SWAT model correction and application method based on remote sensing data
By applying various remote sensing data and advanced data processing techniques to the SWAT model, the problems of spatiotemporal resolution and parameter calibration of traditional models have been solved, enabling high-precision simulation of hydrological processes and analysis of climate change impacts in complex watersheds.
Patent Information
- Application Number
- CN202511099658.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-08-06
- Publication Date
- 2025-11-21
AI Technical Summary
Traditional SWAT models rely on ground-based measured data, which suffer from low spatiotemporal resolution, incomplete coverage, time-consuming parameter calibration, insufficient adaptability to complex underlying surfaces, and lack of real-time response capabilities to vegetation dynamics and soil moisture, leading to a decline in prediction accuracy under the influence of climate change and human activities.
By acquiring various remote sensing data and performing preprocessing, a SWAT model is established. The model parameters are corrected using local sensitivity analysis and ensemble Kalman filtering techniques, and deep reinforcement learning is combined to achieve adaptive parameter adjustment. Physical constraints are introduced, and the model is applied to the simulation and monitoring of hydrological processes.
It improves the accuracy and rationality of hydrological process simulation, optimizes computational efficiency, enhances the ability to compensate for dynamic errors in complex watersheds, and improves the accuracy of climate change impact analysis.
Smart Images
Figure CN120997671A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of hydrological and water resources model optimization technology, and more specifically to a method for correcting and applying SWAT models based on remote sensing data. Background Technology
[0002] The SWAT (Soil and Water Assessment Tool) model was initially developed to predict the long-term impacts of land management on water, sediment, and chemicals under complex and varied soil types, land use patterns, and management practices in large watersheds. The SWAT model uses continuous daily calculations. It is a distributed watershed hydrological model based on GIS and has seen rapid development and application in recent years. It primarily utilizes spatial information provided by remote sensing and geographic information systems to simulate various hydrophysical and chemical processes, such as water quantity, water quality, and pesticide transport and transformation processes.
[0003] SWAT is a physically based model capable of simulating continuous time series. The watershed hydrological processes simulated by SWAT are divided into the terrestrial phase of the hydrological cycle (i.e., runoff generation and slope runoff) and the confluence phase (i.e., channel runoff). The former controls the input of water, sediment, nutrients, and chemicals into the main channel of each sub-basin; the latter determines the transport of water, sediment, and other substances from the river network to the watershed outlet. The entire water cycle system follows the law of water balance.
[0004] Traditional SWAT models have the following technical shortcomings: they rely on ground-based measured data to obtain hydrological parameters (such as evapotranspiration and runoff coefficient), resulting in low spatiotemporal resolution and incomplete coverage; the model parameter calibration process is time-consuming and lacks adaptability to complex underlying surface conditions; they lack the ability to respond in real time to key elements such as vegetation dynamics and soil moisture; and their prediction accuracy decreases under the influence of climate change and human activities.
[0005] Therefore, proposing a method for correcting and applying SWAT models based on remote sensing data to address the difficulties in existing technologies is a problem that urgently needs to be solved by those skilled in the art. Summary of the Invention
[0006] In view of this, the present invention provides a method for SWAT model correction and application based on remote sensing data. The method corrects parameters by establishing a SWAT model and applies various remote sensing data to the established SWAT model to analyze and monitor topographic and river information in hydrological processes.
[0007] To achieve the above objectives, the present invention provides the following technical solution:
[0008] A method for correcting and applying SWAT model based on remote sensing data, comprising the following steps:
[0009] S1, obtaining multiple remote sensing data in the hydrological process;
[0010] S2, preprocessing the multiple remote sensing data;
[0011] S3, establishing a corresponding SWAT model, and correcting parameters through the established SWAT model;
[0012] S4, applying the multiple remote sensing data to the established SWAT model, and analyzing and monitoring the topography and river information in the hydrological process.
[0013] Optionally, the multiple remote sensing data in the hydrological process obtained in S1 includes optical remote sensing data, microwave remote sensing data and thermal infrared remote sensing data.
[0014] Optionally, the multiple remote sensing data in S2 is subjected to data screening, radiation correction and atmospheric correction, invalid value processing, conversion and clipping preprocessing.
[0015] Optionally, the specific content of establishing a corresponding SWAT model in S3 is:
[0016] Collecting basic data including basin boundary vector, digital elevation model, land use / cover map, soil type map and meteorological data;
[0017] Dividing homogeneous sub-basins and sub-basins based on the basic data;
[0018] Selecting a model according to the characteristics of the target basin;
[0019] In the modeling process, input meteorological data, soil parameter data and vegetation parameter data, run SWAT using default parameters, and output time series of basin runoff, evapotranspiration and soil moisture content.
[0020] Optionally, the input meteorological data includes collected precipitation, air temperature, wind speed, relative humidity and solar radiation, and daily meteorological data is generated by interpolation through a SWAT meteorological generator;
[0021] The soil parameter data includes soil texture, organic matter content, saturated hydraulic conductivity, field moisture capacity and permanent wilting point extracted based on the soil type map, and a soil hydrological property table input into SWAT;
[0022] The vegetation parameter data includes determining vegetation types based on the land use map, and initializing vegetation coverage, root depth and surface roughness in combination with the NDVI / EVI time series or LAI inverted by remote sensing.
[0023] Optionally, the specific content of correcting parameters through the established SWAT model in S3 is:
[0024] Identify the core parameters that have a significant impact on the simulation results through local sensitivity analysis;
[0025] Establish a quantitative mapping relationship between the key hydrological parameters and the core parameters of the SWAT model through inversion;
[0026] Use the ensemble Kalman filter to assimilate real-time parameters inverted from remote sensing into the state variables of the SWAT model to correct the initial conditions and parameter deviations of the model, and introduce deep reinforcement learning to achieve adaptive adjustment of parameters, introduce physical constraints and prior knowledge to avoid parameter deviation from the physical reasonable range in the optimization process.
[0027] Optionally, in S4, the multiple remote sensing data are applied to the established SWAT model to analyze and monitor the specific content of the topography and river information in the hydrological process:
[0028] The response of runoff to future climate change is analyzed, the grid precipitation and temperature data of various scenarios of the climate model are interpolated to each sub-basin, the SWAT model is driven, the runoff process of the typical section in the past and the future is simulated, and the influence of climate change is compared and analyzed.
[0029] According to the above technical solution, compared with the prior art, the present application provides a SWAT model correction and application method based on remote sensing data, which has the following beneficial effects:
[0030] (1) The present application realizes the correction of parameters through the established SWAT model, applies multiple remote sensing data to the established SWAT model, and analyzes and monitors the topography and river information in the hydrological process;
[0031] (2) The remote sensing data of the present application can provide large-scale and high-frequency data to make up for the shortcomings of traditional data, which is beneficial to improve the simulation accuracy and rationality of the hydrological process;
[0032] (3) The present application develops a cross-scale fusion method of multi-source remote sensing data and the SWAT model, constructs a dynamic error compensation model correction system for complex basins, and realizes the optimization of calculation efficiency and the improvement of prediction accuracy. BRIEF DESCRIPTION OF DRAWINGS
[0033] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed in the embodiment or prior art description will be briefly introduced below. Obviously, the drawings in the following description are only embodiments of the present application, and those skilled in the art can obtain other drawings according to the provided drawings without creating any creative labor.
[0034] Figure 1A flow chart of a method for correcting and applying a SWAT model based on remote sensing data is provided. DETAILED DESCRIPTION
[0035] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by a person of ordinary skill in the art without creative effort belong to the scope of protection of the present application.
[0036] Reference Figure 1 As shown in the drawings, the present application discloses a method for correcting and applying a SWAT model based on remote sensing data, comprising the following steps:
[0037] S1, obtaining multiple remote sensing data in a hydrological process;
[0038] S2, preprocessing the multiple remote sensing data;
[0039] S3, establishing a corresponding SWAT model, and correcting parameters through the established SWAT model;
[0040] S4, applying the multiple remote sensing data to the established SWAT model, and analyzing and monitoring topography and river information in the hydrological process.
[0041] Further, the multiple remote sensing data in the hydrological process obtained in S1 comprises optical remote sensing data, microwave remote sensing data and thermal infrared remote sensing data.
[0042] Further, the multiple remote sensing data in S2 is subjected to data screening, radiation correction and atmospheric correction, invalid value processing, conversion and clipping preprocessing.
[0043] Specifically, data screening: through data checking, cloud mask (MODIS MYD35 / MYD09 product) is used to remove images with cloud coverage > 30%, and abnormal data such as strip noise and bad lines are filtered out;
[0044] Radiation correction and atmospheric correction: remote sensing data is affected by factors such as sensor response, atmospheric scattering / absorption and solar elevation angle, and DN value (digital quantization value) needs to be converted into physical meaning clear radiation brightness or surface reflectivity / temperature through correction, eliminating the influence of atmospheric molecules (O2, H2O) and aerosol scattering / absorption on the surface signal, and obtaining the real reflectivity or brightness temperature of the ground;
[0045] Invalid value processing: noise removal and invalid value marking are performed;
[0046] Conversion and clipping: convert the projection of different sensors (such as the sinusoidal projection of MODIS, the UTM projection of Sentinel-2) into the projection required by the SWAT model (usually Albers equal-area conic projection or UTM projection), ensuring spatial scale consistency; according to the boundary of the study basin (extracted by DEM or vector boundary file), clip the image to reduce redundant calculation.
[0047] Further, the specific content of establishing a corresponding SWAT model in S3 is:
[0048] Collecting basic data including basin boundary vector, digital elevation model, land use / cover map, soil type map, and meteorological data;
[0049] Dividing homogeneous sub-basins and sub-basins based on basic data;
[0050] Selecting the model according to the characteristics of the target basin;
[0051] During the modeling process, input meteorological data, soil parameter data, and vegetation parameter data, run SWAT using default parameters, and output time series of basin runoff, evapotranspiration, and soil moisture.
[0052] Further, the input meteorological data includes collecting precipitation, air temperature, wind speed, relative humidity, and solar radiation, and generating daily meteorological data by interpolation through the SWAT weather generator;
[0053] Soil parameter data includes extracting soil texture, organic matter content, saturated hydraulic conductivity, field capacity, and permanent wilting point based on soil type map, and inputting soil hydrological property table to SWAT;
[0054] Vegetation parameter data includes determining vegetation type based on land use map, combining remote sensing inversion of NDVI / EVI time series or LAI, and initializing vegetation coverage, root depth, and surface roughness.
[0055] Further, the specific content of parameter correction through the established SWAT model in S3 is:
[0056] Identify the core parameters that significantly affect the simulation results through local sensitivity analysis;
[0057] Invert key hydrological parameters and establish quantitative mapping relationship with core parameters of SWAT model;
[0058] Use ensemble Kalman filter to assimilate real-time parameters inverted by remote sensing into SWAT model state variables, correct model initial conditions and parameter deviation, and introduce deep reinforcement learning to realize adaptive adjustment of parameters, introduce physical constraints and prior knowledge to avoid parameter deviation from physical reasonable range in optimization process.
[0059] Further, in S4, multiple remote sensing data are applied to the established SWAT model to analyze and monitor the specific content of topography and river information in the hydrological process:
[0060] The response of runoff to future climate change is analyzed by interpolating the grid precipitation and temperature data of various climate scenarios to each sub-basin to drive the SWAT model, simulate the runoff process of typical sections in the past and future, and compare and analyze the impact of climate change.
[0061] In a specific embodiment, the following is included:
[0062] Remote sensing data: optical remote sensing (NDVI, EVI, 16-day synthesis, 250m resolution), microwave remote sensing Sentinel-1 (VV / VH backscatter coefficient, 10m resolution, 12-day revisit), thermal infrared remote sensing (LST, 30m resolution).
[0063] Ground measured data: daily weather data (precipitation, temperature) from 2020 to 2024, daily runoff of 3 hydrological stations (for calibration and verification), 5 soil moisture observation points (for inversion parameter verification).
[0064] Basin division: 12 sub-basins, 45 homogeneous sub-basins (HRU) (based on LU / LC, soil, slope superposition).
[0065] Initialization parameters: CN value (70 for forest land, 85 for farmland, 95 for city), ESCO (default 0.95), ALPHA_BF (default 0.5).
[0066] Establish the corresponding SWAT model according to the characteristics of the target basin;
[0067] In the modeling process, the input meteorological data includes collecting precipitation, temperature, wind speed, relative humidity and solar radiation, and generating daily meteorological data by interpolating the SWAT meteorological generator;
[0068] Soil parameter data includes extracting soil texture, organic matter content, saturated hydraulic conductivity, field moisture capacity, and permanent wilting point based on the soil type map, and inputting the soil hydrological property table of SWAT;
[0069] Vegetation parameter data includes determining vegetation types based on land use maps, combining remote sensing inversion of NDVI / EVI time series or LAI, and initializing vegetation coverage, root depth, and surface roughness.
[0070] Output time series of basin runoff, evapotranspiration, and soil moisture content.
[0071] The core parameters that have significant influence on the simulation results are identified by local sensitivity analysis; the quantitative mapping relationship between the key hydrological parameters and the core parameters of the SWAT model is established by inversion; the real-time parameters inverted by remote sensing are assimilated into the state variables of the SWAT model by using the ensemble Kalman filter to correct the initial conditions and parameter deviations of the model, and the deep reinforcement learning is introduced to realize adaptive adjustment of the parameters, the physical constraints and prior knowledge are introduced to avoid the parameters deviating from the physically reasonable range in the optimization process; and the parameters are corrected.
[0072] The remote sensing inverted parameters are updated monthly (e.g., the FVC is calculated by using the MODIS NDVI obtained on the 1st day of each month to update the CN value).
[0073] The various embodiments in the specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referred to each other.
[0074] The above description of the disclosed embodiments enables a person skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to these embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.
Claims
1. A method for correcting and applying a SWAT model based on remote sensing data, characterized in that, Includes the following steps: S1. Acquire various remote sensing data in the hydrological process; S2. Preprocessing various remote sensing data; S3. Establish the corresponding SWAT model and use the established SWAT model to correct the parameters; S4. Apply various remote sensing data to the established SWAT model to analyze and monitor topographic and river information during hydrological processes.
2. The method for correcting and applying a SWAT model based on remote sensing data according to claim 1, characterized in that, The hydrological processes acquired in S1 include various remote sensing data such as optical remote sensing data, microwave remote sensing data, and thermal infrared remote sensing data.
3. The method for correcting and applying a SWAT model based on remote sensing data according to claim 1, characterized in that, S2 performs data filtering, radiometric and atmospheric correction, invalid value processing, transformation and cropping preprocessing on various remote sensing data.
4. The method for correcting and applying a SWAT model based on remote sensing data according to claim 1, characterized in that, The specific steps for building the corresponding SWAT model in S3 are as follows: Collect basic data including watershed boundary vectors, digital elevation models, land use / cover maps, soil type maps, and meteorological data; Based on fundamental data, the watershed is divided into homoprotic watersheds and sub-watersheds; Select a model based on the characteristics of the target watershed; During the modeling process, meteorological data, soil parameter data, and vegetation parameter data are input, and SWAT is run using default parameters to output time series of watershed runoff, evapotranspiration, and soil moisture content.
5. A method for correcting and applying a SWAT model based on remote sensing data according to claim 4, characterized in that, The input meteorological data includes collected precipitation, temperature, wind speed, relative humidity, and solar radiation, which are interpolated by the SWAT weather generator to generate daily meteorological data. Soil parameter data includes soil texture, organic matter content, saturated hydraulic conductivity, field capacity, and permanent wilting point extracted from the soil type map and input into the SWAT soil hydrological attribute table; Vegetation parameter data include determining vegetation type based on land use maps, and initializing vegetation cover, root depth, and surface roughness by combining NDVI / EVI time series or LAI retrieved from remote sensing.
6. A method for correcting and applying a SWAT model based on remote sensing data according to claim 1 or 4, characterized in that, The specific details of parameter correction achieved through the SWAT model established in S3 are as follows: Local sensitivity analysis was used to identify the core parameters that significantly affect the simulation results. A quantitative mapping relationship is established between the key hydrological parameters obtained by inversion and the core parameters of the SWAT model; Ensemble Kalman filtering is used to assimilate the real-time parameters retrieved from remote sensing into the state variables of the SWAT model, correcting the initial conditions and parameter deviations of the model. At the same time, deep reinforcement learning is introduced to achieve adaptive parameter adjustment, and physical constraints and prior knowledge are introduced to avoid parameters deviating from the physically reasonable range during the optimization process.
7. The method for correcting and applying a SWAT model based on remote sensing data according to claim 1, characterized in that, S4 applies various remote sensing data to the established SWAT model to analyze and monitor topographic and river channel information during hydrological processes. The specific content includes: The study analyzes the response of runoff to future climate change by interpolating precipitation and temperature data from various climate model scenarios onto each sub-basin, driving the SWAT model to simulate past and future runoff processes at typical cross sections, and comparing and analyzing the impact of climate change.