Snow melting simulation and risk early warning method and system

Abstract

The invention provides a snow melting simulation and risk early warning method and system, and belongs to the technical field of flood early warning. The method comprises the following steps: collecting and preprocessing multi-source weather, geographic information and accumulated snow characteristic data of a target area; a snow melting simulation model considering multiple factors is constructed, a CEEMDAN algorithm and an LSTM network are utilized to predict the snow melting amount, the runoff volume and the soil water content, and a multi-head attention mechanism is introduced; dividing a space-time grid and carrying out hourly simulation calculation to obtain a space-time distribution result; setting a risk assessment index system, and calculating a snow melting risk index; judging whether the risk index exceeds a threshold value or not, outputting an early warning result, matching a risk level when the risk index exceeds the threshold value, drawing a risk partition map, and marking a low-risk or safe state if the risk index does not exceed the threshold value. According to the snow melting simulation and risk early warning method and system, accurate prediction and timely early warning of the snow melting flood risk are achieved, the disaster prevention and reduction capacity is effectively improved, and potential loss and harm caused by the snow melting flood are reduced.

Description

Technical Field

[0001] The present invention relates to the field of flood warning technology, and in particular to a snowmelt simulation and risk warning method and system. Background Art

[0002] Snowmelt floods are a common natural disaster, especially in high-altitude mountainous areas. The melting of snow can cause serious flood disasters. Accurately predicting the risk of snowmelt floods is of great significance for disaster prevention and mitigation.

[0003] Prior art methods for early warning of snowmelt flood risks have made some progress. For example, patent publication number CN111080030B discloses a snowmelt flood prediction method, device, electronic device, and storage medium. This method determines the snowmelt flood risk level by acquiring rain and snowfall distribution data and temperature distribution data for a target area, extracting precipitation and temperature characteristics data. This method combines current temperature and precipitation conditions to assess and provide early warning of snowmelt flood risk over a longer period of time, providing a visual aid for flood monitoring, early warning, and emergency response in areas prone to snowmelt floods.

[0004] Furthermore, patent publication number CN116341208A discloses a snowmelt flood emergency warning method based on snowmelt simulation. This method generates a three-dimensional snow model by acquiring snow and environmental characteristics of a pre-defined area. This model, combined with weather forecast information, simulates snowmelt to generate a snowmelt flood risk index. This method quantifies risk through model simulation, improving the accuracy and timeliness of early warnings.

[0005] However, existing technologies still have some shortcomings. For example, some methods still have room for improvement in data fusion and model building, failing to fully leverage the combined advantages of multi-source data. In terms of risk assessment and early warning response, the reliability of some methods' warning basis is low, failing to fully reflect the extent of snowmelt flood risk. Furthermore, the early warning information dissemination and emergency response processes are not perfect, making it difficult to achieve accurate and timely early warnings and effective emergency response. Summary of the Invention

[0006] The purpose of the present invention is to provide a snowmelt simulation and risk warning method and system, which can achieve accurate prediction and timely warning of snowmelt flood risks, effectively improve disaster prevention and mitigation capabilities, and reduce the potential losses and hazards caused by snowmelt floods.

[0007] To achieve the above objectives, the present invention provides a snowmelt simulation and risk warning method, comprising the following steps:

[0008] Collect multi-source meteorological data, geographic information data and snow characteristics data of the target area, and pre-process them to obtain pre-processed data;

[0009] A snowmelt simulation model was established by comprehensively considering the energy input of solar radiation, air temperature, and wind speed to the snow surface, as well as the heat conduction and phase change processes within the snow layer.

[0010] The pre-processed data is used as the input of the snowmelt simulation model, and the time-space grid is divided to perform hourly simulation calculations to obtain the spatiotemporal distribution results of snowmelt, runoff and soil moisture content.

[0011] A snowmelt risk assessment index system was established. Hourly snowmelt rate data was extracted from snowmelt simulation records. The first, second, and third snowmelt rates were calculated. The snowmelt risk index was calculated by combining the obtained spatiotemporal distribution results of snowmelt volume, runoff volume, and soil moisture content.

[0012] Get the preset risk threshold, determine whether the snowmelt risk index exceeds the preset threshold, and output the warning result.

[0013] Preferably, the multi-source meteorological data includes temperature, precipitation, wind speed and humidity; the geographic information data includes topography, landform and soil type; and the snow characteristic data includes snow depth, density and water content.

[0014] Preferably, preprocessing includes data cleaning, spatial interpolation, data standardization and data fusion.

[0015] Preferably, a snowmelt simulation model is established by comprehensively considering the energy input of solar radiation, air temperature, and wind speed to the snow surface, as well as the heat conduction and phase change process inside the snow layer. The specific operations are as follows:

[0016] Establish the snowmelt physical process equations:

[0017]

[0018] Among them, Q in represents the total energy input to the snow surface, R represents solar radiation, and T a Indicates air temperature, V w represents wind speed, α and β represent empirical coefficients, T s represents the temperature inside the snow layer, t represents time, z represents the depth of the snow layer, K represents the thermal conductivity, L represents the latent heat of phase change, θ represents the water content of the snow layer, Q out represents the longwave radiation loss from the snow surface;

[0019] The preprocessed data is decomposed into multiple IMF components using the CEEMDAN algorithm;

[0020] Each IMF component is predicted through an LSTM network. The input of the LSTM network is the decomposed IMF component, and the output is the predicted snowmelt, runoff, and soil moisture content.

[0021] Introducing a multi-head attention mechanism to enhance the model's attention to different features;

[0022] The specific formula of the snowmelt simulation model is as follows:

[0023]

[0024] in, represents the amount of snowmelt at time t and spatial position (i, j), γ represents the weight coefficient, represents the comprehensive data at time t-1 and spatial position (i, j), i and j represent the rows and columns of the spatial grid, and P represents the set of equations for the snowmelt physical process.

[0025] Preferably, the preprocessed data is used as the input of the snowmelt simulation model, and the time-space grid is divided to perform hourly simulation calculations to obtain the time-space distribution results of snowmelt, runoff, and soil moisture. The specific operations are as follows:

[0026] Divide the space-time grid, each grid cell corresponds to a specific spatial location (i, j);

[0027] For each time step t, the meteorological data, geographic information data, and snow characteristics data of the current time step are extracted from the preprocessed data and then input into the CEEMDAN-LSTM-ATTE model to obtain the predicted value of the IMF component;

[0028] The predicted values of the IMF components are combined with the calculation results of the snowmelt physical process equations to obtain the predicted values of snowmelt, runoff and soil moisture.

[0029] Store the predicted value as the spatiotemporal distribution result of the current time step t;

[0030] Repeat the above steps until the simulation calculation of all time steps is completed and the snowmelt amount is obtained. Runoff and soil moisture content The spatiotemporal distribution results.

[0031] Preferably, the snowmelt risk assessment indicator system includes snowmelt rate, runoff peak, water accumulation depth, terrain slope and soil saturation.

[0032] Preferably, the calculation formula of the snowmelt risk index is as follows:

[0033]

[0034] in, represents the snowmelt risk index, R m represents the snowmelt rate, Q p represents the peak runoff value, D a Indicates the depth of water accumulation, Ss Indicates soil saturation, S t represents the terrain slope, ω1, ω2, ω3, ω4, ω5, and ω6 represent weight coefficients;

[0035] The snowmelt rate is calculated as follows:

[0036]

[0037] Among them, R m1 、R m2 、R m3 Respectively represent the first snowmelt rate, the second snowmelt rate, and the third snowmelt rate, ω m1 、ω m2 、ω m3 Represent the weight coefficients of the first snowmelt rate, the second snowmelt rate, and the third snowmelt rate respectively.

[0038] Snowmelt is usually a gradual phenomenon, and the snowmelt rate and related hydrological processes can vary significantly over different time periods.

[0039] First 12 hours: In the early stage of snowmelt, the snow begins to melt and the amount of snowmelt gradually increases. The snowmelt rate in this stage is relatively low, but it has an important impact on the subsequent snowmelt process.

[0040] 12 hours to 36 hours: As factors such as temperature change, the snowmelt rate usually increases significantly and reaches a peak. This stage contributes the most to runoff.

[0041] 36 to 72 hours: The snowmelt process enters a stable stage, the snowmelt rate tends to be stable, and the snow accumulation gradually decreases. This stage has an important impact on soil moisture content and groundwater recharge.

[0042] In flood warnings, it is important to focus on the contributions of different stages of the snowmelt process to runoff and flood risk. Early snowmelt increases soil moisture, mid-term snowmelt leads to peak runoff, and late snowmelt leads to persistent flooding.

[0043] For water resource management, understanding the temporal distribution of the snowmelt process helps to rationally allocate water resources, prevent flood disasters, and ensure the sustainable use of water resources.

[0044] The calculation formula for the first snowmelt rate is as follows:

[0045]

[0046] in, represents the amount of snowmelt at time t+12 and spatial position (i, j);

[0047] The second snowmelt rate calculation formula is as follows:

[0048]

[0049] in, represents the amount of snowmelt at time t+36 and spatial position (i, j);

[0050]

[0051] in, represents the amount of snowmelt at time t+72 and spatial position (i, j);

[0052] The formula for calculating peak runoff is as follows:

[0053]

[0054] The calculation formula for soil saturation is as follows:

[0055]

[0056] Among them, S max Indicates the maximum water holding capacity of the soil.

[0057] Preferably, when the snowmelt risk index exceeds a preset threshold, the corresponding risk level is matched according to the extent of the excess, and a snowmelt risk zoning map is drawn on the GIS map, marking the risk level and warning signal of each area;

[0058] When the snowmelt risk index does not exceed the preset threshold, a snowmelt risk zoning map is drawn on the GIS map, marking each area as low risk or safe.

[0059] The present invention also provides a snowmelt simulation and risk warning system, comprising:

[0060] Data acquisition and preprocessing module: used to collect multi-source meteorological data, geographic information data and snow characteristics data of the target area, and preprocess them to obtain preprocessed data;

[0061] Snowmelt simulation model building module: This module is used to comprehensively consider the energy input to the snow surface from solar radiation, air temperature, and wind speed, as well as the heat conduction and phase change processes within the snow layer, to establish a snowmelt simulation model.

[0062] Simulation calculation module: used to use the pre-processed data as the input of the snowmelt simulation model, divide the time and space grids to perform hourly simulation calculations, and obtain the spatiotemporal distribution results of snowmelt, runoff and soil moisture content;

[0063] Risk Assessment and Early Warning Classification Module: This module is used to set a snowmelt risk assessment indicator system. Hourly snowmelt rate data is extracted from snowmelt simulation records as the first, second, and third snowmelt rates. Combined with the spatiotemporal distribution results of snowmelt volume, runoff volume, and soil moisture content obtained from the simulation calculation module, the snowmelt risk index is calculated.

[0064] Warning issuance and emergency response module: used to obtain the preset risk threshold, determine whether the snowmelt risk index exceeds the preset threshold, and output the warning result. If it exceeds, the corresponding risk level is matched according to the excess range, and a snowmelt risk zoning map is drawn on the GIS map, marking the risk level and warning signal of each area; if it does not exceed, a snowmelt risk zoning map is drawn on the GIS map, marking each area as low risk or safe.

[0065] Preferably, the snowmelt simulation model construction module includes a physical equation group establishment unit, a data decomposition unit, a model prediction unit and an attention mechanism introduction unit, which are respectively used to establish the physical equation group of the snowmelt process, decompose the preprocessed data into multiple IMF components through the CEEMDAN algorithm, predict each IMF component through the LSTM network, and introduce a multi-head attention mechanism.

[0066] Therefore, the present invention adopts the above-mentioned snowmelt simulation and risk warning method and system, and the beneficial technical effects are as follows:

[0067] (1) This invention integrates multi-source meteorological data, geographic information data, and snow characteristics data. Compared with relying solely on meteorological data or a single data source, the data dimension is richer and can more comprehensively characterize the driving factors and basic conditions of the snowmelt process, providing a more solid data foundation for accurate simulation and early warning. For example, by combining meteorological factors such as temperature, precipitation, and wind speed with geographic information such as topography and soil type, it can more accurately analyze the snowmelt characteristics and risk differences in different regions.

[0068] (2) The physical equations for the snowmelt process, established based on the principles of energy balance and phase change, comprehensively consider the effects of solar radiation, air temperature, wind speed, etc. on the energy input of the snow surface, as well as the heat conduction and phase change processes within the snow layer. Compared with some simple models that only consider the single influence of some meteorological factors on snowmelt, this equation group can more accurately describe the physical mechanism of the snowmelt process, thereby improving the scientificity and accuracy of the simulation.

[0069] Advantages of the CEEMDAN algorithm: Using the CEEMDAN algorithm to decompose snowmelt time series data yields multiple IMF components and residual terms. Compared to traditional decomposition methods like EEMD, CEEMDAN effectively avoids the problem of mode mixing by adding white noise of varying amplitudes and performing ensemble empirical mode decomposition, followed by least squares fitting to remove the noise. The decomposition results are more stable and reliable, enabling more accurate extraction of the different frequency features in snowmelt data, providing a more favorable foundation for subsequent modeling and prediction.

[0070] Combining an LSTM network with a multi-head attention mechanism: An LSTM network is constructed for each IMF component and residual term, and a multi-head attention mechanism is introduced. LSTM networks can effectively handle long-term dependencies in time series data, while the multi-head attention mechanism further enhances the model's focus on different feature dimensions and time steps, enabling the model to automatically learn and highlight important feature information. Compared to a single LSTM model or other traditional time series models such as ARIMA, this model has stronger feature expression capabilities and predictive performance, and can more accurately capture the complex dynamics of the snowmelt process.

[0071] (3) A risk assessment indicator system including snowmelt rate, runoff peak, water accumulation depth, terrain slope and soil saturation was established to comprehensively assess snowmelt risk from multiple perspectives. Compared with the assessment method based on a single indicator such as snowmelt volume or runoff volume, it can more comprehensively and deeply reflect the risk level and potential impact of snowmelt floods, and provide a more sufficient basis for early warning decisions.

[0072] Accurate early warning release and emergency response: The system can accurately determine the risk level based on whether the snowmelt risk index exceeds the preset threshold, and intuitively display the risk zoning through GIS maps. At the same time, it uses multiple communication channels to accurately push early warning information and preset emergency response processes and measures under different risk levels, realizing the timely and accurate release of early warning information and effective emergency response. Compared with the existing technology where the early warning information release and emergency response processes are not perfect, the system has significantly improved the prevention and response capabilities of snowmelt flood disasters. BRIEF DESCRIPTION OF THE DRAWINGS

[0073] Figure 1 This is a flow chart of a snowmelt simulation and risk warning method of the present invention;

[0074] Figure 2 A flow chart for snowmelt simulation;

[0075] Figure 3 This is a structural diagram of a snowmelt simulation and risk warning system of the present invention. DETAILED DESCRIPTION

[0076] The technical solution of the present invention is further described below with reference to the accompanying drawings and embodiments.

[0077] Unless otherwise defined, technical or scientific terms used in the present invention shall have the same meaning as commonly understood by one of ordinary skill in the art to which the present invention belongs.

[0078] Example 1

[0079] like Figure 1 As shown, the present invention provides a snowmelt simulation and risk warning method, comprising the following steps:

[0080] 1. Taking a certain mountainous area as the target area, multi-source meteorological data, geographic information data and snow characteristics data of the target area are collected, and preprocessed to obtain preprocessed data.

[0081] 1.1 Experimental data

[0082] Meteorological data: temperature, precipitation, wind speed, and humidity, with a time resolution of 1 hour. The experimental period is from January 1, 2022 to April 30, 2022, a total of 121 days. The data come from 10 meteorological stations in the region.

[0083] Geographic information data: topography (altitude, slope, aspect), landforms (mountains, plains, river valleys) and soil types (sand, loam, clay), with a spatial resolution of 30 meters. The data are derived from geographic information databases and field surveys.

[0084] Snow characteristic data: snow depth, density and water content, with a time resolution of 1 day. A total of 90 sets of data were obtained during the experimental period. The data came from snow monitoring stations and field sampling.

[0085] 1.2. Data preprocessing

[0086] Data cleaning: remove outliers and erroneous records in meteorological data; supplement missing soil type information in geographic information data; smooth snow characteristics data and eliminate outliers with obvious deviations.

[0087] Spatial interpolation: The weather station data is interpolated using the Kriging method to generate spatial distribution data for the entire area.

[0088] Data normalization: All data are normalized to the range [0,1] for easy model input.

[0089] Data fusion: The processed meteorological, geographic information and snow characteristics data are integrated to form a complete data set.

[0090] Second, a snowmelt simulation model was established, taking into account the energy input to the snow surface from solar radiation, air temperature, and wind speed, as well as the heat conduction and phase change processes within the snow layer. The preprocessed data was used as input to the snowmelt simulation model, which was divided into a spatiotemporal grid and simulated hourly to obtain the spatiotemporal distribution of snowmelt, runoff, and soil moisture.

[0091] Establishment of a set of physical equations: Based on the principles of energy balance and phase change, a set of physical equations for the snowmelt process is established, including equations for energy input to the snow surface from solar radiation, air temperature, and wind speed, as well as equations for heat conduction and phase change within the snow layer.

[0092]

[0093] Among them, Q in represents the total energy input to the snow surface, R represents solar radiation, and T a Indicates air temperature, V w represents wind speed, α and β represent empirical coefficients, T s represents the temperature inside the snow layer, t represents time, z represents the depth of the snow layer, K represents the thermal conductivity, L represents the latent heat of phase change, θ represents the water content of the snow layer, Q out Represents the longwave radiation loss from the snow surface.

[0094] like Figure 2 The figure shows a snowmelt simulation flow chart. In this embodiment, the CEEMDAN algorithm is used to decompose the snowmelt time series data into five IMF components and one residual term. The CEEMDAN algorithm adds white noise of varying amplitudes and performs ensemble empirical mode decomposition (EEMD), ultimately removing the added noise through least squares fitting to obtain more stable IMF components. For example, white noise with an amplitude of 0.2 is added to the snowmelt time series data, and 100 EEMD decompositions are performed to obtain multiple IMF component sets. The noise effect is then removed through fitting, ultimately resulting in five stable IMF components and one residual term.

[0095] An LSTM network is constructed for each IMF component and residual term for prediction. The LSTM network has an input layer of 10 neurons, a hidden layer of 50 neurons, and an output layer of 1 neuron. A multi-head attention mechanism is introduced with four attention heads to enhance the model's focus on different features. The multi-head attention mechanism calculates attention weights for different feature combinations of the input data, enabling the model to simultaneously focus on multiple important feature dimensions. For example, for each IMF component, the multi-head attention mechanism calculates its importance weight at different time steps. These weights are then applied to the hidden layer output of the LSTM network, highlighting important time step features and improving prediction accuracy.

[0096] The specific formula of the snowmelt simulation model is as follows:

[0097]

[0098] in, represents the amount of snowmelt at time t and spatial position (i, j), γ represents the weight coefficient, represents the comprehensive data at time t-1 and spatial position (i, j), i and j represent the rows and columns of the spatial grid, and P represents the set of equations for the snowmelt physical process.

[0099] Model training: Data from 2017-2021 was used as the training set, and data from 2022 was used as the test set. Training was performed for 100 epochs, with a learning rate of 0.001 and a mean squared error loss function. During training, batch gradient descent was used with a batch size of 32. The validation loss was calculated after each epoch. Training was terminated early if the validation loss did not decrease for five consecutive epochs to prevent overfitting.

[0100] 3. Establish a snowmelt risk assessment index system, extract hourly snowmelt rate data from snowmelt simulation records as the first snowmelt rate, second snowmelt rate and third snowmelt rate, and calculate the snowmelt risk index by combining the spatiotemporal distribution results of snowmelt volume, runoff volume and soil moisture content.

[0101] Spatiotemporal grid division: The target area is divided into a 1000×1000 spatiotemporal grid, with each grid cell being 5km×5km in size.

[0102] Hourly simulation calculation: The preprocessed data is input into the model and hourly simulation calculation is performed to obtain the spatiotemporal distribution results of snowmelt, runoff and soil moisture content.

[0103] The risk assessment indicator system is set as follows: snowmelt rate, runoff peak, water accumulation depth, terrain slope and soil saturation.

[0104] Snowmelt risk index calculation: Calculate the snowmelt risk index for each grid cell based on the indicator system;

[0105] The calculation formula for the snowmelt risk index is as follows:

[0106]

[0107] in, represents the snowmelt risk index, R m represents the snowmelt rate, Q p represents the peak runoff value, D a Indicates the depth of water accumulation, S s Indicates soil saturation, S trepresents the terrain slope, ω1, ω2, ω3, ω4, ω5, and ω6 represent weight coefficients;

[0108] Snowmelt is usually a gradual phenomenon, and the snowmelt rate and related hydrological processes can vary significantly over different time periods.

[0109] First 12 hours: In the early stage of snowmelt, the snow begins to melt and the amount of snowmelt gradually increases. The snowmelt rate in this stage is relatively low, but it has an important impact on the subsequent snowmelt process.

[0110] 12 hours to 36 hours: As factors such as temperature change, the snowmelt rate usually increases significantly and reaches a peak. This stage contributes the most to runoff.

[0111] 36 to 72 hours: The snowmelt process enters a stable stage, the snowmelt rate tends to be stable, and the snow accumulation gradually decreases. This stage has an important impact on soil moisture content and groundwater recharge.

[0112] In flood warnings, it is important to focus on the contributions of different stages of the snowmelt process to runoff and flood risk. Early snowmelt increases soil moisture, mid-term snowmelt leads to peak runoff, and late snowmelt leads to persistent flooding.

[0113] For water resource management, understanding the temporal distribution of the snowmelt process helps to rationally allocate water resources, prevent flood disasters, and ensure the sustainable use of water resources.

[0114] The calculation formula for the first snowmelt rate is as follows:

[0115]

[0116] in, represents the amount of snowmelt at time t+12 and spatial position (i, j);

[0117] The second snowmelt rate calculation formula is as follows:

[0118]

[0119] in, represents the amount of snowmelt at time t+36 and spatial position (i, j);

[0120]

[0121] in, represents the amount of snowmelt at time t+72 and spatial position (i, j);

[0122] The formula for calculating peak runoff is as follows:

[0123]

[0124] The calculation formula for soil saturation is as follows:

[0125]

[0126] Among them, S max Indicates the maximum water holding capacity of the soil.

[0127] 4. Obtain the preset risk threshold, determine whether the snowmelt risk index exceeds the preset threshold, and output the warning result.

[0128] Warning results: On March 15, 2022, the model predicted that the snowmelt risk index in some areas of the target area would be high in the next 48 hours, exceeding the preset threshold. The system automatically drew a snowmelt risk zoning map on the GIS map, marked high-risk areas, and pushed warning information to relevant departments and residents via SMS and APP.

[0129] The root mean square error (RMSE) of the prediction results and the actual observation values of the proposed method is 0.25, the mean absolute error (MAE) is 0.18, and the coefficient of determination (R 2 ) is 0.89; the RMSE of the multiple linear regression is 0.45, MAE is 0.36, R 2 is 0.72; the RMSE of the support vector machine is 0.38, MAE is 0.30, and R 2 This indicates that the proposed method can more accurately simulate the snowmelt process and provide a more reliable basis for snowmelt flood risk warning.

[0130] like Figure 3 As shown, a snowmelt simulation and risk warning system includes:

[0131] Data acquisition and preprocessing module: used to collect multi-source meteorological data, geographic information data and snow characteristics data of the target area, and preprocess them to obtain preprocessed data;

[0132] Snowmelt simulation model building module: This module is used to comprehensively consider the energy input to the snow surface from solar radiation, air temperature, and wind speed, as well as the heat conduction and phase change processes within the snow layer, to establish a snowmelt simulation model.

[0133] Simulation calculation module: used to use the pre-processed data as the input of the snowmelt simulation model, divide the time and space grids to perform hourly simulation calculations, and obtain the spatiotemporal distribution results of snowmelt, runoff and soil moisture content;

[0134] Risk Assessment and Early Warning Classification Module: This module is used to set a snowmelt risk assessment indicator system. Hourly snowmelt rate data is extracted from snowmelt simulation records as the first, second, and third snowmelt rates. Combined with the spatiotemporal distribution results of snowmelt volume, runoff volume, and soil moisture content obtained from the simulation calculation module, the snowmelt risk index is calculated.

[0135] Warning issuance and emergency response module: used to obtain the preset risk threshold, determine whether the snowmelt risk index exceeds the preset threshold, and output the warning result. If it exceeds, the corresponding risk level is matched according to the excess range, and a snowmelt risk zoning map is drawn on the GIS map, marking the risk level and warning signal of each area; if it does not exceed, a snowmelt risk zoning map is drawn on the GIS map, marking each area as low risk or safe.

[0136] The specific implementation of the early warning release and emergency response module is as follows:

[0137] Risk level classification.

[0138] Level I (extremely high risk): When the snowmelt risk index exceeds the preset threshold by 30% or more, it indicates that the area faces an extremely high risk of snowmelt floods, which may cause serious flood disasters.

[0139] Level II (high risk): When the snowmelt risk index exceeds the preset threshold of 15% to 30%, it means that the area has a high risk of snowmelt floods and major floods may occur.

[0140] Level III (Medium Risk): When the snowmelt risk index exceeds the preset threshold by 5% to 15%, it indicates that the area has a moderate risk of snowmelt floods and a certain degree of flooding may occur.

[0141] Level IV (low risk): When the snowmelt risk index exceeds the preset threshold but is within 5%, it means that the risk of snowmelt flooding in the area is low, but vigilance is still needed.

[0142] Draw a snowmelt risk zoning map.

[0143] Map data preparation: Obtain GIS map data of the target area, including geographic information (such as topography, landforms, water systems, etc.), administrative boundaries, and settlement distribution information, so as to accurately mark risk areas and related details on the map.

[0144] Risk area coloring: Based on the calculated snowmelt risk index for each grid cell, each grid cell is colored according to the risk level classification criteria described above. For example, Level I (extremely high risk) areas are represented by red, Level II (high risk) areas by orange, Level III (medium risk) areas by yellow, Level IV (low risk) areas by blue, and areas that do not exceed the threshold are represented by green.

[0145] Mark risk levels and warning signals: On the colored map, clearly mark the risk level of each risk area, such as "extremely high risk (level I)", "high risk (level II)", etc.

[0146] Map output and display: The drawn snowmelt risk zoning map will be output through geographic information system software to generate maps of different scales to meet decision-making needs at different levels.

[0147] Early warning information release and emergency response.

[0148] Warning information push: When the risk index exceeds the threshold, the system automatically matches the corresponding warning information template based on the risk level and accurately pushes the warning information to relevant departments (such as emergency management, water conservancy, and meteorological departments) and residents in the risk area through multiple communication channels such as text messages, emails, mobile apps, and social media. Warning information includes the location of the risk area, risk level, expected impact time, possible disaster type, and corresponding preventive measures.

[0149] Emergency Response Activation: When an early warning is issued, the system automatically triggers the pre-set emergency response process. For each level of risk, the corresponding emergency plan is activated. The emergency plan is implemented based on the actual situation in each region.

[0150] Warning cancellation and summary evaluation: When the snowmelt process ends, the warning will be cancelled based on the actual flood situation and risk changes.

[0151] When the snowmelt risk index does not exceed the preset threshold, a snowmelt risk zoning map is drawn on the GIS map, marking each area as low risk or safe. The specific implementation is as follows:

[0152] Map data preparation: Similar to the above process when the threshold is exceeded, obtain GIS map data of the target area, including geographic information, administrative boundaries, and settlement distribution information.

[0153] Area coloring and labeling: Based on the calculated snowmelt risk index, mark areas that do not exceed the threshold as low risk or safe. You can color these areas green and label them as "low risk" or "safe" on the map.

[0154] Map output and display: Maps are also output through professional geographic information system software.

[0155] The snowmelt simulation model construction module includes a physical equation group establishment unit, a data decomposition unit, a model prediction unit and an attention mechanism introduction unit, which are respectively used to establish the physical equation group of the snowmelt process, decompose the preprocessed data into multiple IMF components through the CEEMDAN algorithm, predict each IMF component through the LSTM network, and introduce the multi-head attention mechanism.

[0156] The data acquisition and preprocessing module includes a data cleaning unit, a spatial interpolation unit, a data standardization unit and a data fusion unit, which are used to clean, interpolate, standardize and fuse the collected data respectively.

[0157] The simulation calculation module is specifically used to input the preprocessed data as model input, divide the time and space grids to perform hourly simulation calculations, and obtain the time and space distribution results of snowmelt, runoff and soil moisture content.

[0158] The snowmelt risk assessment indicator system in the risk assessment and early warning classification module includes snowmelt rate, runoff peak, water accumulation depth, terrain slope and soil saturation.

[0159] It is worth noting that the contents not elaborated in detail in the present invention are all prior art and are well known to those skilled in the art.

[0160] Therefore, the present invention adopts the above-mentioned snowmelt simulation and risk warning method and system, and constructs an accurate snowmelt simulation model by integrating multi-source meteorological data, geographic information data and snow characteristics data, comprehensively considering the influence of multiple factors such as solar radiation, air temperature, wind speed on the energy input of the snow surface, as well as the heat conduction and phase change process inside the snow layer. The model uses the advanced CEEMDAN algorithm to decompose the data, combines the LSTM network and multi-head attention mechanism for prediction, and improves the accuracy and reliability of the simulation. On this basis, a comprehensive snowmelt risk assessment indicator system is set up, and risk assessment is carried out from multiple dimensions such as snowmelt rate, runoff peak, water accumulation depth, terrain slope and soil saturation to calculate the snowmelt risk index. By comparing with the preset risk threshold, accurate prediction and timely warning of snowmelt flood risks can be achieved. When the risk index exceeds the threshold, the system can match the corresponding risk level according to the degree of excess, and draw a snowmelt risk zoning map on the GIS map, marking the risk level and warning signal of each area. At the same time, it can accurately push warning information through multiple communication channels, preset emergency response processes and measures under different risk levels, and ensure that the corresponding plan is automatically activated after the warning is issued, effectively improving disaster prevention and mitigation capabilities, and reducing the potential losses and hazards caused by snowmelt floods.

[0161] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit the same. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that they can still modify or replace the technical solutions of the present invention with equivalents, and these modifications or equivalent replacements cannot cause the modified technical solutions to deviate from the spirit and scope of the technical solutions of the present invention.

Claims

1. A snowmelt simulation and risk warning method, characterized in that: The following steps are involved: Collect multi-source meteorological data, geographic information data and snow characteristics data of the target area, and pre-process them to obtain pre-processed data; A snowmelt simulation model was established by comprehensively considering the energy input of solar radiation, air temperature, and wind speed to the snow surface, as well as the heat conduction and phase change processes within the snow layer. The pre-processed data is used as the input of the snowmelt simulation model, and the time-space grid is divided to perform hourly simulation calculations to obtain the spatiotemporal distribution results of snowmelt, runoff and soil moisture content. A snowmelt risk assessment index system was established. Hourly snowmelt rate data was extracted from snowmelt simulation records. The first, second, and third snowmelt rates were calculated. The snowmelt risk index was calculated by combining the obtained spatiotemporal distribution results of snowmelt volume, runoff volume, and soil moisture content. Get the preset risk threshold, determine whether the snowmelt risk index exceeds the preset threshold, and output the warning result.

2. A snowmelt simulation and risk warning method according to claim 1, characterized in that: Multi-source meteorological data include temperature, precipitation, wind speed and humidity; geographic information data include topography, landform and soil type; snow characteristics data include snow depth, density and moisture content.

3. A snowmelt simulation and risk warning method according to claim 1, characterized in that: Preprocessing includes data cleaning, spatial interpolation, data standardization and data fusion.

4. A snowmelt simulation and risk warning method according to claim 1, characterized in that: A snowmelt simulation model was established by comprehensively considering the energy input of solar radiation, air temperature, and wind speed to the snow surface, as well as the heat conduction and phase change processes within the snow layer. The specific operations are as follows: Establish the snowmelt physical process equations: Among them, Q in represents the total energy input to the snow surface, R represents solar radiation, and T a Indicates air temperature, V w represents wind speed, α and β represent empirical coefficients, T s represents the temperature inside the snow layer, t represents time, z represents the depth of the snow layer, K represents the thermal conductivity, L represents the latent heat of phase change, θ represents the water content of the snow layer, Q out represents the longwave radiation loss from the snow surface; The preprocessed data is decomposed into multiple IMF components using the CEEMDAN algorithm; Each IMF component is predicted through an LSTM network. The input of the LSTM network is the decomposed IMF component, and the output is the predicted snowmelt, runoff, and soil moisture content. Introducing a multi-head attention mechanism to enhance the model's attention to different features; The specific formula of the snowmelt simulation model is as follows: in, represents the amount of snowmelt at time t and spatial position (i, j), γ represents the weight coefficient, represents the comprehensive data at time t-1 and spatial position (i, j), i and j represent the rows and columns of the spatial grid, and P represents the set of equations for the snowmelt physical process.

5. A snowmelt simulation and risk warning method according to claim 4, characterized in that: The preprocessed data is used as the input of the snowmelt simulation model. The time-space grid is divided and the hourly simulation calculation is performed to obtain the spatiotemporal distribution results of snowmelt, runoff, and soil moisture. The specific operation is as follows: Divide the space-time grid, each grid cell corresponds to a specific spatial location (i, j); For each time step t, the meteorological data, geographic information data, and snow characteristics data of the current time step are extracted from the preprocessed data and then input into the CEEMDAN-LSTM-ATTE model to obtain the predicted value of the IMF component; The predicted values of the IMF components are combined with the calculation results of the snowmelt physical process equations to obtain the predicted values of snowmelt, runoff and soil moisture. Store the predicted value as the spatiotemporal distribution result of the current time step t; Repeat the above steps until the simulation calculation of all time steps is completed and the snowmelt amount is obtained. Runoff and soil moisture content The spatiotemporal distribution results.

6. A snowmelt simulation and risk warning method according to claim 5, characterized in that: The snowmelt risk assessment indicator system includes snowmelt rate, runoff peak, water accumulation depth, terrain slope and soil saturation.

7. A snowmelt simulation and risk warning method according to claim 6, characterized in that: The calculation formula for the snowmelt risk index is as follows: in, represents the snowmelt risk index, R m represents the snowmelt rate, Q p represents the peak runoff value, D a Indicates the depth of water accumulation, S s Indicates soil saturation, S t represents the terrain slope, ω1, ω2, ω3, ω4, ω5, and ω6 represent weight coefficients; The snowmelt rate is calculated as follows: Among them, R m1 、R m2 、R m3 Respectively represent the first snowmelt rate, the second snowmelt rate, and the third snowmelt rate, ω m1 、ω m2 、ω m3 Represent the weight coefficients of the first snowmelt rate, the second snowmelt rate, and the third snowmelt rate respectively.

8. The snowmelt simulation and risk warning method according to claim 1, characterized in that: When the snowmelt risk index exceeds the preset threshold, the corresponding risk level is matched according to the extent of the excess, and a snowmelt risk zoning map is drawn on the GIS map, marking the risk level and warning signal of each area; When the snowmelt risk index does not exceed the preset threshold, a snowmelt risk zoning map is drawn on the GIS map, marking each area as low risk or safe.

9. A snowmelt simulation and risk warning system, characterized in that: include: The data acquisition and preprocessing module is used to collect multi-source meteorological data, geographic information data and snow characteristics data of the target area, and preprocess them to obtain preprocessed data; The snowmelt simulation model construction module is used to comprehensively consider the energy input of solar radiation, air temperature, wind speed to the snow surface, as well as the heat conduction and phase change process within the snow layer to establish a snowmelt simulation model; The simulation calculation module is used to use the pre-processed data as the input of the snowmelt simulation model, divide the time and space grids to perform hourly simulation calculations, and obtain the spatiotemporal distribution results of snowmelt, runoff and soil moisture content; The risk assessment and early warning classification module is used to set the snowmelt risk assessment indicator system. Hourly snowmelt rate data is extracted from the snowmelt simulation records as the first, second, and third snowmelt rates. Combined with the spatiotemporal distribution results of snowmelt volume, runoff volume, and soil moisture content obtained by the simulation calculation module, the snowmelt risk index is calculated. The early warning and emergency response module is used to obtain the preset risk threshold, determine whether the snowmelt risk index exceeds the preset threshold, and output the early warning result. If it exceeds the threshold, the corresponding risk level is matched according to the excess degree, and a snowmelt risk zoning map is drawn on the GIS map, marking the risk level and early warning signal of each area; If it does not exceed the limit, a snowmelt risk zoning map will be drawn on the GIS map, marking each area as low risk or safe.

10. A snowmelt simulation and risk warning system according to claim 9, characterized in that: The snowmelt simulation model construction module includes a physical equation group establishment unit, a data decomposition unit, a model prediction unit and an attention mechanism introduction unit, which are respectively used to establish the physical equation group of the snowmelt process, decompose the preprocessed data into multiple IMF components through the CEEMDAN algorithm, predict each IMF component through the LSTM network, and introduce the multi-head attention mechanism.

Images