Snow melting runoff prediction method, snow melting runoff prediction device, medium and electronic equipment

Abstract

The invention discloses a snow melting runoff prediction method, a snow melting runoff prediction device, a medium and electronic equipment, and relates to the field of data processing. The accuracy of runoff prediction can be improved. The snow melting runoff prediction method comprises the following steps: acquiring hydrological data of each drainage basin, and extracting block data of a target drainage basin from the hydrological data based on satellite remote sensing data; dividing the block data into a plurality of level blocks, and determining clustering features of each level block, the level blocks being used for indicating precipitation levels of each region in a target drainage basin; and inputting the clustering features of each level block into a snow-melting runoff prediction model, and predicting the snow-melting runoff of the target drainage basin to obtain runoff prediction data of the target drainage basin.

Description

Technical Field

[0001] The present application relates to the field of data processing technology, and in particular to a snowmelt runoff prediction method, snowmelt runoff prediction device, medium and electronic equipment. Background Art

[0002] As the impact of climate change continues to expand, the stability of the climate system continues to decline, which in turn has a profound impact on the global water cycle system. The variation in the hydrological cycle caused by climate change will lead to the reconfiguration of water resources in terms of temporal and spatial distribution and changes in the total amount of water resources, which may increase the frequency and intensity of extreme climate and hydrological events in some basins, thereby exacerbating the risk of frequent floods and droughts. As an important part of the cryosphere, snow has a significant impact on the surface energy balance due to its unique physical properties, such as high reflectivity, high latent heat and low thermal conductivity. Snowmelt runoff in snowy areas is an important part of surface runoff and has a significant impact on the runoff generation process. In recent years, global climate change has had a huge impact on the runoff generation and convergence process in snowy areas.

[0003] With the rapid progress of computer technology, deep learning technology with artificial neural networks as the core has been widely used in various fields. Data-driven machine learning models are gradually becoming a key tool for runoff simulation and forecasting. Therefore, the accuracy of machine learning models in runoff prediction needs to be improved urgently. Summary of the invention

[0004] The present application provides a snowmelt runoff prediction method, a snowmelt runoff prediction device, a medium and an electronic device, which can improve the prediction accuracy of snowmelt runoff.

[0005] In a first aspect, the present application provides a snowmelt runoff prediction method, comprising:

[0006] Obtaining hydrological data of each river basin, and extracting block data of the target river basin from the hydrological data based on satellite remote sensing data;

[0007] Dividing the block data into a plurality of level blocks, and determining a clustering feature of each level block, wherein the level block is used to indicate the precipitation level of each area in the target watershed;

[0008] The clustering characteristics of each level block are input into the snowmelt runoff prediction model, the snowmelt runoff of the target watershed is predicted, and the runoff prediction data of the target watershed is obtained.

[0009] According to the snowmelt runoff prediction method in the present embodiment, the block data of the target watershed to be predicted is extracted from the hydrological data of the watershed, and the block data is divided according to the precipitation level to obtain the level blocks of different precipitation levels, and then the snowmelt runoff is predicted based on the corresponding clustering characteristics of the precipitation levels of the level blocks in the target watershed, which can reduce the input data; and, compared with the method of directly using the hydrological data as input for prediction, the present embodiment integrates the clustering characteristics of various precipitation levels, and predicts according to the influence of the precipitation level on the snowmelt runoff, which can improve the prediction accuracy.

[0010] In a second aspect, the present application provides a snowmelt runoff prediction device, comprising:

[0011] A data acquisition module, used to acquire hydrological data of each river basin, and extract block data of the target river basin from the hydrological data based on satellite remote sensing data;

[0012] A data partitioning module, used for partitioning the block data into a plurality of level blocks and determining a clustering feature of each level block, wherein the level block is used for indicating the precipitation level of each area in the target watershed;

[0013] The runoff prediction module is used to input the clustering characteristics of each level block into the snowmelt runoff prediction model, predict the snowmelt runoff of the target watershed, and obtain the runoff prediction data of the target watershed.

[0014] In a third aspect, the present application provides an electronic device, the electronic device comprising a memory and one or more processors. The memory stores one or more computer programs, the computer programs comprising instructions, and when the instructions are executed by the processor, the electronic device can execute the snowmelt runoff prediction method as in the first aspect.

[0015] In a fourth aspect, the present application provides a computer-readable storage medium having instructions stored therein, which, when executed on an electronic device, enables the electronic device to execute the snowmelt runoff prediction method as in the first aspect.

[0016] In a fifth aspect, the present application provides a computer program product, which, when executed on an electronic device, enables the electronic device to execute the snowmelt runoff prediction method as described in the first aspect.

[0017] It can be understood that the beneficial effects that can be achieved by the snowmelt runoff prediction device, electronic device, computer-readable storage medium, and computer program product provided above can refer to the beneficial effects in the first aspect and will not be repeated here. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1A schematic diagram of a snowmelt runoff prediction method provided in an embodiment of the present application;

[0019] Figure 2 A schematic diagram of the model structure in the snowmelt runoff prediction method provided in an embodiment of the present application;

[0020] Figure 3 A schematic diagram of another model structure in the snowmelt runoff prediction method provided in an embodiment of the present application;

[0021] Figure 4 A schematic diagram of the structure of a snowmelt runoff prediction device provided in an embodiment of the present application;

[0022] Figure 5 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0023] In order to facilitate the clear description of the technical solutions of the embodiments of the present application, in the embodiments of the present application, the words "first", "second" and the like are used to distinguish the same items or similar items with substantially the same functions and effects. For example, the first chip and the second chip are only used to distinguish different chips, and their order is not limited. Those skilled in the art can understand that the words "first", "second" and the like do not limit the quantity and execution order, and the words "first", "second" and the like do not necessarily limit the difference. It should be noted that in the embodiments of the present application, the words "exemplary" or "for example" are used to indicate examples, illustrations or explanations. Any embodiment or design described as "exemplary" or "for example" in the present application should not be interpreted as being more preferred or more advantageous than other embodiments or design solutions. Specifically, the use of words such as "exemplary" or "for example" is intended to present related concepts in a specific way. In the embodiments of the present application, "at least one" refers to one or more, and "more" refers to two or more.

[0024] It should be noted that the “at…” in the embodiments of the present application may be the instant when a certain situation occurs, or may be a period of time after the occurrence of a certain situation, and the embodiments of the present application do not specifically limit this.

[0025] The implementation of this embodiment will be described in detail below with reference to the accompanying drawings.

[0026] The present embodiment provides a snowmelt runoff prediction method. By way of example, the snowmelt runoff prediction method can be applied to various electronic devices such as computers (PCs), tablet computers, virtual reality / augmented reality devices, wearable devices, industrial computers, and vehicle computers; it can also be applied to servers, cloud computing, server clusters, etc., and the present embodiment does not make any special limitations on this.

[0027] Figure 1A schematic flow chart of a snowmelt runoff prediction method provided in an embodiment of the present application is shown.

[0028] like Figure 1 As shown, the snowmelt runoff prediction method may include the following steps:

[0029] Step 101: Acquire hydrological data of each river basin, and extract block data of the target river basin from the hydrological data based on satellite remote sensing data.

[0030] Hydrological data refers to raster data of hydrological elements of each river basin. Exemplarily, the hydrological data includes historical meteorological raster data of the river basin, forecast raster data for the future forecast period, multi-source remote sensing data, and digital elevation model; the historical meteorological raster data includes: precipitation, air temperature, temperature, air pressure, wind speed, potential evapotranspiration, and actual evapotranspiration; the multi-source remote sensing data includes vegetation type, ground type, soil moisture, surface temperature, and snow thickness; the forecast raster data includes daily precipitation, air temperature, and snow thickness for the future forecast period.

[0031] The required hydrological data are collected through open data platforms, such as geospatial data cloud, meteorological information center and other platforms. Specifically, the historical data that need to be collected include: local meteorological grid data of the basin, including the measured data of precipitation, temperature, humidity, air pressure, wind speed, potential evapotranspiration, and actual evapotranspiration in the basin, multi-source remote sensing data including vegetation type, ground type, soil moisture, surface temperature, and snow thickness, including all available historical daily runoff and flood runoff data before the forecast date, and digital elevation model data (DEM) of the study basin; future data include: daily scale precipitation, temperature, and snow thickness forecast grid data within the future forecast period provided by meteorological forecast.

[0032] The geographical location of the target basin can be determined based on satellite remote sensing data, and then the block data within the geographical location can be extracted from the hydrological data. Block data refers to the raster data of the target basin, which can include historical meteorological raster data, such as precipitation, temperature, air pressure, wind speed, potential evapotranspiration, actual evapotranspiration, etc.; raster data of ground information, such as vegetation type, ground type, soil moisture, surface temperature, snow thickness; forecast raster data, such as daily scale precipitation, temperature, and snow thickness forecast raster data within the future forecast period provided by meteorological forecasts.

[0033] In an exemplary implementation, after extracting the block data of the target watershed from the raster data, the block data can be normalized. The snow depth, precipitation, temperature and runoff data in the block data are not uniform in terms of dimension and magnitude. Therefore, in order to make the model converge quickly during the simulation process, a linear function normalization method can be used to reduce the weight effect of numerical differences on variable inputs.

[0034] Step 102: Divide the block data into a plurality of level blocks, and determine the clustering characteristics of each level block, wherein the level blocks are used to indicate the precipitation level of each area in the target watershed.

[0035] Exemplarily, precipitation levels can be divided into five levels: dry, semi-dry, balanced, prone to runoff, and extremely prone to runoff.

[0036] In this embodiment, a hybrid convolutional Mamba model is used to segment the block data to obtain multiple level blocks.

[0037] Before using the Hybrid Convolutional Mamba (HC-Mamba) model to divide the block data, the HC-Mamba model needs to be trained. Specifically, the historical hydrological data of the basin and the multi-source remote sensing data are obtained; the historical hydrological data of the basin is divided based on the multi-source remote sensing data to obtain multiple level blocks, wherein each level block corresponds to a precipitation level; the historical meteorological raster data and the multi-source remote sensing data are input into the hybrid convolutional Mamba model, and the hybrid convolutional Mamba model is trained, so that the segmentation result of the hybrid convolutional Mamba model on the historical meteorological raster data of the basin is the same as the level block.

[0038] Among them, the multi-source remote sensing data includes the historical precipitation of each area in the basin. Based on the historical precipitation, the area in the basin can be divided to obtain multiple level blocks divided according to the precipitation level. Then, the historical hydrological data and multi-source remote sensing data are used as input, and the divided level blocks are used as expected outputs to train the hybrid convolutional Mamba model, so that the segmentation result of the hybrid convolutional Mamba model for the historical hydrological raster data of the basin is the same as the level block or the difference between the two is less than the preset value, and the training is completed.

[0039] The structure of HC-Mamba can be described as the patch embedding layer, HC-SSM module and patch merging layer. In the patch embedding layer of HC-Mamba, the input block data x∈R (H×W×N) Divide into non-overlapping grids of size 4×4. H is the number of vertical grids of the block data; W is the number of horizontal grids of the block data; N is the number of channels of the block data, which is determined by the number of types of block data. Map the original grid data to C size (usually C = 32N) to get the embedded grid data It is then input into the backbone network of HC-Mamba, which consists of four stages. After the output of the first three stages, a patch merging layer is used to reduce the size of the input features and increase the number of channels. The four stages use 2, 4, 2, and 2 HC-SSM modules respectively, and each stage has C, 2C, 4C, and 8C channels respectively.

[0040] The HC-SSM module is the core module of HC-Mamba. Figure 2 The structural diagram of the HC-SSM module is shown in FIG. Figure 2 As shown in the figure, the input of the HC-SSM module is evenly divided into two sub-inputs X1 and X2 of equal size, which are input to the state space model (SSM) branch and the HC-Conv branch respectively. In the SSM branch, the input features are first dimensionally enhanced by linear mapping, and then enter a convolution layer using depthwise separable convolution, which improves the localization processing capability of features by grouping them while maintaining the dimension. Next, the SiLU function is used as the activation function and input to the SS2D module, and nonlinear transformation is introduced to enhance the expression capability of the model. Finally, the processed features are remapped to the original feature space to obtain the output of the SSM branch. In the HC-Conv branch, a dilated convolution is introduced to expand the learning range of the convolution kernel.

[0041] The SS2D module is the core of the SSM branch of the HC-SSM module, which contains three key parts: scan expansion, selective state-space sequence model (S6), and scan merging. Scan expansion decomposes the input image into independent sequences of up, down, left, and right. This process ensures a wide spatial coverage of information, thereby achieving the capture of multi-directional features. The S6 module is a component of the selective state-space model (SSM), which is specifically designed to handle sequence modeling tasks and can effectively capture long-range dependencies. The S6 module accurately identifies and extracts useful information by selecting the parameters of the state-space model through a selection mechanism, while filtering out irrelevant parts. The input feature format of this module is [B, L, D] (B is the batch size; L is the sequence length; D is the feature dimension). The features are first transformed by the linear layer, and then the update and output equations in the state-space model are applied to finally generate the output features; scan merging will operate and reconfigure these transformed sequences to generate output images that match the original input image.

[0042] The SSM model is a classic continuous system model. Figure 3 As shown, the SSM model includes a structured state space sequence model (S4) and a Mamba model. The system takes the input function or sequence as Mapping to output The specific formula is as follows:

[0043]

[0044] in, is the state matrix, is the projection parameter, and the process is as shown in the figure below. In this figure, the symbol D represents a skip connection, which can be understood as a transformed residual connection. Therefore, the part of the graph excluding D is usually designated as a state space model (SSM).

[0045] In order to adapt these continuous systems to deep learning applications, S4 and Mamba discretized the systems. Specifically, a time scale parameter or step size Δ was introduced, and fixed discretization rules such as zero-order hold (ZOH) were used to convert parameters A and B into discrete parameters.

[0046]

[0047] After discretization, the computation of the state space model can be achieved by linear recursion:

[0048]

[0049] It can also be achieved through global convolution:

[0050]

[0051] in, is a structured convolution kernel, and L represents the length of the input sequence x.

[0052] The HC-Mamba model uses a comprehensive weighted loss function As a model effect evaluation indicator, it evaluates the overlap, boundary accuracy and similarity between the segmentation result and the true result. This function combines mIoU loss, Dice loss and boundary loss. The mIoU loss function evaluates the overlap of the segmented area, the Dice loss function measures the similarity of the segmented area, and the boundary loss function focuses on the accuracy of the segmentation boundary.

[0053]

[0054] in, Represent mIoU loss, Dice loss and boundary loss respectively, w mIoU 、w Dice 、w Boundary Then is the corresponding weight coefficient. P and B G represent the boundary point sets of the predicted segmentation result and the true result, respectively, and d(p,q) is the Euclidean distance between point p and point q. In order to comprehensively evaluate the index, the HC-Mamba model is trained using historical basin measured data and multi-source remote sensing data to divide the data. The trained model is then used to divide the input future weather forecast grid data of the target basin into five level blocks: dry, semi-dry, balanced, easy to produce runoff, and extremely easy to produce runoff according to the historical data characteristics. Clustering features are extracted in different level blocks. Clustering features of each level block can be extracted by various clustering algorithms, such as K-means, etc., which are not specifically limited in this embodiment.

[0055] Step 103: Input the clustering features of each level block into the snowmelt runoff prediction model, predict the snowmelt runoff of the target watershed, and obtain the runoff prediction data of the target watershed.

[0056] The runoff prediction data may be the daily runoff of the target watershed. The snowmelt runoff prediction model may include various machine learning models, and illustratively, the snowmelt runoff prediction model includes random forest, support vector machine, multi-layer perceptron, extreme gradient random tree, and Bayesian ridge regression.

[0057] The clustering features of each of the hierarchical blocks are respectively input into random forest, support vector machine, multi-layer perceptron, extreme gradient random tree, and Bayesian ridge regression to obtain multiple prediction results; then the multiple prediction results are weighted to obtain the runoff prediction data of the target basin.

[0058] Random forest (RF) This model makes predictions by combining multiple decision trees. Each decision tree is built based on a randomly selected feature subset. As the number of decision trees in the random forest increases, the generalization error of the model gradually stabilizes.

[0059] eXtreme Gradient Boosting (XGB) is an efficient gradient boosting algorithm that is extended from the decision tree model and improved based on Boosting. This model can use multi-core CPUs for parallel computing, significantly improving the prediction speed and accuracy.

[0060] The regression algorithm of Support Vector Machine (SVM) is called SVR. This model solves regression problems based on the "VC" dimension theory and risk minimization principles. The parameter selection and model structure are relatively flexible and it has the ability to process high-dimensional data.

[0061] Multilayer Perceptron (MLP) is a feedforward neural network model, which consists of multiple neuron layers including input layer, output layer and hidden layer. Multiple hidden layers can be set. The model uses methods such as Adam optimizer and stochastic gradient descent to continuously optimize the internal parameters of the model to minimize the loss, and updates the position and weight of each neuron through back propagation until the loss function is minimized or the accuracy requirement is met.

[0062] Bayesian Ridge Regression (BRR) is a machine learning regression algorithm based on Bayesian theory. This method has a high utilization rate of data samples and solves problems such as multicollinearity and maximum likelihood estimation overfitting in the linear regression process. The BRR algorithm automatically introduces a regularization term in the prediction process, and avoids overfitting in the maximum likelihood estimation process by obtaining the posterior distribution of the parameters, thereby making the prediction more accurate.

[0063] After obtaining the prediction results of each model, the final runoff prediction data, that is, the daily runoff data of the target basin, can be obtained by weighted summation.

[0064] In an exemplary embodiment, three commonly used statistical indicators are introduced into the prediction results of each model: Nash-Sutcliffe Efficiency Coefficient (NSE), Root Mean Square Error (RMSE) and Pbias, which are used to evaluate the prediction effects of the above five machine learning models under different forecast periods, so that the prediction results with high accuracy are used as the final runoff prediction data. Different models may have different prediction effects under different forecast periods, so the prediction results of multiple models can be combined under different forecast periods to select the prediction results with higher accuracy as the final runoff prediction data.

[0065] Exemplarily, the watershed size and historical daily runoff data of the target watershed are extracted from the block data of the target watershed, and the longest lag time of the target watershed is determined based on the watershed size and the historical daily runoff data; the clustering characteristics of each of the hierarchical blocks and the longest lag time are input into the snowmelt runoff prediction model, and the snowmelt runoff of the target watershed is predicted to obtain the runoff prediction data of the target watershed.

[0066] The basin size refers to the size of the target basin area, and the longest lag time refers to the time from when water flows into the target basin to when it flows out. The longest lag time of the target basin can be calculated by analyzing the correlation between historical daily runoff data and basin size. In this embodiment, the longest lag time of the target basin and the clustering feature are used as inputs to predict the runoff, which can further improve the accuracy of the prediction.

[0067] Furthermore, this embodiment also provides a snowmelt runoff prediction device, which can be used to execute the above-mentioned snowmelt runoff prediction method.

[0068] like Figure 4 As shown, the snowmelt runoff prediction device 400 may include: a data acquisition module 410, used to acquire hydrological data of each basin, and extract block data of the target basin from the hydrological data based on satellite remote sensing data; a data division module 420, used to divide the block data into multiple level blocks, and determine the clustering characteristics of each level block, wherein the level block is used to indicate the precipitation level of each area in the target basin; a runoff prediction module 430, used to input the clustering characteristics of each level block into the snowmelt runoff prediction model, predict the snowmelt runoff of the target basin, and obtain the runoff prediction data of the target basin.

[0069] In an exemplary embodiment, the data partitioning module 420 is specifically used to: use a hybrid convolutional Mamba model to partition the block data to obtain multiple level blocks.

[0070] In an exemplary embodiment, the snowmelt runoff prediction device 400 also includes: a model training module, used to obtain historical hydrological data of the basin, and multi-source remote sensing data; based on the multi-source remote sensing data, the historical hydrological data of the basin is divided to obtain a plurality of level blocks, wherein each level block corresponds to a precipitation level; the historical hydrological data and the multi-source remote sensing data are input into a hybrid convolutional Mamba model, and the hybrid convolutional Mamba model is trained so that the segmentation result of the hybrid convolutional Mamba model for the historical hydrological raster data of the basin is the same as the level block.

[0071] In an exemplary embodiment, the hydrological data includes historical meteorological raster data of the basin, forecast raster data for the future forecast period, multi-source remote sensing data, and a digital elevation model; the historical meteorological raster data includes: precipitation, air temperature, temperature, air pressure, wind speed, potential evapotranspiration, actual evapotranspiration; the multi-source remote sensing data includes vegetation type, ground type, soil moisture, surface temperature, and snow thickness; the forecast raster data includes daily precipitation, air temperature, and snow thickness for the future forecast period.

[0072] In an exemplary embodiment, the snowmelt runoff prediction model includes random forest, support vector machine, multi-layer perceptron, extreme gradient random tree, and Bayesian ridge regression.

[0073] In an exemplary embodiment, the runoff prediction module 430 is specifically used to: input the clustering features of each of the level blocks into a random forest, a support vector machine, a multi-layer perceptron, an extreme gradient random tree, and a Bayesian ridge regression, respectively, to obtain multiple prediction results; and perform weighted processing on the multiple prediction results to obtain the runoff prediction data of the target watershed.

[0074] In an exemplary embodiment, the runoff prediction module 430 is further specifically used to: extract the watershed size and historical daily runoff data of the target watershed from the block data of the target watershed, and determine the longest lag time of the target watershed based on the watershed size and the historical daily runoff data; input the clustering characteristics of each of the hierarchical blocks and the longest lag time into the snowmelt runoff prediction model, predict the snowmelt runoff of the target watershed, and obtain the runoff prediction data of the target watershed.

[0075] The specific details of each module or unit in the above-mentioned snowmelt runoff prediction device have been described in detail in the corresponding snowmelt runoff prediction method, so they will not be repeated here.

[0076] The embodiment of the present application also provides an electronic device, Figure 5 A schematic diagram of the structure of an electronic device suitable for implementing the embodiments of the present disclosure is shown. Figure 5 The electronic device 600 shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present disclosure.

[0077] like Figure 5 As shown, electronic device 600 includes a central processing unit (CPU) 601, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 602 or a program loaded from a storage part 608 into a random access memory (RAM) 603. In RAM 603, various programs and data required for system operation are also stored. CPU 601, ROM 602 and RAM 603 are connected to each other via a bus 604. An input / output (I / O) interface 605 is also connected to bus 604.

[0078] The following components are connected to the I / O interface 605: an input section 606 including a keyboard, a mouse, etc.; an output section 607 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker, etc.; a storage section 608 including a hard disk, etc.; and a communication section 609 including a network interface card such as a LAN card, a modem, etc. The communication section 609 performs communication processing via a network such as the Internet. A drive 610 is also connected to the I / O interface 605 as needed. A removable medium 611, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 610 as needed, so that a computer program read therefrom is installed into the storage section 608 as needed.

[0079] In particular, according to an embodiment of the present disclosure, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present disclosure includes a computer program product, which includes a computer program carried on a computer-readable storage medium, and the computer program includes a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from the network through the communication part 609, and / or installed from the removable medium 611. When the computer program is executed by the central processing unit (CPU) 601, the above-mentioned functions defined in the embodiment of the present application are executed.

[0080] It should be noted that the computer-readable medium shown in the present disclosure may be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present disclosure, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, device or device. In the present disclosure, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, in which a computer-readable program code is carried. This propagated data signal may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. Computer-readable signal media may also be any computer-readable medium other than computer-readable storage media, which may send, propagate or transmit a program for use by or in conjunction with an instruction execution system, apparatus or device. The program code contained on the computer-readable medium may be transmitted using any appropriate medium, including but not limited to: wireless, wire, optical cable, RF, etc., or any suitable combination of the above.

[0081] The flow charts and block diagrams in the accompanying drawings illustrate the possible architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present disclosure. In this regard, each box in the flow chart or block diagram can represent a module, a program segment, or a part of a code, and the above-mentioned module, program segment, or a part of a code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram or flow chart, and the combination of the boxes in the block diagram or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0082] The units involved in the embodiments described in the present disclosure may be implemented by software or hardware, and the units described may also be arranged in a processor. The names of these units do not, in some cases, constitute limitations on the units themselves.

[0083] As another aspect, the present application also provides a computer-readable medium, which may be included in the electronic device described in the above embodiment; or may exist independently without being assembled into the electronic device. The above computer-readable medium carries one or more programs, and the one or more programs include instructions, and when the instructions are executed by the electronic device, the electronic device implements the method described in the above embodiment.

[0084] It should be noted that, although several modules or units of the equipment for action execution are mentioned in the above detailed description, this division is not mandatory. In fact, according to the embodiments of the present application, the features and functions of two or more modules or units described above can be embodied in one module or unit. On the contrary, the features and functions of one module or unit described above can be further divided into being embodied by multiple modules or units.

[0085] The above contents are only specific implementation methods of the present application, but the protection scope of the present application is not limited thereto. Any changes or substitutions within the technical scope disclosed in the present application shall be included in the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims.

Claims

1. A snowmelt runoff prediction method, characterized in that: include: Obtaining hydrological data of each river basin, and extracting block data of the target river basin from the hydrological data based on satellite remote sensing data; Dividing the block data into a plurality of level blocks, and determining a clustering feature of each level block, wherein the level block is used to indicate the precipitation level of each area in the target watershed; The clustering characteristics of each level block are input into the snowmelt runoff prediction model, the snowmelt runoff of the target watershed is predicted, and the runoff prediction data of the target watershed is obtained.

2. The snowmelt runoff prediction method according to claim 1, characterized in that: The step of dividing the block data into a plurality of level blocks comprises: The block data is segmented using a hybrid convolutional Mamba model to obtain multiple level blocks.

3. The snowmelt runoff prediction method according to claim 2, characterized in that: Before segmenting the block data using the hybrid convolutional Mamba model, the method further includes: Obtain historical hydrological data of the basin, as well as multi-source remote sensing data; Based on the multi-source remote sensing data, the historical hydrological data of the basin is divided to obtain a plurality of grade blocks, wherein each grade block corresponds to a precipitation grade; The historical hydrological data and the multi-source remote sensing data are input into a hybrid convolutional Mamba model, and the hybrid convolutional Mamba model is trained so that the segmentation result of the hybrid convolutional Mamba model on the historical hydrological raster data of the watershed is the same as the hierarchical block.

4. The snowmelt runoff prediction method according to claim 1, characterized in that: The hydrological data include historical meteorological grid data of the basin, forecast grid data for the future forecast period, multi-source remote sensing data, and digital elevation model; the historical meteorological grid data include: precipitation, air temperature, temperature, air pressure, wind speed, potential evapotranspiration, actual evapotranspiration; the multi-source remote sensing data include vegetation type, ground type, soil moisture, surface temperature, and snow thickness; the forecast grid data include daily precipitation, air temperature, and snow thickness for the future forecast period.

5. The snowmelt runoff prediction method according to claim 1, characterized in that: The snowmelt runoff prediction model includes random forest, support vector machine, multi-layer perceptron, extreme gradient random tree, and Bayesian ridge regression.

6. The snowmelt runoff prediction method according to claim 5, characterized in that: The clustering features of each level block are input into the snowmelt runoff prediction model to predict the snowmelt runoff of the target watershed, thereby obtaining the runoff prediction data of the target watershed, including: The clustering features of each of the hierarchical blocks are respectively input into a random forest, a support vector machine, a multi-layer perceptron, an extreme gradient random tree, and a Bayesian ridge regression to obtain multiple prediction results; The multiple prediction results are weighted to obtain the runoff prediction data of the target watershed.

7. The snowmelt runoff prediction method according to claim 1, characterized in that: The clustering features of each level block are input into the snowmelt runoff prediction model to predict the snowmelt runoff of the target watershed, thereby obtaining the runoff prediction data of the target watershed, including: Extract the watershed size and historical daily runoff data of the target watershed from the block data of the target watershed, Determining a maximum lag time of a target watershed based on the watershed size and the historical daily runoff data; The clustering characteristics of each of the graded blocks and the longest lag time are input into a snowmelt runoff prediction model to predict the snowmelt runoff of a target watershed, thereby obtaining runoff prediction data for the target watershed.

8. A snowmelt runoff prediction device, characterized in that: include: A data acquisition module, used to acquire hydrological data of each river basin, and extract block data of the target river basin from the hydrological data based on satellite remote sensing data; A data partitioning module, used for partitioning the block data into a plurality of level blocks and determining a clustering feature of each level block, wherein the level block is used for indicating the precipitation level of each area in the target watershed; The runoff prediction module is used to input the clustering characteristics of each level block into the snowmelt runoff prediction model, predict the snowmelt runoff of the target watershed, and obtain the runoff prediction data of the target watershed.

9. A computer-readable storage medium storing a computer program, wherein when the computer program is executed by a processor, the processor executes the snowmelt runoff prediction method according to any one of claims 1 to 7.

10. An electronic device, characterized in that: It includes a processor and a memory, wherein one or more computer programs are stored in the memory, and the one or more computer programs include instructions. When the instructions are executed by the electronic device, the electronic device executes the snowmelt runoff prediction method described in any one of claims 1 to 7.

Images