the identified land cover category is class i; main diagonal elements. This represents the number of pixels whose actual land cover category is the kth class and whose identified land cover category is the kth class, i.e., the number of pixels that are correctly classified. ;

[0023] The total number of pixels belonging to the real feature category k is:

[0024]

[0025] The total number of pixels belonging to the identified land cover category i is:

[0026]

[0027] The total number of pixels involved in the accuracy assessment is:

[0028] N=

[0029] In one implementation of the first aspect, based on a confusion matrix, the accuracy of each of the underlying surface classification results is evaluated, and the underlying surface classification results that meet a preset accuracy standard are selected as candidate underlying surface classification results, including:

[0030] Based on the confusion matrix, a classification accuracy evaluation index is calculated; the classification accuracy evaluation index includes overall accuracy, user accuracy, mapping accuracy, and Kappa coefficient; wherein:

[0031] The overall accuracy is defined as the proportion of the total number of correctly classified pixels to the total number of pixels, and the value is equal to the ratio of the sum of the main diagonal elements of the confusion matrix to the total number of pixels.

[0032] The mapping accuracy is defined as the proportion of correctly classified pixels in the pixels of the true land cover category k, and the value is equal to the ratio of the main diagonal element of the kth row of the confusion matrix to the sum of the elements in the same row.

[0033] The user accuracy is defined as the proportion of correctly classified pixels among the pixels identifying land cover category i, and the value is equal to the ratio of the main diagonal element of the i-th column of the confusion matrix to the sum of the elements in that column.

[0034] The Kappa coefficient is used to measure the consistency between the classification results of each underlying surface and the land cover sample data, and its value range is (-1, 1).

[0035] If one or more of the classification accuracy evaluation indicators meet the preset accuracy standard, then the corresponding underlying surface classification result is taken as the candidate underlying surface classification result.

[0036] One implementation of the first aspect also includes:

[0037] Based on the simulated runoff data and combined with the measured runoff data, an objective function is constructed; the objective function is used to quantify the deviation between the simulated runoff data and the measured runoff data.

[0038] The parameters of the coupled distributed hydrological model are calibrated with the goal of minimizing the objective function in mind.

[0039] In one implementation of the first aspect, before performing land cover classification on the satellite remote sensing image, the satellite remote sensing image is further preprocessed; the preprocessing steps include: radiometric calibration, atmospheric correction, image mosaicking, and image cropping.

[0040] Secondly, this application provides a watershed runoff simulation device, comprising:

[0041] The image acquisition module is used to acquire satellite remote sensing images of the target watershed;

[0042] The supervised classification module is used to classify land cover in the satellite remote sensing images based on multiple supervised classification algorithms, and obtain multiple underlying surface classification results; the supervised classification algorithms include maximum likelihood method, neural network method and support vector machine method.

[0043] The accuracy evaluation module is used to evaluate the accuracy of each of the underlying surface classification results based on the confusion matrix, and to take the underlying surface classification results that meet the preset accuracy standard as candidate underlying surface classification results.

[0044] The model coupling module is used to couple the candidate underlying surface classification results with the distributed hydrological model to obtain the coupled distributed hydrological model.

[0045] The runoff simulation module is used to simulate the runoff process of the target watershed based on the coupled distributed hydrological model, and obtain simulated runoff data.

[0046] Thirdly, this application provides an electronic device, comprising:

[0047] The memory is used to store computer programs;

[0048] A processor, the processor being configured to execute a computer program stored in the memory, so as to cause the electronic device to perform the method described in any of the preceding descriptions.

[0049] Fourthly, this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method described in any of the preceding claims.

[0050] As described above, the watershed runoff simulation method, apparatus, electronic equipment, and medium described in this application achieve a fine characterization of the spatial heterogeneity of the watershed underlying surface by coupling high-resolution satellite remote sensing images with a distributed hydrological model, and on this basis, realize high spatiotemporal resolution runoff process simulation, effectively improving the spatiotemporal accuracy of runoff simulation. Attached Figure Description

[0051] Figure 1 The flowchart shown is a watershed runoff simulation method according to an embodiment of this application.

[0052] Figure 2 The flowchart shown is a watershed runoff simulation method according to another embodiment of this application.

[0053] Figure 3 The flowchart shown is a watershed runoff simulation method according to another embodiment of this application.

[0054] Figure 4 The diagram shown is a structural schematic of a watershed runoff simulation device according to an embodiment of this application.

[0055] Figure 5 The diagram shown is a structural schematic of an electronic device according to an embodiment of this application. Detailed Implementation

[0056] The following specific examples illustrate the implementation of this application. Those skilled in the art can easily understand other advantages and effects of this application from the content disclosed in this specification. This application can also be implemented or applied through other different specific embodiments, and various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of this application. It should be noted that, unless otherwise specified, the following embodiments and features in the embodiments can be combined with each other.

[0057] It should be noted that the illustrations provided in the following embodiments are only schematic representations of the basic concept of this application. Therefore, the drawings only show the components related to this application and are not drawn according to the actual number, shape and size of the components in the actual implementation. In the actual implementation, the form, quantity and proportion of each component can be arbitrarily changed, and the layout of the components may also be more complex.

[0058] The following embodiments of this application provide a watershed runoff simulation method, apparatus, electronic device, and medium. The method achieves a fine characterization of the spatial heterogeneity of the watershed underlying surface by coupling high-resolution satellite remote sensing images with a distributed hydrological model, and on this basis, realizes runoff process simulation with high spatiotemporal resolution, effectively improving the spatiotemporal accuracy of runoff simulation.

[0059] The technical solution of this application can be widely applied to key areas such as water resource optimization, flood forecasting and early warning, water ecological environment assessment, water conservancy project planning and design, and climate change impact analysis. Furthermore, because the data source used has global coverage, the model structure has good portability, and the parameter calibration process supports localization, this application is also applicable to runoff simulation tasks in different climate zones and watersheds of different scales, demonstrating strong versatility, adaptability, and engineering practical value.

[0060] The following will describe in detail the principles and implementation methods of a watershed runoff simulation method, apparatus, electronic device and medium of this embodiment, so that those skilled in the art can understand the watershed runoff simulation method, apparatus, electronic device and medium of this embodiment without creative effort.

[0061] Please see Figure 1 The above is a flowchart of a watershed runoff simulation method according to an embodiment of this application.

[0062] like Figure 1 As shown in the figure, this application provides a watershed runoff simulation method, which includes the following steps S100 to S500.

[0063] In step S100, satellite remote sensing images of the target watershed are acquired.

[0064] In this embodiment, the target watershed refers to a specific hydrological unit for which runoff simulation is to be performed.

[0065] Specifically, satellite remote sensing images covering the target watershed can be downloaded by accessing a publicly available remote sensing data platform. In this embodiment, the satellite remote sensing images are acquired by the OLI (Operational LandImager) / TIRS (Thermal Infrared Sensor) sensor carried by the Landsat 8 satellite. These are Level-1 products with a spatial resolution of 30 meters and contain multispectral data across multiple visible, near-infrared, and shortwave infrared bands.

[0066] In step S200, the satellite remote sensing image is classified for land cover based on multiple supervised classification algorithms, resulting in multiple underlying surface classification results.

[0067] In this embodiment, different supervised classification algorithms are used to identify land cover in the acquired satellite remote sensing images. Each algorithm independently outputs a classification result of the underlying surface. The underlying surface categories are defined according to the actual needs of watershed hydrological simulation and typically include: water bodies, forest land, grassland, cultivated land, built-up land, and bare land. These categories can effectively reflect the spatial heterogeneity of the underlying surface in the watershed and provide key input parameters for subsequent runoff process modeling.

[0068] Please see Figure 2 The flowchart shown is a watershed runoff simulation method according to another embodiment of this application.

[0069] like Figure 2 As shown, the supervised classification algorithms include Maximum Likelihood (ML), Neural Network (NN), and Support Vector Machine (SVM).

[0070] It should be noted that the maximum likelihood method, neural network method and support vector machine method described in this embodiment are all mature technologies widely used in the field of remote sensing image processing. Their basic principles and implementation methods have been described in detail in existing literature and technical materials, so this application will not repeat their specific classification process.

[0071] This implementation approach, by introducing a multi-algorithm parallel classification strategy, fully leverages the complementary advantages of different classifiers in feature learning and discrimination capabilities, significantly enhancing the robustness and spatial consistency of land cover information extraction results. These diverse underlying surface classification results will serve as the foundational data source for subsequent steps, providing strong support for improving the accuracy and stability of the entire runoff simulation system.

[0072] Please see Figure 3 The flowchart shown is a watershed runoff simulation method according to another embodiment of this application.

[0073] like Figure 3 As shown, the watershed runoff simulation method described in this application further includes the following steps S201 to S204.

[0074] In step S201, using preset global land cover data as a reference dataset, several regions of interest (ROIs) are selected for training the supervised classification algorithm.

[0075] Specifically, this embodiment uses the domestically developed 30-meter spatial resolution global land cover data GlobeLand30-2010 version as the reference dataset. This dataset is one of the most accurate and widely used global-scale land cover products internationally. Its classification system covers major land cover types such as water bodies, forest land, grassland, cultivated land, construction land, and bare land, which is highly consistent with the land use / cover categories required for watershed hydrological simulation.

[0076] In practice, the GlobeLand30-2010 data is first spatially registered and aligned with the coordinate system of the satellite remote sensing imagery to ensure precise spatial alignment. Then, based on the land cover category labels of each pixel in the GlobeLand30-2010 imagery, several representative homogeneous regions within the target watershed are manually or semi-automatically selected as training samples for each category, i.e., regions of interest (ROIs). The selected ROIs should meet the following requirements:

[0077] (1) The spatial distribution is uniform, covering different landforms and land use patterns within the basin;

[0078] (2) The spectral characteristics within the region are highly consistent, avoiding interference from mixed pixels;

[0079] (3) The area is large enough to meet the basic requirements of each supervised classification algorithm for the number of training samples.

[0080] In this implementation, the training sample set constructed in the above manner not only has clear semantic labels for land cover, but also has good matching with the remote sensing images to be classified in terms of spatial scale and observation time, thereby effectively ensuring the training quality and classification reliability of various supervised classification algorithms.

[0081] In step S202, each of the regions of interest is divided into multiple typical land cover categories.

[0082] In this embodiment, to ensure the accuracy of each supervised classification algorithm during the training process and the consistency of category semantics, the selected regions of interest need to be finely divided according to a predefined land cover classification system.

[0083] Specifically, the typical land cover categories include water bodies, bare land, cultivated land, grassland, forest land, urban land, and snow cover, wherein the water bodies include rivers and lakes. This classification system fully considers the influence mechanism of the underlying surface on hydrological processes such as rainfall generation, infiltration, and evapotranspiration during watershed runoff formation and confluence, and can effectively support subsequent hydrological model parameterization and simulation calculations.

[0084] In practice, based on the land cover label corresponding to each ROI pixel in the GlobeLand30-2010 reference dataset, and combined with manual visual interpretation or auxiliary spectral feature analysis, the original ROI is divided into sub-regions, ensuring that each sub-region contains only a single and typical land cover type. For example, for an initially defined mixed region that contains both cultivated land and forest land, pure cultivated land sub-regions and pure forest land sub-regions should be extracted as independent training samples. This process can be achieved using techniques such as mask extraction, clustering segmentation, or object-oriented classification in remote sensing image processing software.

[0085] In this implementation, by performing refined category division on the region of interest, not only is the purity and representativeness of the training samples improved, but also the structured and semantically clear labeled data is provided for each supervised classification algorithm, thereby significantly enhancing the spatial accuracy and hydrological applicability of the underlying surface classification results.

[0086] In step S203, the separability between the regions of interest is verified using Jeffries-Matusita (JM) distance and Transformed Divergence (TD) as quantitative indicators.

[0087] This embodiment introduces two classic spectral separability evaluation indices, JM and TD, to quantitatively assess the spectral distinguishability among different categories of ROIs.

[0088] Specifically, the JM distance is calculated based on the statistical difference between the two types of samples using their spectral probability density functions. The value ranges from [0, 2]. A larger value indicates stronger separability between the two types of land cover in the multispectral feature space. For example, when the JM distance is greater than 1.9, the two types of land cover are generally considered to have good separability.

[0089] The TD index projects high-dimensional spectral data to the optimal discrimination direction through linear transformation and calculates the separation degree based on the difference in inter-class means and the covariance within classes, with a value range of [0, 2000]. For example, when the TD value is greater than 1800, it is generally considered that the two types of land cover can be effectively distinguished.

[0090] In this implementation, by introducing a separability verification mechanism, a high-quality training sample set with stable spectral features and clear category boundaries can be effectively screened, thereby significantly improving the classification accuracy and robustness of various supervised classification algorithms and providing reliable land cover input data for subsequent watershed runoff simulation.

[0091] In step S204, the region of interest that has passed the separability verification is used as a training sample and input into the maximum likelihood method, the neural network method, and the support vector machine method for supervised classification.

[0092] In this embodiment, since the training samples used have passed rigorous separability verification, the risk of class confusion is effectively reduced, and the discrimination accuracy and spatial consistency of each classifier are significantly improved.

[0093] In one embodiment of this application, the watershed runoff simulation method further includes: post-processing the classification results of each underlying surface based on the primary analysis method and / or the secondary analysis method to eliminate noise patches generated during the land cover classification process.

[0094] In practical applications, considering that the classification results of the supervised classification algorithms may contain salt-and-pepper noise or isolated patches (i.e., noise patches) with excessively small areas caused by factors such as spectral confusion, mixed pixels, or classification boundary jitter, these fragmented patches not only affect the spatial continuity and visual readability of the land cover map, but may also adversely affect the accuracy of subsequent hydrological parameter extraction and runoff simulation. Therefore, spatial post-processing of the classification results is necessary.

[0095] Specifically, this embodiment uses Majority Analysis and / or Minority Analysis to optimize the underlying surface classification results.

[0096] The main analysis method involves using a defined neighborhood window (e.g., 3×3 or 5×5 pixels) as the center, counting the frequency of each category within the window, and then reassigning the center pixel to the category with the highest frequency within that window. This method can effectively smooth local noise and enhance the spatial clustering of similar land features.

[0097] Secondary analysis rules are used to identify and remove isolated patches with an area smaller than a preset threshold (e.g., an area of ​​less than 9 pixels, which is equivalent to about 8100 m², corresponding to 3×3 Landsat pixels), and merge them into the surrounding dominant neighboring categories, thereby eliminating unreasonable small-area category fragments.

[0098] In practice, the size of the neighborhood window and the minimum patch area threshold can be flexibly set according to the hydrological sensitivity and spatial distribution characteristics of different land cover types. For example, for categories with clear boundaries and strong spatial continuity, such as water bodies and urban land, a smaller filtering window (e.g., 3×3) can be used to preserve details; while for categories with strong transitions, such as grassland and bare land, the window can be appropriately increased or the area threshold can be raised to enhance stability.

[0099] In this implementation, the post-processing step effectively eliminates noise patches in the underlying surface classification results, thereby improving the classification quality of each supervised classification algorithm.

[0100] In one embodiment of this application, before performing land cover classification on the satellite remote sensing image, the method further includes preprocessing the satellite remote sensing image.

[0101] Specifically, the preprocessing steps include: radiometric calibration, atmospheric correction, image mosaicking, and image cropping.

[0102] The radiometric calibration process converts the raw digital number (DN) values ​​of satellite remote sensing imagery into physically meaningful radiance or apparent reflectance values. This process uses metadata (such as gain coefficients and offsets) provided by the Landsat 8 OLI / TIRS sensor to linearly convert DN values ​​into Top of Atmosphere (TOA) reflectance or radiance.

[0103] Atmospheric correction can eliminate the effects of atmospheric scattering, absorption, and aerosols on surface reflectance signals, further correcting the TOA reflectance to the true surface reflectance. This operation can improve the comparability between images from different times or regions and enhance the authenticity of the spectral characteristics of ground features.

[0104] If the target watershed spans multiple Landsat 8 images, adjacent images must be seamlessly stitched together. During image mosaicking, the radiometric reference of each image should be unified, and feathering blending or histogram matching techniques should be used to eliminate brightness or color differences at the seams, ensuring the spatial continuity and spectral consistency of the stitched image.

[0105] Image cropping refers to spatially cropping the processed remote sensing image based on the vector boundary of the target watershed, retaining only the image data within the target watershed area. This operation not only reduces the subsequent computational load but also avoids interference from irrelevant regions on the underlying surface classification results.

[0106] In step S300, the accuracy of each of the underlying surface classification results is evaluated based on the confusion matrix, and the underlying surface classification results that meet the preset accuracy standard are taken as candidate underlying surface classification results.

[0107] In this embodiment, the confusion matrix is ​​also called the error matrix. By comparing the correspondence between the predicted labels and the true reference labels of various surface cover categories in the underlying surface classification results, the system reflects the classifier's recognition ability and misjudgment situation in each category.

[0108] In one embodiment of this application, the steps for constructing the confusion matrix include steps S301 and S302.

[0109] In step S301, the actual land cover categories of the target watershed are determined based on the land cover sample data that are spatiotemporally matched with the satellite remote sensing image.

[0110] In step S302, the actual land cover category is compared with the identified land cover category corresponding to each underlying surface classification result by pixel comparison to form the corresponding confusion matrix.

[0111] Please refer to Table 1, which shows a confusion matrix according to an embodiment of this application. The confusion matrix uses four typical land cover categories as examples (such as water bodies, forest land, cultivated land, and urban land) to demonstrate the correspondence between the identified land cover categories and the actual land cover categories.

[0112] Table 1 Confusion Matrix

[0113]

[0114] As shown in Table 1, the confusion matrix is ​​a C×C square matrix, where C represents the total number of the true land cover categories or the identified land cover categories. The actual land cover categories are consistent with the total number of identified land cover categories.

[0115] The rows of the confusion matrix represent the true land cover categories, the columns represent the identified land cover categories, and the elements... This represents the number of pixels where the actual land cover category is class k and the identified land cover category is class i; main diagonal elements. This represents the number of pixels whose actual land cover category is the kth class and whose identified land cover category is the kth class, i.e., the number of pixels that are correctly classified. .

[0116] The total number of pixels belonging to the real feature category k is:

[0117]

[0118] The total number of pixels belonging to the identified land cover category i is:

[0119]

[0120] The total number of pixels involved in the accuracy assessment is:

[0121] N=

[0122] In one embodiment of this application, the accuracy evaluation of each of the underlying surface classification results is performed based on the confusion matrix, and the underlying surface classification results that meet the preset accuracy standard are used as candidate underlying surface classification results, including the following steps S301 and S302.

[0123] In step S301, several classification accuracy evaluation indicators are calculated based on the confusion matrix.

[0124] Specifically, the classification accuracy evaluation indicators include overall accuracy, user accuracy, mapping accuracy, and Kappa coefficient.

[0125] The overall accuracy is defined as the proportion of the total number of correctly classified pixels to the total number of pixels, and the value is equal to the ratio of the sum of the main diagonal elements of the confusion matrix to the total number of pixels.

[0126] The mapping accuracy is defined as the proportion of correctly classified pixels in the real land cover category k, and the value is equal to the ratio of the main diagonal element of the kth row of the confusion matrix to the sum of the elements in that row.

[0127] The user accuracy is defined as the proportion of correctly classified pixels among the pixels identifying land cover category i, and the value is equal to the ratio of the main diagonal element of the i-th column of the confusion matrix to the sum of the elements in that column.

[0128] The Kappa coefficient is used to measure the consistency between the classification results of each underlying surface and the land cover sample data, and its value range is (-1, 1).

[0129] For example, when the Kappa coefficient is greater than 0.75, it is generally considered that the underlying surface classification results are highly consistent with the actual land cover categories, and the classification quality reaches a relatively ideal level; when the Kappa coefficient is between 0.40 and 0.75, it indicates that the underlying surface classification results have moderate consistency; while below 0.40, it indicates that the reliability of the underlying surface classification results is low, and there may be significant misjudgments or systematic biases.

[0130] In step S302, if one or more of the classification accuracy evaluation indicators meet the preset accuracy standard, the corresponding underlying surface classification result is taken as the candidate underlying surface classification result.

[0131] In one embodiment of this application, if a certain underlying surface classification result simultaneously satisfies Kappa ≥ 0.75 and overall accuracy ≥ 85%, it is marked as a candidate underlying surface classification result.

[0132] If multiple underlying surface classification results meet the above criteria, a further horizontal comparison is performed, and the one with the highest Kappa coefficient is selected as the candidate underlying surface classification result. This strategy prioritizes ensuring the statistical consistency and generalization ability of the underlying surface classification results, and is particularly suitable for hydrological simulation applications sensitive to land cover boundaries and category purity.

[0133] If no underlying surface classification result meets the preset standard, a feedback mechanism can be triggered, such as optimizing training samples, adjusting classifier parameters, or introducing a fusion classification strategy to improve classification performance.

[0134] In this implementation, by introducing a confusion matrix and a multi-dimensional accuracy evaluation system including overall accuracy, mapping accuracy, user accuracy, and Kappa coefficient, the underlying surface classification results output by each supervised classification algorithm can be systematically and quantitatively evaluated. Only underlying surface classification results that meet the preset accuracy threshold are included in the candidate set, ensuring that the land cover data entering the runoff simulation stage have sufficient credibility and hydrological representativeness.

[0135] In step S400, the candidate underlying surface classification results are coupled with the distributed hydrological model to obtain the coupled distributed hydrological model.

[0136] In this embodiment, typical distributed hydrological models include SWAT (Soil and Water Assessment Tool), VIC (Variable Infiltration Capacity), TOPMODEL, HEC-HMS, and other models.

[0137] Specifically, coupling the candidate underlying surface classification results with the distributed hydrological model includes replacing the original underlying surface-related parameter information in the distributed hydrological model with candidate underlying surface information obtained from supervised classification.

[0138] During the coupling process, the following basic data are simultaneously integrated to construct a complete input system for the distributed hydrological model:

[0139] (1) Digital Elevation Model (DEM): used to extract topographic parameters such as river network, sub-basin boundaries, slope, and flow direction;

[0140] (2) Measured meteorological driving data: including precipitation, temperature, solar radiation, relative humidity, wind speed and water vapor pressure, etc., with a time resolution of daily or hourly scale.

[0141] This implementation method achieves deep integration of remote sensing surface information and distributed hydrological model, which significantly improves the model's ability to characterize the spatial heterogeneity of the underlying surface of the watershed and enhances the physical rationality and spatiotemporal accuracy of the runoff generation and confluence process simulation.

[0142] In step S500, based on the coupled distributed hydrological model, the runoff process of the target watershed is simulated to obtain simulated runoff data.

[0143] Specifically, based on the coupled distributed hydrological model, simulating the runoff process of the target watershed involves dividing the target watershed into several computational units, combining measured meteorological driving data within each computational unit, candidate underlying surface classification results, and the hydrological response function obtained from model coupling, and solving the water balance equation and hydrodynamic control equations such as the kinematic wave or Saint-Venant equation at time steps to obtain the hourly (or daily) runoff sequence of the target watershed outlet section or a designated sub-watershed. This simulation result is the simulated runoff data, which can be used for subsequent comparison and verification with measured runoff data, model calibration, or uncertainty analysis, thereby supporting applications such as watershed water resources management, flood forecasting, and eco-hydrological assessment.

[0144] In one embodiment of this application, the watershed runoff simulation method further includes the following steps S600 and S700.

[0145] In step S600, an objective function is constructed based on the simulated runoff data and combined with the measured runoff data.

[0146] In one embodiment of this application, the objective function is used to quantify the deviation between the simulated runoff data and the measured runoff data.

[0147] In step S700, the parameters of the coupled distributed hydrological model are calibrated with the goal of minimizing the objective function as the guide.

[0148] A calibrated distributed hydrological model can output high-precision flow process curves at the watershed outlet section, which can be used for practical applications such as water resource allocation, flood warning, or ecological flow assessment.

[0149] It should be noted that the scope of protection of the watershed runoff simulation method described in this application is not limited to the order of steps listed in this embodiment. Any solution implemented by adding, subtracting, or replacing steps in the prior art based on the principles of this application is included within the scope of protection of this application.

[0150] Please see Figure 4 The image shown is a schematic diagram of a watershed runoff simulation device according to an embodiment of this application.

[0151] like Figure 4 As shown, this embodiment provides a watershed runoff simulation device, including an image acquisition module, a supervised classification module, an accuracy evaluation module, a model coupling module, and a runoff simulation module.

[0152] The image acquisition module is used to acquire satellite remote sensing images of the target watershed.

[0153] The supervised classification module is used to classify land cover in the satellite remote sensing imagery based on multiple supervised classification algorithms, resulting in multiple underlying surface classification results. The supervised classification algorithms include maximum likelihood estimation, neural network methods, and support vector machine methods.

[0154] The accuracy evaluation module is used to evaluate the accuracy of each of the underlying surface classification results based on the confusion matrix, and to take the underlying surface classification results that meet the preset accuracy standard as candidate underlying surface classification results.

[0155] The model coupling module is used to couple the candidate underlying surface classification results with the distributed hydrological model to obtain the coupled distributed hydrological model.

[0156] The runoff simulation module is used to simulate the runoff process of the target watershed based on the coupled distributed hydrological model, and obtain simulated runoff data.

[0157] It should be noted that the structure and principle of the image acquisition module, the supervised classification module, the accuracy evaluation module, the model coupling module and the runoff simulation module described in this embodiment correspond one-to-one with the steps in the above-mentioned watershed runoff simulation method, so they will not be repeated here.

[0158] The watershed runoff simulation device provided in this application embodiment can implement the watershed runoff simulation method described in this application. However, the implementation device of the watershed runoff simulation method described in this application includes, but is not limited to, the structure of the watershed runoff simulation device listed in this embodiment. All structural modifications and substitutions of the prior art made based on the principles of this application are included within the protection scope of this application.

[0159] Please see Figure 5 The image shown is a schematic diagram of the structure of an electronic device according to an embodiment of this application.

[0160] like Figure 5 As shown, this embodiment provides an electronic device. The electronic device of this application is embodied in the form of a general-purpose computing device. The components of the electronic device may include, but are not limited to: a memory; a bus connecting different system components (including memory and processing units) for storing computer programs; one or more processors or processing units for executing the computer programs stored in the memory, so that the electronic device performs any of the methods described above.

[0161] A bus refers to one or more of several bus architectures, including a memory bus or memory controller, a peripheral bus, a graphics acceleration port, a processor, or a local bus using any of the various bus architectures. Examples of these architectures include, but are not limited to, the Industry Standard Architecture (ISA) bus, the Micro Channel Architecture (MAC) bus, the Enhanced ISA bus, the Video Electronics Standards Association (VESA) local bus, and the Peripheral Component Interconnect (PCI) bus.

[0162] Electronic devices typically include a variety of computer-readable media. These media can be any available media that can be accessed by the electronic device, including volatile and non-volatile media, and removable and non-removable media.

[0163] Memory may include computer system readable media in the form of volatile memory, such as random access memory (RAM) and / or cache memory. Electronic devices may further include other removable / non-removable, volatile / non-volatile computer system storage media. By way of example only, a storage system may be used to read and write non-removable, non-volatile magnetic media (… Figure 5 Not shown; usually referred to as a "hard drive"). Although Figure 5 As not shown, a disk drive for reading and writing to a removable non-volatile disk (e.g., a "floppy disk") and an optical disk drive for reading and writing to a removable non-volatile optical disk (e.g., a CD-ROM, DVD-ROM, or other optical media) may be provided. In these cases, each drive may be connected to a bus via one or more data media interfaces. The memory may include at least one program product having a set (e.g., at least one) of program modules configured to perform the functions of the embodiments of this application.

[0164] A program / utility having a set (at least one) of program modules can be stored, for example, in memory. Such program modules include, but are not limited to, an operating system, one or more applications, other program modules, and program data. Each or some combination of these examples may include an implementation of a network environment. The program modules typically perform the functions and / or methods described in the embodiments of this application.

[0165] Electronic devices can also communicate with one or more external devices (such as keyboards, pointing devices, displays, etc.), one or more devices that enable user interaction with the electronic device, and / or any device that enables the electronic device to communicate with one or more other computing devices (such as network interface cards, modems, etc.). This communication can be performed through input / output (I / O) interfaces. Furthermore, electronic devices can communicate with one or more networks (such as local area networks (LANs), wide area networks (WANs), and / or public networks, such as the Internet) via network adapters. Figure 5 As shown, the network adapter communicates with other modules of the electronic device via a bus. It should be understood that, although not shown in the figure, other hardware and / or software modules can be used in conjunction with the electronic device, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.

[0166] It should be noted that the watershed runoff simulation method described above can be applied to various types of hardware devices. The hardware device can be a computer including components such as a memory, memory controller, one or more processing units (CPUs), peripheral interfaces, RF circuits, audio circuits, speakers, microphones, input / output (I / O) subsystems, displays, other output or control devices, and external ports; the computer includes, but is not limited to, personal computers such as desktop computers, laptops, tablets, smartphones, smart TVs, and personal digital assistants (PDAs). In other embodiments, the hardware device can also be a local server or a cloud server. The server can be deployed on one or more physical servers depending on factors such as function and load, or it can be composed of distributed or centralized server clusters; this embodiment does not impose any limitations.

[0167] In the embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, or methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative. For instance, the division of modules / units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple modules or units may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection of apparatuses or modules or units may be electrical, mechanical, or other forms.

[0168] The modules / units described as separate components may or may not be physically separate. The components shown as modules / units may or may not be physical modules; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules / units can be selected to achieve the objectives of the embodiments of this application, depending on actual needs. For example, the functional modules / units in the various embodiments of this application may be integrated into one processing module, or each module / unit may exist physically separately, or two or more modules / units may be integrated into one module / unit.

[0169] Those skilled in the art will further recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0170] This application also provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the methods described in any of the above embodiments. Those skilled in the art will understand that all or part of the steps in the methods of the above embodiments can be implemented by a program instructing a processor. The program can be stored in a computer-readable storage medium, which is a non-transitory medium, such as random access memory, read-only memory, flash memory, hard disk, solid-state drive, magnetic tape, floppy disk, optical disk, and any combination thereof. The storage medium can be any available medium accessible to a computer or a data storage device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., digital video disc (DVD)), or a semiconductor medium (e.g., solid-state disk (SSD)).

[0171] The descriptions of the processes or structures corresponding to the above figures each have their own emphasis. For parts of a process or structure that are not described in detail, please refer to the relevant descriptions of other processes or structures.

[0172] The above embodiments are merely illustrative of the principles and effects of this application and are not intended to limit this application. Any person skilled in the art can modify or alter the above embodiments without departing from the spirit and scope of this application. Therefore, all equivalent modifications or alterations made by those skilled in the art without departing from the spirit and technical concept disclosed in this application should still be covered by the claims of this application.

Claims

1. A method for simulating watershed runoff, characterized in that, include: Acquire satellite remote sensing images of the target watershed; Based on multiple supervised classification algorithms, the satellite remote sensing images are classified for land cover, resulting in multiple underlying surface classification results. The supervised classification algorithms include maximum likelihood estimation, neural network methods, and support vector machine methods. Based on the confusion matrix, the accuracy of each underlying surface classification result is evaluated, and the underlying surface classification result that meets the preset accuracy standard is taken as the candidate underlying surface classification result. The candidate underlying surface classification results are coupled with the distributed hydrological model to obtain the coupled distributed hydrological model. Based on the coupled distributed hydrological model, the runoff process of the target watershed is simulated to obtain simulated runoff data.

2. The method according to claim 1, characterized in that, Also includes: Using a preset global land cover data as a reference dataset, several regions of interest are selected for training the supervised classification algorithm; Each region of interest is divided into multiple typical land cover categories; the typical land cover categories include water bodies, bare land, cultivated land, grassland, woodland, urban land and snow cover, wherein the water bodies include rivers and lakes; The separability between the regions of interest was verified using Jeffries-Matusita distance and transformation separation degree as quantitative indicators. The regions of interest that have passed the separability verification are used as training samples and input into the maximum likelihood method, the neural network method, and the support vector machine method for supervised classification, respectively.

3. The method according to claim 1, characterized in that, Also includes: Based on primary and / or secondary analysis methods, the classification results of each underlying surface are post-processed to eliminate noise patches generated during the land cover classification process.

4. The method according to claim 1, characterized in that, The steps for constructing the confusion matrix include: Based on the land cover sample data that is spatiotemporally matched with the satellite remote sensing image, the true land cover categories of the target watershed are determined; The actual land cover categories are compared with the identified land cover categories corresponding to the underlying surface classification results using pixels to form the corresponding confusion matrix; wherein: The confusion matrix is ​​a C×C square matrix, where C represents the total number of the real feature categories or the total number of the identified feature categories, and the real feature categories are the same as the total number of the identified feature categories; The rows of the confusion matrix represent the true land cover categories, the columns represent the identified land cover categories, and the elements... This represents the number of pixels where the actual land cover category is class k and the identified land cover category is class i; main diagonal elements. This represents the number of pixels whose actual land cover category is the kth class and whose identified land cover category is the kth class, i.e., the number of pixels that are correctly classified. ; The total number of pixels belonging to the real feature category k is: The total number of pixels belonging to the identified land cover category i is: The total number of pixels participating in the accuracy assessment is: N= 5. The method according to claim 4, characterized in that, Based on the confusion matrix, the accuracy of each underlying surface classification result is evaluated, and the underlying surface classification results that meet the preset accuracy standard are selected as candidate underlying surface classification results, including: Based on the confusion matrix, a classification accuracy evaluation index is calculated; the classification accuracy evaluation index includes overall accuracy, user accuracy, mapping accuracy, and Kappa coefficient; wherein: The overall accuracy is defined as the proportion of the total number of correctly classified pixels to the total number of pixels, and the value is equal to the ratio of the sum of the main diagonal elements of the confusion matrix to the total number of pixels. The mapping accuracy is defined as the proportion of correctly classified pixels in the pixels of the true land cover category k, and the value is equal to the ratio of the main diagonal element of the kth row of the confusion matrix to the sum of the elements in the same row. The user accuracy is defined as the proportion of correctly classified pixels among the pixels identifying land cover category i, and the value is equal to the ratio of the main diagonal element of the i-th column of the confusion matrix to the sum of the elements in that column. The Kappa coefficient is used to measure the consistency between the classification results of each underlying surface and the land cover sample data, and its value range is (-1, 1). If one or more of the classification accuracy evaluation indicators meet the preset accuracy standard, then the corresponding underlying surface classification result is taken as the candidate underlying surface classification result.

6. The method according to claim 1, characterized in that, Also includes: Based on the simulated runoff data and combined with the measured runoff data, an objective function is constructed; the objective function is used to quantify the deviation between the simulated runoff data and the measured runoff data. The parameters of the coupled distributed hydrological model are calibrated with the goal of minimizing the objective function in mind.

7. The method according to claim 1, characterized in that, Before performing land cover classification on the satellite remote sensing images, the process also includes preprocessing the satellite remote sensing images; the preprocessing steps include: radiometric calibration, atmospheric correction, image mosaicking, and image cropping.

8. A watershed runoff simulation device, characterized in that, include: The image acquisition module is used to acquire satellite remote sensing images of the target watershed; The supervised classification module is used to classify land cover in the satellite remote sensing images based on multiple supervised classification algorithms, and obtain multiple underlying surface classification results; the supervised classification algorithms include maximum likelihood method, neural network method and support vector machine method. The accuracy evaluation module is used to evaluate the accuracy of each of the underlying surface classification results based on the confusion matrix, and to take the underlying surface classification results that meet the preset accuracy standard as candidate underlying surface classification results. The model coupling module is used to couple the candidate underlying surface classification results with the distributed hydrological model to obtain the coupled distributed hydrological model. The runoff simulation module is used to simulate the runoff process of the target watershed based on the coupled distributed hydrological model, and obtain simulated runoff data.

9. An electronic device, characterized in that, include: The memory is used to store computer programs; A processor for executing a computer program stored in the memory to cause the electronic device to perform the method of any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the method of any one of claims 1 to 7.