Mountain area drainage basin rainfall-runoff refined simulation system and method

Abstract

The invention relates to the technical field of runoff simulation, in particular to a rainfall-runoff refined simulation system and method for a mountainous area drainage basin, and the method comprises the following steps: 1, preprocessing a data set and data; 2, designing a U-Net super-resolution model TA-UNet based on a terrain attention mechanism as a rainfall downscaling model, and training the rainfall downscaling model; 3, verifying the precision; step 4, rainfall downscaling; and step 5, rainfall-runoff simulation. According to the invention, the spatial rainfall characteristics of the mountainous area drainage basin can be better captured, and the rainfall-runoff propagation process can be simulated in detail.

Description

Technical Field

[0001] The invention relates to the technical field of runoff simulation, and in particular to a system and method for fine-grained simulation of precipitation and runoff in a mountainous watershed. Background Art

[0002] Runoff from small watersheds triggered by heavy rainfall can lead to secondary disasters such as flash floods, landslides, and debris flows, which have caused a large number of casualties and huge economic losses worldwide. The triggering of these disasters depends on multiple factors such as hydrological conditions, geomorphological characteristics, slope characteristics, and sediment source volume, among which precipitation is the key driving factor. The observation of precipitation data directly affects the timeliness and accuracy of geological disaster forecasts. However, sparsely distributed ground rain gauges and coarse-resolution meteorological radars are difficult to accurately reflect local precipitation patterns at the watershed scale. By analyzing field measurements and historical data, the researchers found that unique local precipitation patterns were formed under the micro-topography effect. These characteristics indicate that the precipitation intensity in the debris flow source area may be much higher than the disaster threshold provided by local meteorological stations at lower altitudes. The significant changes in local precipitation greatly increase the unpredictability of secondary disasters caused by runoff. Therefore, it is urgent to accurately obtain high-resolution precipitation data that reflects the local precipitation patterns of small watersheds under complex terrain.

[0003] In recent years, the advancement of satellite remote sensing technology has made it an effective tool for obtaining spatial continuity and temporal full coverage estimates of global precipitation. Many meteorological reanalysis products obtained through satellite remote sensing technology, such as TRMM, IMERG GPM, ERA5-Land and CLDAS, can effectively perform statistical analysis of large-scale spatial precipitation distribution and trends, and are key resources for regional precipitation analysis. Researchers then used dynamic downscaling and statistical downscaling techniques to improve the spatial resolution of meteorological satellite observation data. Dynamic downscaling technology, namely regional climate models, establish and simulate atmospheric thermodynamic equations by integrating topography, meteorological factors and other mathematical and physical factors to calculate local climate fields. It can accurately simulate climate models with clear physical mechanisms, but it requires a lot of computing resources and has limited spatial generalization capabilities. In contrast, statistical downscaling methods use machine learning methods, such as random forests, support vector machines and neural networks, based on the spatial correlation and scale invariance assumptions between precipitation and various meteorological elements. Many statistical models have been developed to describe the nonlinear relationship between coarse and fine resolutions. These models use machine learning methods to extract the characteristic relationship between coarse-resolution meteorological underlying surface factors and precipitation to predict the spatial distribution of precipitation at fine-resolution meteorological underlying surface factors. Statistical models based on point-to-point precipitation prediction usually ignore the spatial information between environmental factors and precipitation. Nowadays, deep learning methods have been shown to better capture high-dimensional and complex spatial features. Single image super-resolution models generate super-resolution images by learning the intrinsic features of images and the mapping relationship between coarse and fine resolutions. The most representative model is SRCNN. Considering meteorological downscaling as an image super-resolution task, researchers proposed DeepSD, which uses stacked SRCNNs to demonstrate its superior downscaling performance on Earth system models (ESMs), which is more advantageous than other statistical downscaling models. Subsequent studies have made significant progress in extreme precipitation and precipitation forecasting by integrating additional meteorological auxiliary information into neural network models. In addition, some scholars have scientifically explained how CNN-based downscaling models learn the physical possible relationship between large-scale atmospheric environment and extreme local precipitation events. More improved downscaling models based on super-resolution algorithms, such as FSRCNN, EPSCN, ResLap, etc., are superior to traditional statistical models in terms of computational efficiency, prediction accuracy and feature extraction. Related research mainly focuses on the downscaling of precipitation on annual and monthly scales, where the interaction between precipitation and meteorological factors is obvious. However, as the resolution of precipitation increases, its spatial discontinuity increases, making the downscaling of daily or hourly precipitation more challenging. At the same time, runoff simulation currently still uses coarse-resolution rain gauge observation data, meteorological bureau forecast data, and simple interpolation results (such as bilinear interpolation and Kriging interpolation). Studies have shown that such precipitation data will greatly limit the calculation accuracy.Due to the different focus scales of meteorological and hydrological research, few studies have considered satellite precipitation downscaling based on basin climate characteristics and applied it to runoff simulation. Summary of the invention

[0004] The present invention provides a system and method for fine-grained simulation of precipitation-runoff in mountainous watersheds, which can better capture the spatial precipitation characteristics of mountainous watersheds and simulate the precipitation-runoff propagation process in detail.

[0005] A method for fine-tuning precipitation-runoff simulation in a mountainous watershed according to the present invention comprises the following steps:

[0006] Step 1: Dataset and data preprocessing;

[0007] Step 2: Design a U-Net super-resolution model TA-UNet based on terrain attention mechanism as a precipitation downscaling model and train it;

[0008] Step 3: Accuracy verification;

[0009] Step 4: downscaling precipitation;

[0010] Step 5: Precipitation-runoff simulation.

[0011] Preferably, in step 1, the IMERG 0.1° precipitation data set every half hour in recent years and related topographic meteorological elements and ground rain gauge observation data are collected and collated; the satellite precipitation data and topography are resampled to 0.5° to establish a training data set.

[0012] As a preference, in step 2, the TA-UNet model is improved as follows:

[0013] 1) Add an additional upsampling module to the Attention U-Net model to upsample the coarse-resolution data to match the fine-resolution data grid;

[0014] 2) Incorporate multi-scale terrain data through attention gates; the attention gates serve as the self-attention feature layer between the encoder and decoder feature layers, and the attention feature layer between terrain and precipitation information in the TA-UNet model. The representation of the attention gates is as follows:

[0015] q att =ψ(σ1((w x x input +b x )+(w g g+b g )))+b ψ

[0016] α=σ2(q att (x input ,g;Θatt ))

[0017] σ1(x)=max(0,x)

[0018]

[0019] where q att represents the attention feature layer; σ1 and σ2 are ReLU and Sigmoid activation functions respectively, x var is the independent variable, c is a constant; Θ att Represents the trainable parameters of the neural network, i.e., ψ,w x ,b x ,w g and b g ;q att Normalized by sigmoid function to obtain the attention coefficient α, the attention mechanism is for the feature layer input x input The attention coefficient is g, which further weights the feature layer x input ;

[0020] 3) Replace the deconvolution layer used for upsampling with PixelShuffle.

[0021] Preferably, in step 2, the training process is as follows:

[0022] The dataset was used to train the 5x precipitation downscaling model until the error converged, the model was saved and a 0.02° downscaling prediction was performed. The PyTorch deep learning framework was used, the training batch size was set to 64, and the epochs were set to 50. The ADAM optimizer was used with a learning rate of 1e-4. The Exponential LR learning rate adjustment strategy was set, and the gamma value was set to 0.95.

[0023] Preferably, in step 4, specifically: using satellite precipitation data of 0.1° as model input, predicting 0.02° precipitation downscaled data.

[0024] Preferably, in step 5, specifically: coupling the precipitation downscaling model with the hydrodynamic model, and performing runoff forecast simulation on a 10m Cartesian coordinate grid.

[0025] As a preferred method, the hydrodynamic model is based on the two-dimensional depth-averaged shallow water equations, assuming that the pressure distribution of the water body is hydrostatic pressure, the channel slope is very small, the fluid density is uniform, and it has a free surface and a fixed bottom boundary; in matrix form, the shallow water equations SWEs are expressed as:

[0026]

[0027] Where t represents time; x and y are Cartesian coordinates; U is the vector of conservative physical variables; F and G are flux vectors along the x and y directions respectively; source vectors include precipitation R, riverbed slope S b Friction slope S f ; These vectors are represented as follows:

[0028]

[0029] Where h represents the water depth; u and v are the depth-averaged velocity components in the x and y directions, respectively; I represents the precipitation intensity; g is the gravitational acceleration; Z b is the slope height; S fx and S fy is the friction stress estimated by Manning's formula, expressed as n b is the Manning friction coefficient, set to 0.01.

[0030] As a preferred method, the hydrodynamic numerical simulation method adopts the Godunov finite volume method to solve the shallow water equation. The Godunov finite volume method can handle the exact or approximate Riemann problem at the boundary of the calculation unit; the HLLC approximate method is selected to handle the Riemann problem at the unit interface; the finite volume format derived from the shallow water equation is as follows:

[0031]

[0032] That is, the differential form of the time step after numerical discretization is:

[0033]

[0034] The superscript n represents the time step; the subscript i represents the grid unit number, and f E and f W corresponds to the flux to the east and west respectively, and g N and g S represent the flux to the north and south respectively, Δt is the time step selected by the CFL condition, and its expression is Δt = cflmin (Δt x ,Δt y ),in cfl∈[0,1]; the iterative process is performed uniformly over the overall time T, and the precipitation intensity R(I,t) is synchronized on the time scale based on the downscaled precipitation every consecutive half hour.

[0035] The present invention provides a system for fine-grained simulation of precipitation and runoff in a mountainous area, which adopts the above-mentioned method for fine-grained simulation of precipitation and runoff in a mountainous area.

[0036] The present invention proposes a system and method for fine-grained simulation of precipitation-runoff in mountainous watersheds, which is suitable for small watershed scales in mountainous areas. The present invention couples an improved downscaling model and a hydrodynamic model, which can better capture the spatial precipitation characteristics of mountainous watersheds and simulate the propagation process of rainfall-runoff in detail. By introducing an attention mechanism that considers terrain characteristics, a TA-UNet downscaling model based on a terrain attention mechanism is proposed, which can efficiently and conveniently downscale the IMERG satellite precipitation data to 0.02°. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 It is a flow chart of a method for fine-tuning precipitation-runoff simulation in a mountainous watershed in an embodiment;

[0038] Figure 2 It is a structural diagram of the TA-UNet model in the embodiment;

[0039] Figure 3 This is a comparison chart of the TA-UNet downscaling results in the embodiment and the Sichuan ground station verification;

[0040] Figure 4 This is a comparison chart of the TA-UNet downscaling results in the embodiment and the Jiangjiagou rain gauge verification. DETAILED DESCRIPTION

[0041] In order to further understand the content of the present invention, the present invention is described in detail in conjunction with the accompanying drawings and embodiments. It should be understood that the embodiments are only for explaining the present invention and are not intended to limit it.

[0042] Example

[0043] like Figure 1 As shown, this embodiment provides a method for fine-tuning simulation of precipitation and runoff in a mountainous area. Specifically, a method for fine-tuning simulation of precipitation and runoff in a mountainous area based on satellite precipitation downscaling and a hydrodynamic model is provided, which includes the following steps:

[0044] Step 1: Dataset and data preprocessing;

[0045] In step 1, the half-hourly IMERG 0.1° precipitation dataset in recent years and related topographic meteorological elements and ground rain gauge observation data are collected and sorted; the satellite precipitation data and terrain are resampled to 0.5° to establish a training dataset.

[0046] Step 2: Design a U-Net super-resolution model TA-UNet based on terrain attention mechanism as a precipitation downscaling model and train it;

[0047] In step 2, the improvements of the TA-UNet model are as follows:

[0048] 1) The Attention U-Net model was originally designed for classification tasks to identify spatially distributed features in images, but it does not have super-resolution capabilities. In order to adapt to downscaling tasks, an additional upsampling module (nearest neighbor interpolation) is added to the original Attention U-Net model to upsample the coarse-resolution data to match the fine-resolution data grid;

[0049] 2) Incorporating multi-scale terrain data through an attention gate; By applying the attention mechanism, the network’s ability to focus on spatial features in terrain and precipitation data is enhanced to improve the efficiency and accuracy of feature extraction; The modified attention gate can specifically extract features from terrain elements based on the attention mechanism.

[0050] The attention gate acts as a self-attention feature layer between the encoder and decoder feature layers in the TA-UNet model, as well as an attention feature layer between terrain and precipitation information. The representation of the attention gate is as follows:

[0051] q att =ψ(σ1((w x x input +b x )+(w g g+b g )))+b ψ

[0052] α=σ2(q att (x input ,g;Θ att ))

[0053] σ1(x)=max(0,x)

[0054]

[0055] where q att represents the attention feature layer, σ1 and σ2 are ReLU and Sigmoid activation functions respectively, Θ att Represents the trainable parameters of the neural network, i.e., ψ,w x ,b x ,w g and b g ;q att Normalized by sigmoid function to obtain the attention coefficient α, the attention mechanism is for the feature layer input x input The attention coefficient is g, which further weights the feature layer x input ;

[0056] Compared with the original Attention U-Net model structure, the attention gate, as a self-attention mechanism, has excellent performance in identifying the potential inherent features between feature layers;

[0057] 3) The “deconvolution layer” used for upsampling usually generates discontinuous data and causes a series of strange checkerboard artifacts. In this embodiment, the deconvolution layer used for upsampling is replaced with pixel reorganization PixelShuffle, so that the model can generate smoother and higher quality downscaled prediction data.

[0058] like Figure 2 Figure 1 shows a schematic diagram of the TA-UNet architecture. "Addition" refers to the element-by-element addition of tensors, "multiplication" refers to the Hadamard product of tensors, and "concatenation" refers to the concatenation of multiple tensors along a given dimension. "Downsampling" can reduce the spatial resolution of the input tensor, using the maximum pooling algorithm to achieve 2x downsampling. Similarly, "upsampling" uses a 5x nearest neighbor interpolation algorithm and a 2x PixelShuffle algorithm.

[0059] The training process is as follows:

[0060] The dataset was used to train the 5x precipitation downscaling model until the error converged, the model was saved and a 0.02° downscaling prediction was performed. The PyTorch deep learning framework was used, the training batch size was set to 64, and the epochs were set to 50. The ADAM optimizer was used with a learning rate of 1e-4. The Exponential LR learning rate adjustment strategy was set, and the gamma value was set to 0.95.

[0061] Step 3: Accuracy verification;

[0062] The performance of the TA-UNet model is better than that of the bilinear interpolation method, as well as benchmark models such as SRCNN, FSRCNN, and EPSCN. At the same time, through verification of observation data from Sichuan Meteorological Station and Jiangjiagou Ground Rain Gauge, the TA-UNet downscaling results have improved accuracy compared to the original satellite precipitation data.

[0063] Step 4: downscaling precipitation;

[0064] In step 4, specifically: use the satellite precipitation data of 0.1° as the model input and predict the precipitation downscaled data of 0.02°.

[0065] Step 5: precipitation-runoff simulation;

[0066] In step 5, specifically: couple the precipitation downscaling model with the hydrodynamic model and perform runoff forecast simulation on a 10m Cartesian coordinate grid.

[0067] The hydrodynamic model is based on the two-dimensional depth-averaged shallow water equations, assuming that the pressure distribution of the water body is hydrostatic pressure, the channel slope is very small, the fluid density is uniform, and it has a free surface and a fixed bottom boundary; in matrix form, the shallow water equations SWEs are expressed as:

[0068]

[0069] Where t represents time; x and y are Cartesian coordinates; U is the vector of conservative physical variables; F and G are flux vectors along the x and y directions respectively; source vectors include precipitation R, riverbed slope S b Friction slope S f ; These vectors are represented as follows:

[0070]

[0071] Where h represents the water depth; u and v are the depth-averaged velocity components in the x and y directions, respectively; I represents the precipitation intensity; g is the gravitational acceleration; Z b is the slope height; S fx and S fy is the friction stress estimated by Manning's formula, expressed as n b is the Manning friction coefficient, set to 0.01.

[0072] The hydrodynamic numerical simulation method uses the Godunov finite volume method to solve the shallow water equation. The Godunov finite volume method can handle the exact or approximate Riemann problem at the boundary of the calculation unit. In order to improve the efficiency and robustness of solving the Riemann problem, the HLLC approximation method is selected to handle the Riemann problem at the unit interface. The finite volume format derived from the shallow water equation is as follows:

[0073]

[0074] That is, the differential form of the time step after numerical discretization is:

[0075]

[0076] The superscript n represents the time step; the subscript i represents the grid unit number, and f E and f W corresponds to the flux to the east and west respectively, and g N and g S represent the flux to the north and south respectively, Δt is the time step selected by the CFL condition, and its expression is Δt = cflmin (Δt x ,Δt y ),in cfl∈[0,1]; the iterative process is performed uniformly over the overall time T, and the precipitation intensity R(I,t) is synchronized on the time scale based on the downscaled precipitation every consecutive half hour.

[0077] This embodiment provides a system for fine-grained simulation of precipitation and runoff in a mountainous watershed, which adopts the above-mentioned method for fine-grained simulation of precipitation and runoff in a mountainous watershed.

[0078] This embodiment proposes a refined precipitation-runoff simulation system and method combining a precipitation downscaling model and a hydrodynamic model. By introducing an attention mechanism that focuses on terrain features, the U-Net model (TA-UNet) based on the terrain attention mechanism realizes an efficient and convenient data preprocessing process of satellite precipitation downscaling from 0.1° to 0.02°. Compared with bilinear interpolation, TA-UNet improves by 34.0%, 29.9%, 4.4% and 0.6% in mean square error (MSE), mean absolute error (MAE), peak signal-to-noise ratio (PSNR) and structural similarity (SSIM), respectively. In addition, it performs well in meteorological forecast indicators such as VHI, VMI, VFAR and VCSI, as shown in Table 1.

[0079] Table 1 Performance evaluation index table of downscaling methods

[0080]

[0081]

[0082] By verifying the ground observation data of 156 national meteorological stations in Sichuan Province and 8 rain gauges in Jiangjiagou, TA-UNet can effectively downscale the IMERG GPM products and provide higher-precision local precipitation. Figure 3 As shown in the figure, the downscaling results of TA-UNet are compared with the verification of ground stations in Sichuan Province. Figure 4 The figure shows the comparison between the TA-UNet downscaling results and the Jiangjiagou rain gauge verification.

[0083] At the same time, the high-precision downscaled precipitation data showed a high sensitivity to runoff response in the rainfall-runoff simulation of Jiangjiagou. This refined method better captures the spatial precipitation characteristics of mountainous watersheds and more accurately simulates the propagation process of rainfall-runoff, providing an innovative and efficient forecasting strategy for hydrological management.

[0084] The present invention and its embodiments are described schematically above, and the description is not restrictive. The drawings show only one embodiment of the present invention, and the actual structure is not limited thereto. Therefore, if a person skilled in the art is inspired by it and designs a structural method and an embodiment similar to the technical solution without creativity without departing from the purpose of the invention, they shall all fall within the protection scope of the present invention.

Claims

1. A method for fine-tuning precipitation-runoff simulation in a mountainous watershed, characterized by: The following steps are involved: Step 1: Dataset and data preprocessing; Step 2: Design a U-Net super-resolution model TA-UNet based on terrain attention mechanism as a precipitation downscaling model and train it; Step 3: Accuracy verification; Step 4: downscaling precipitation; Step 5: Precipitation-runoff simulation.

2. The method for fine-tuning precipitation-runoff simulation in a mountainous watershed according to claim 1, characterized in that: In step 1, the half-hourly IMERG 0.1° precipitation dataset in recent years and related topographic meteorological elements and ground rain gauge observation data are collected and sorted; the satellite precipitation data and terrain are resampled to 0.5° to establish a training dataset.

3. The method for fine-tuning precipitation-runoff simulation in a mountainous watershed according to claim 2, characterized in that: In step 2, the improvements of the TA-UNet model are as follows: 1) Add an additional upsampling module to the Attention U-Net model to upsample the coarse-resolution data to match the fine-resolution data grid; 2) Incorporating multi-scale terrain data through attention gates; The attention gate acts as a self-attention feature layer between the encoder and decoder feature layers in the TA-UNet model, as well as an attention feature layer between terrain and precipitation information. The representation of the attention gate is as follows: q att =ψ(σ1((w x x input +b x )+(w g g+b g )))+b ψ α=σ2(q att (x input ,g;Θ att )) σ1(x var )=max(0,x var ) where q att Represents the attention feature layer; σ1 and σ2 are ReLU and Sigmoid activation functions respectively, x var is the independent variable, c is a constant; Θ att Represents the trainable parameters of the neural network, i.e., ψ,w x ,b x ,w g and b g ;q att Normalized by sigmoid function to obtain the attention coefficient α, the attention mechanism is for the feature layer input x input The attention coefficient is g, which further weights the feature layer x input ; 3) Replace the deconvolution layer used for upsampling with PixelShuffle.

4. The method for fine-tuning precipitation-runoff simulation in a mountainous watershed according to claim 3, characterized in that: In step 2, the training process is as follows: The dataset was used to train the 5x precipitation downscaling model until the error converged, the model was saved and a 0.02° downscaling prediction was performed. The PyTorch deep learning framework was used, the training batch size was set to 64, and the epochs were set to 50. The ADAM optimizer was used with a learning rate of 1e-4. The Exponential LR learning rate adjustment strategy was set, and the gamma value was set to 0.

95.

5. The method for fine-tuning precipitation-runoff simulation in a mountainous watershed according to claim 4, characterized in that: In step 4, specifically: use the satellite precipitation data of 0.1° as the model input and predict the precipitation downscaled data of 0.02°.

6. A method for fine-tuning precipitation-runoff simulation in a mountainous watershed according to claim 5, characterized in that: In step 5, specifically: couple the precipitation downscaling model with the hydrodynamic model and perform runoff forecast simulation on a 10m Cartesian coordinate grid.

7. The method for fine-tuning precipitation-runoff simulation in a mountainous watershed according to claim 6, characterized in that: The hydrodynamic model is based on the two-dimensional depth-averaged shallow water equations, assuming that the pressure distribution of the water body is hydrostatic pressure, the channel slope is very small, the fluid density is uniform, and it has a free surface and a fixed bottom boundary; in matrix form, the shallow water equations SWEs are expressed as: Where t represents time; x and y are Cartesian coordinates; U is the vector of conservative physical variables; F and G are flux vectors along the x and y directions respectively; source vectors include precipitation R, riverbed slope S b Friction slope S f ; These vectors are represented as follows: Where h represents the water depth; u and v are the depth-averaged velocity components in the x and y directions, respectively; I represents the precipitation intensity; g is the gravitational acceleration; Z b is the slope height; S fx and S fy is the friction stress estimated by Manning's formula, expressed as n b is the Manning friction coefficient, set to 0.

01.

8. The method for fine-tuning precipitation-runoff simulation in a mountainous watershed according to claim 7, characterized in that: The hydrodynamic numerical simulation method uses the Godunov finite volume method to solve the shallow water equation. The Godunov finite volume method can handle the exact or approximate Riemann problem at the boundary of the calculation unit; the HLLC approximate method is selected to handle the Riemann problem at the unit interface; the finite volume format derived from the shallow water equation is as follows: That is, the differential form of the time step after numerical discretization is: The superscript n represents the time step; the subscript i represents the grid unit number, and f E and f W corresponds to the flux to the east and west respectively, and g N and g S represent the flux to the north and south respectively, Δt is the time step selected by the CFL condition, and its expression is Δt = cflmin (Δt x ,Δt y ),in The iterative process is performed uniformly over the overall time T, and the precipitation intensity R(I,t) is synchronized in time based on the downscaled precipitation of each consecutive half-hour.

9. A system for fine-tuning precipitation and runoff simulation in a mountainous watershed, characterized by: It adopts a method for fine-tuning precipitation-runoff simulation in a mountainous watershed as described in any one of claims 1-8.

Images