Satellite remote sensing high-temperature heat wave risk early warning method, device, equipment and medium

Abstract

The invention provides a satellite remote sensing high-temperature heat wave risk early warning method, device and equipment and a medium. The method comprises the following steps: acquiring satellite remote sensing data, numerical mode forecast data, cloud coverage data and land utilization type analysis data corresponding to a research area; according to a model dynamic allocation mechanism, dynamically allocating a corresponding high-temperature heat wave risk early-warning model for each grid in the research area according to the cloud coverage data, and determining a high-temperature heat wave early-warning result corresponding to each grid based on satellite remote sensing data and numerical mode forecasting data through the high-temperature heat wave risk early-warning model corresponding to each grid; and determining a susceptibility weight corresponding to each grid according to the land utilization type analysis data, and determining a high-temperature heat wave risk level corresponding to each grid in combination with the high-temperature heat wave early warning result corresponding to each grid. According to the method, the high-temperature heat wave prediction precision can be remarkably improved, and meanwhile risk early warning is more regional and targeted.

Description

Technical Field

[0001] The present invention relates to the technical field of disaster early warning, and in particular to a satellite remote sensing high temperature heat wave risk early warning method, device, equipment and medium. Background Art

[0002] As an extreme meteorological event, high temperature heat waves not only have a serious impact on human health, agricultural production, and ecological environment, but may also cause damage to infrastructure and economic losses. Traditional high temperature heat wave early warning methods mainly rely on single data sources or static modeling techniques, which have certain limitations. Especially in the case of complex cloud cover conditions and diverse land use types, it is difficult to meet the requirements of refined early warning. Among them, the high temperature early warning model based on remote sensing data is sensitive to cloud cover conditions. When there is cloud cover, the quality of remote sensing data deteriorates or even disappears, resulting in incomplete results. And limited by the spatial resolution, it is difficult to conduct refined analysis for small-scale areas. The meteorological early warning model based on reanalysis data and model prediction data has a low spatial resolution, is difficult to adapt to local characteristics such as urban heat island effect, and lacks a direct reflection of surface temperature, resulting in large regional deviations in risk estimation. The traditional early warning model based on ground meteorological observations is limited by the spatial distribution of stations, and the blank areas between stations are difficult to be covered, with low prediction accuracy. Summary of the Invention

[0003] In view of this, the purpose of the present invention is to provide a satellite remote sensing high temperature heat wave risk early warning method, device, equipment and medium, which can significantly improve the accuracy of high temperature heat wave prediction, and at the same time make the risk early warning more regional and targeted.

[0004] In the first aspect, the present invention provides a satellite remote sensing high temperature heat wave risk early warning method, including:

[0005] Obtain satellite remote sensing data, numerical model forecast data, cloud cover data, and land use type analysis data corresponding to the research area;

[0006] According to the model dynamic allocation mechanism, dynamically allocate corresponding high temperature heat wave risk early warning models for each grid in the research area according to the cloud cover data, so as to determine the high temperature heat wave early warning results corresponding to each grid based on the satellite remote sensing data and numerical model forecast data through the high temperature heat wave risk early warning models corresponding to each grid;

[0007] Determine the susceptibility weight corresponding to each grid according to the land use type analysis data, and combine the high temperature heat wave early warning results corresponding to each grid to determine the high temperature heat wave risk level corresponding to each grid. The susceptibility weight is used to optimize the high temperature heat wave early warning results corresponding to each grid.

[0008] In one implementation, according to the model dynamic allocation mechanism, a corresponding high temperature and heat wave risk warning model is dynamically allocated for each grid in the study area according to the cloud cover data, including:

[0009] Perform the following operations on any grid in the study area:

[0010] Extract the cloud cover ratio value corresponding to the grid according to the cloud cover data;

[0011] Based on the threshold interval where the cloud cover ratio value is located, determine the cloud cover degree corresponding to the grid. The cloud cover degree is divided into light cloud cover, partial cloud cover, and cloudy cover;

[0012] According to the mapping relationship between the cloud cover degree and the high temperature and heat wave risk warning model, allocate the corresponding high temperature and heat wave risk warning model for the grid;

[0013] Among them, different cloud cover degrees correspond to different high temperature and heat wave risk warning models, and the model types of different high temperature and heat wave risk warning models are different, and the sample data sets used for training different high temperature and heat wave risk warning models are different.

[0014] In one implementation, when the cloud cover degree is light cloud cover, the high temperature and heat wave risk warning model adopts a random forest model. The training steps of the random forest model include: extracting the surface temperature, vegetation index, and radiation intensity based on the satellite remote sensing data of the study area at historical times, using the surface temperature, vegetation index, radiation intensity, and numerical model forecast data of the study area at historical times as model inputs, using the high temperature and heat wave event data of the study area at historical times as training labels, constructing a first sample data set, and training the random forest model using the first sample data set;

[0015] In the case where the cloud cover degree is partial cloud cover, the high temperature and heat wave risk warning model adopts a support vector machine model. The training steps of the support vector machine model include: based on the cloud cover data of the study area at historical times, performing weighted fusion on the satellite remote sensing data, ground station observation data, and reanalysis data of the study area at historical times to obtain temperature characteristics, using the temperature characteristics and numerical model forecast data of the study area at historical times as model inputs, using the high temperature and heat wave event data of the study area at historical times as training labels, constructing a second sample data set, and training the support vector machine model using the second sample data set;

[0016] In the case of moderate cloud cover, the high-temperature heatwave risk warning model uses the XGBoost model. The training steps of the high-temperature heatwave risk warning model include: using the reanalysis data and numerical model forecast data of the study area at historical times as model inputs, using the high-temperature heatwave event data of the study area at historical times as training labels, constructing a third sample dataset, and using the third sample dataset to train the XGBoost model.

[0017] In one implementation, based on the cloud cover data of the study area at historical times, the satellite remote sensing data, ground station observation data, and reanalysis data of the study area at historical times are weighted and fused to obtain temperature features, including:

[0018] Perform the following operations for any grid in the study area:

[0019] Based on the satellite remote sensing data, ground station observation data, and reanalysis data of the study area at historical times, respectively determine the land surface temperature, meteorological station observation temperature, and reanalysis temperature of the study area at historical times;

[0020] According to the cloud cover data of the study area at historical times, extract the cloud cover ratio value of this grid at historical times, and determine the standard weight value corresponding to this grid based on the cloud cover ratio value;

[0021] According to the cloud cover ratio value and the standard weight value, respectively determine the fusion weight coefficients corresponding to the land surface temperature, meteorological station observation temperature, and reanalysis temperature, so as to use the fusion weight coefficients to perform weighted fusion on the land surface temperature, meteorological station observation temperature, and reanalysis temperature to obtain temperature features.

[0022] In one implementation, determine the susceptibility weight corresponding to each grid according to the land use type analysis data, including:

[0023] Perform the following operations for any grid in the study area:

[0024] Determine the land use type to which this grid belongs according to the land use type analysis data;

[0025] According to the mapping relationship between the land use type, seasonal factors, and susceptibility weights, based on the land use type to which this grid belongs and the season it is in, determine the susceptibility weight corresponding to this grid.

[0026] In one implementation, combine the high-temperature heatwave warning results corresponding to each grid to determine the high-temperature heatwave risk level corresponding to each grid, including:

[0027] Perform the following operations for any grid in the study area:

[0028] Based on the susceptibility weight corresponding to the grid and the high-temperature heatwave warning result, determine the refined high-temperature risk index corresponding to the grid;

[0029] According to the refined high-temperature risk index corresponding to the grid, determine the high-temperature heatwave risk level corresponding to the grid.

[0030] In one implementation, based on the susceptibility weight corresponding to the grid and the high-temperature heatwave warning result, determining the refined high-temperature risk index corresponding to the grid includes:

[0031] Take the product between the susceptibility weight corresponding to the grid and the high-temperature heatwave warning result as the refined high-temperature risk index corresponding to the grid.

[0032] In a second aspect, the present invention further provides a satellite remote sensing high-temperature heatwave risk warning device, including:

[0033] A data acquisition module, configured to acquire satellite remote sensing data, numerical model prediction data, cloud coverage data, and land use type analysis data corresponding to the study area;

[0034] A model allocation and intensity prediction module, configured to dynamically allocate a corresponding high-temperature heatwave risk warning model for each grid in the study area according to the cloud coverage data according to the model dynamic allocation mechanism, so as to determine the high-temperature heatwave warning result corresponding to each grid based on the satellite remote sensing data and the numerical model prediction data through the high-temperature heatwave risk warning model corresponding to each grid;

[0035] A refined level determination module, configured to determine the susceptibility weight corresponding to each grid according to the land use type analysis data, and combine the high-temperature heatwave warning result corresponding to each grid to determine the high-temperature heatwave risk level corresponding to each grid, and the susceptibility weight is used to optimize the high-temperature heatwave warning result corresponding to each grid.

[0036] In a third aspect, the present invention further provides an electronic device, including a processor and a memory, the memory stores computer-executable instructions that can be executed by the processor, and the processor executes the computer-executable instructions to implement the method according to any one of the first aspect.

[0037] In a fourth aspect, the present invention further provides a computer-readable storage medium, the computer-readable storage medium stores computer-executable instructions, and when the computer-executable instructions are called and executed by the processor, the computer-executable instructions cause the processor to implement the method according to any one of the first aspect.

[0038] A method, device, equipment and medium for satellite remote sensing high-temperature heat wave risk early warning provided by the present invention first obtains satellite remote sensing data, numerical model prediction data, cloud coverage data and land use type analysis data corresponding to the research area; then, according to the model dynamic allocation mechanism, the corresponding high-temperature heat wave risk early warning model is dynamically allocated to each grid in the research area according to the cloud coverage data, so as to determine the high-temperature heat wave early warning result corresponding to each grid based on the satellite remote sensing data and the numerical model prediction data through the high-temperature heat wave risk early warning model corresponding to each grid; finally, the susceptibility weight corresponding to each grid is determined according to the land use type analysis data, and the high-temperature heat wave risk level corresponding to each grid is determined in combination with the high-temperature heat wave early warning result corresponding to each grid. The susceptibility weight is used to optimize the high-temperature heat wave early warning result corresponding to each grid. The above method dynamically adapts to the cloud cover condition, and improves the accuracy of high-temperature heat wave prediction in the next 1-3 days through the multi-source data fusion of satellite remote sensing data, land use type analysis data and numerical model prediction data; at the same time, combined with the refined analysis of land use type analysis data, the risk early warning is more regional and targeted, and can provide more reliable technical support for urban planning, agricultural production, drought resistance and disaster prevention, and disaster emergency and other fields. The present invention significantly improves the timeliness, spatial resolution and evaluation accuracy of high-temperature heat wave risk early warning, and has high innovation and application value.

[0039] Other features and advantages of the present invention will be described in the following specification, and, in part, will be obvious from the specification, or will be understood by implementing the present invention. The objectives and other advantages of the present invention are realized and obtained by the structures specifically pointed out in the specification, claims and drawings.

[0040] In order to make the above objectives, features and advantages of the present invention more obvious and understandable, the following specifically gives preferred embodiments and, in conjunction with the accompanying drawings, makes a detailed description as follows. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following will briefly introduce the drawings required for the description of the specific embodiments or the prior art. Obviously, the following drawings are some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0042] Figure 1 It is a schematic flow chart of a method for satellite remote sensing high-temperature heat wave risk early warning provided by an embodiment of the present invention;

[0043] Figure 2 It is a technical framework diagram of a method for satellite remote sensing high-temperature heat wave risk early warning provided by an embodiment of the present invention;

[0044] Figure 3 The structural schematic diagram of a satellite remote sensing high temperature heat wave risk early warning device provided by an embodiment of the present invention;

[0045] Figure 4 The structural schematic diagram of an electronic device provided by an embodiment of the present invention. Detailed implementation manners

[0046] To make the objectives, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below in conjunction with the embodiments. Apparently, the described embodiments are some but not all of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0047] Currently, the prior art has the problem of low prediction accuracy of high temperature heat wave risks. Moreover, traditional methods usually do not consider the differences in the responses of different land use types to high temperature heat waves, and it is impossible to refine regional risks, making it difficult to meet the needs of disaster emergency responses and different production activity entities, resulting in the lack of pertinence in risk early warning results. With the continuous development of remote sensing technology and meteorological observation means, as well as the increasing demand for refined early warning services for high temperature heat wave risks, a multi-mode refined high temperature heat wave risk early warning model based on multi-source data can give full play to the advantages of multi-source observation data, get rid of the dependence on data quality to a certain extent, and can provide more refined high temperature heat wave risk early warning results for various regions in the next 1-3 days, with strong technical feasibility and broad application prospects. Based on this, the embodiments of the present invention provide a satellite remote sensing high temperature heat wave risk early warning method, device, equipment and medium, which can significantly improve the prediction accuracy of high temperature heat waves and make the risk early warning more regional and targeted.

[0048] To facilitate the understanding of this embodiment, first, a satellite remote sensing high temperature heat wave risk early warning method disclosed in the embodiments of the present invention will be introduced in detail. Refer to Figure 1 The flowchart of a satellite remote sensing high temperature heat wave risk early warning method shown below. This method mainly includes the following steps S102 to step S106:

[0049] Step S102, obtain satellite remote sensing data, numerical model forecast data, cloud cover data, and land use type analysis data corresponding to the research area.

[0050] Step S104: According to the model dynamic allocation mechanism, dynamically allocate the corresponding high-temperature heatwave risk warning model for each grid in the study area based on the cloud cover data, so as to determine the high-temperature heatwave warning result corresponding to each grid through the high-temperature heatwave risk warning model corresponding to each grid, based on satellite remote sensing data and numerical model prediction data.

[0051] Among them, the high-temperature heatwave warning result includes the probability and intensity distribution of high-temperature heatwave events in the next 1 to 3 days. In one example, the cloud cover data is used to describe the cloud cover situation corresponding to each grid in the study area, such as various situations such as few cloud covers (or no cloud covers), partial cloud covers, and cloudy covers. The cloud cover data is used to identify the cloud cover situation corresponding to each grid, and on this basis, the corresponding high-temperature heatwave risk warning model is dynamically allocated for each grid.

[0052] Different cloud cover situations correspond to different high-temperature heatwave risk warning models, and the model types of different high-temperature heatwave risk warning models are different, and the sample data sets used for training different high-temperature heatwave risk warning models are different. For example, in the case of few cloud covers, the sample data set used for model training is constructed based on satellite remote sensing data and numerical model prediction data at historical moments; in the case of partial cloud covers, the sample data set used for model training is constructed based on satellite remote sensing data, ground station observation data, reanalysis data, and numerical model prediction data at historical moments; in the case of cloudy covers, the sample data set used for model training is constructed based on reanalysis data and numerical model prediction data at historical moments.

[0053] After the training of the high-temperature heatwave risk warning model is completed, the inputs of the model are all constructed based on the latest satellite remote sensing data and numerical model prediction data, and the outputs are all the probability and intensity distribution of high-temperature heatwave events in the next 1 to 3 days.

[0054] Step S106: Determine the susceptibility weight corresponding to each grid according to the land use type analysis data, and combine the high-temperature heatwave warning result corresponding to each grid to determine the high-temperature heatwave risk level corresponding to each grid.

[0055] Among them, the susceptibility weight is used to optimize the high-temperature heatwave warning result corresponding to each grid, and the specific value of the susceptibility weight is related to the current season and the land use type to which the grid belongs. In one example, the susceptibility weight corresponding to the grid is determined by combining the current season and the land use type to which the grid belongs, and the risk is refined based on the susceptibility weight and the high-temperature heatwave warning result to determine the high-temperature heatwave risk level corresponding to each grid.

[0056] The satellite remote sensing high-temperature heat wave risk early warning method provided by the embodiments of the present invention constructs a high-temperature heat wave risk early warning model for predicting the next 1-3 days based on multi-source data fusion (remote sensing satellite data, reanalysis data, ground station observation data, and numerical model prediction data), and realizes the dynamic adaptation of high-temperature heat wave risk early warning for different meteorological conditions and underlying surface areas with different characteristics through cloud cover classification (cloudless, partly cloudy, cloudy) and land use type analysis. This method takes the rapid perception, prediction, and early warning of high-temperature heat wave risk at a short time scale as the core, combines the refined analysis of land use types, and provides high-precision risk division and level estimation for different regions. Its advantage lies in being able to maintain high early warning accuracy under complex cloud conditions and conduct refined risk level division for different land use types. It solves the problems of insufficient accuracy and poor adaptability in cloud cover, data loss, and regional difference processing of traditional methods, thereby improving the pertinence and accuracy of high-temperature heat wave risk early warning.

[0057] In a specific implementation manner, the general idea of the embodiments of the present invention is as follows: First, use satellite remote sensing data, reanalysis data, and ground station observation data to model high-temperature heat waves under different cloud conditions. The data relied on for constructing the early warning model is different in different situations. In the cloudless case, it mainly relies on remote sensing satellite data for modeling. When it is cloudy, it relies on reanalysis data for modeling. In the case of partly cloudy, according to the cloud cover degree and the information of meteorological observation stations under the cloud, satellite remote sensing data and ground station observation data are fused for joint modeling. During use, the model is automatically and intelligently selected and dynamically adjusted according to real-time satellite remote sensing data, cloud conditions, and numerical model prediction data, and the optimal model is selected for high-temperature heat wave prediction. Then, combined with land use types, regional refinement is carried out, and finally, the high-temperature heat wave risk levels of different regions in the next 1-3 days are obtained to ensure the accuracy and timeliness of the early warning results.

[0058] Specifically, refer to Figure 2 As shown in the technical framework diagram of a satellite remote sensing high-temperature heat wave risk early warning method, meteorological factors are extracted and spatio-temporally matched for satellite remote sensing data, spatio-temporal matching and quality control are performed on ground station observation data, spatio-temporal matching and resampling are performed on reanalysis data. In addition, data format conversion, spatio-temporal matching, cloud cover detection and cloud amount division, research area division, etc. are performed on cloud cover data, and data cleaning and spatio-temporal matching, etc. are performed on historical high-temperature heat wave events. Three sets of sample sets are comprehensively constructed to generate labels + feature quantities for training random forest machine learning models, support vector machine models, and XGBoost models, and model evaluation is performed on the above models; when the scores of the above models meet the standards, the high-temperature heat wave early warning results are output using the models, and the high-temperature risk refinement assessment intensity is combined with heat wave susceptibility weights and land use type analysis data to obtain the high-temperature heat wave risk level division results.

[0059] For the convenience of understanding, an embodiment of the present invention provides a specific implementation manner of a satellite remote sensing high-temperature heat wave risk early warning method.

[0060] Before performing the foregoing step S104, it is necessary to pre-train the high-temperature heat wave risk early warning model. The training includes:

[0061] (1) Obtain multi-source historical high-temperature heat wave event data, that is, satellite remote sensing data, reanalysis data, ground station observation data, historical high-temperature heat wave event data, and numerical model forecast data in the study area at historical times, and perform data preprocessing, including:

[0062] (1.1) Use satellite remote sensing data provided by satellites such as Fengyun satellites or Himawari to obtain the land surface temperature (LST), and convert the data into a format suitable for analysis according to requirements. Ensure that the satellite remote sensing data is spatially consistent with other data sources, and align the satellite remote sensing data in time according to the time information of the high-temperature heat wave event.

[0063] (1.2) Use reanalysis data such as CRA40 and ERA5 to obtain meteorological variables. Spatial resolution adjustment: Perform spatial resampling on the reanalysis data to adapt to the resolution requirements of the study area, ensure time alignment with the satellite remote sensing data and ground station observation data, and perform missing value filling and interpolation processing.

[0064] (1.3) Collect ground station observation data, use the interpolation method to fill in missing values, ensure the integrity of the data, ensure time alignment with the satellite remote sensing data and reanalysis data, and remove outliers to ensure the accuracy of the data.

[0065] (1.4) Collect historical high-temperature heat wave event data, identify historical high-temperature heat wave events according to temperature thresholds and durations, and ensure that the time format of the historical event data is consistent with other data sources.

[0066] (1.5) Collect numerical model forecast data. Spatial resolution adjustment: Perform spatial resampling on the numerical model forecast data to adapt to the resolution requirements of the study area and remove outliers.

[0067] (2) Obtain cloud cover data, and divide the study area according to cloud amount conditions, including: extracting the cloud cover ratio value corresponding to the grid according to the cloud cover data, determining the cloud cover degree corresponding to the grid based on the threshold interval where the cloud cover ratio value is located, and assigning the corresponding high-temperature heat wave risk early warning model to the grid according to the mapping relationship between the cloud cover degree and the high-temperature heat wave risk early warning model.

[0068] In the embodiments of the present invention, cloud cover data is obtained through satellite remote sensing to obtain the surface cloud amount situation. The obtained cloud cover data is preprocessed, such as data format conversion, spatial alignment, and temporal alignment, to ensure its suitability for subsequent analysis and high-temperature heatwave risk assessment. Finally, the cloud cover ratio value is extracted from the cloud cover data. In specific implementation, according to the cloud cover data, the cloud amount situation in the study area is divided into three cases: cloudless, partly cloudy, and cloudy. This division will affect the selection of subsequent high-temperature heatwave modeling.

[0069] Few clouds: The cloud cover ratio value is lower than a certain threshold (between 0% and 30%). In this case, the influence of clouds on the surface temperature is small, and high-temperature heatwave modeling mainly relies on satellite remote sensing data.

[0070] Partly cloudy: The cloud cover ratio value is between 30% and 70%. In this case, clouds partly affect the surface temperature, and it is necessary to combine cloud cover data, satellite remote sensing data, ground station observation data, and reanalysis data for fusion modeling.

[0071] Cloudy: The cloud cover ratio value is higher than 70%. In this case, the shielding effect of clouds is large, and high-temperature heatwave modeling mainly relies on reanalysis data.

[0072] In the embodiments of the present invention, the cloud cover data is applied to the study area for spatial division to determine the cloud cover situation of each grid, and the study area is divided into several spatial grids. The size of each grid can be set according to research needs. According to the cloud cover data in each grid, each grid is classified into the category of few clouds, partly cloudy, or cloudy.

[0073] (3) Construction of a short-term (next 1 - 3 days) early warning model for high-temperature heatwaves and establishment of a dynamic adjustment mechanism under different cloud conditions, including:

[0074] (3.1) In the case of few cloud covers, the random forest model is adopted for the high-temperature heatwave risk early warning model. The training steps of the random forest model include: extracting surface temperature, vegetation index, and radiation intensity based on the satellite remote sensing data of the study area at historical moments. Using the surface temperature, vegetation index, radiation intensity, and numerical model forecast data of the study area at historical moments as model inputs, and using the high-temperature heatwave event data of the study area at historical moments as training labels to construct the first sample dataset, and training the random forest model with the first sample dataset.

[0075] In the case of cloudless skies, the influence of clouds is relatively small, and the surface temperature is mainly directly affected by solar radiation. Therefore, relying on satellite remote sensing data, a short-term prediction of high-temperature heatwaves is carried out based on the random forest model. Taking surface temperature, vegetation index, precipitation, wind speed, and radiation intensity as input features, after feature standardization and data partitioning of the data, combined with real-time satellite observation data and numerical model forecast data for data partitioning, a random forest model with high prediction accuracy and fast response ability is constructed to achieve the forecast of high-temperature heatwaves in the next 1 - 3 days.

[0076] In practical applications, the data can be partitioned by time window. For example: Input time window: Surface temperature, vegetation index, radiation intensity from the previous moment to the current moment (1 day), future meteorological variables (i.e., precipitation and wind speed) predicted by the numerical model from the previous moment to the current moment (1 day). Target time window: Probability and intensity distribution of high-temperature heatwave events in the next 1 to 3 days.

[0077] Furthermore, a real-time update strategy can be adopted. Every day, the latest satellite observation data and numerical model forecast data are added to the sample dataset to dynamically optimize the random forest model. Among them, the training set is used for model training, accounting for 70% of the data; the validation set is used for parameter tuning and model validation, accounting for 20% of the data; the test set is used for final model performance evaluation, accounting for 10% of the data.

[0078] Furthermore, the random forest hyperparameters are set as follows: Number of decision trees (n_estimators): 100 - 300; Maximum number of features (max_features): sqrt; Maximum depth of the tree (max_depth): [None, 10, 20]; Minimum number of samples for splitting (min_samples_split): 2 - 10; Minimum number of samples in a leaf node (min_samples_leaf): 1 - 5; Maximum number of leaf nodes (max_leaf_nodes): None; Sample weights (class_weight): balanced.

[0079] During the training process, the hyperparameters of the random forest model can be tuned through grid search (GridSearchCV) to finally obtain the best model. The code is as follows:

[0080] param_grid = {

[0081] 'n_estimators': 100 - 300,

[0082] 'max_depth': [None, 10, 20],

[0083] 'min_samples_split': 2 to 10,

[0084] 'min_samples_leaf': 1 to 5

[0085] }

[0086] grid_search = GridSearchCV(estimator = rf_model, param_grid = param_grid, cv = 3)

[0087] Furthermore, the importance evaluation of the model can be utilized to explain the contribution of each feature to the high-temperature heatwave warning result, so as to optimize the model subsequently.

[0088] Furthermore, the embodiment of the present invention also provides a dynamic update function: regularly retrain the model to incorporate the latest data and maintain the adaptability of the model.

[0089] Furthermore, after an extreme high-temperature heatwave event occurs in the research area, the feature weights of the model can be optimized through retrospective analysis.

[0090] (3.2) In the case of partial cloud cover, the high-temperature heatwave risk warning model adopts a support vector machine model. The training steps of the support vector machine model include: based on the cloud cover data of the research area at historical moments, performing weighted fusion on the satellite remote sensing data, ground station observation data, and reanalysis data of the research area at historical moments to obtain temperature features, using the temperature features and numerical model prediction data of the research area at historical moments as model inputs, using the high-temperature heatwave event data of the research area at historical moments as training labels, constructing a second sample dataset, and training the support vector machine model using the second sample dataset.

[0091] In the case of partial cloud cover, satellite remote sensing data, ground station observation data, and numerical model prediction data are used, combined with cloud cover data to jointly construct a high-temperature heatwave risk warning model for the next 1 - 3 days based on the support vector machine model. The input features are the surface temperature, meteorological station observation temperature, cloud cover, reanalysis temperature, temperature for the next 1 - 3 days predicted by the numerical model, precipitation, and other meteorological factors affecting high-temperature heatwaves in the research area at historical moments. The data is standardized, and weights are set for the data according to the cloud amount:

[0092] (3.21) According to the cloud cover data of the research area at historical moments, extract the cloud cover ratio value of this grid at historical moments, and determine the standard weight value corresponding to this grid based on the cloud cover ratio value. The calculation process of the standard weight value is as follows:

[0093] normalization_factor = 1 + 0.5 * cloud_fraction

[0094] Among them, normalization_factor is the standard weight value, and cloud_fraction is the cloud cover ratio value.

[0095] (3.22) According to the cloud cover ratio value and the standard weight value, determine the fusion weight coefficients corresponding to the land surface temperature, the meteorological station observed temperature, and the reanalysis temperature respectively, so as to use the fusion weight coefficients to perform weighted fusion on the land surface temperature, the meteorological station observed temperature, and the reanalysis temperature to obtain the temperature characteristics.

[0096] The calculation process of the fusion weight coefficient corresponding to the land surface temperature is as follows: w1 = (1 - cloud_fraction) / normalization_factor;

[0097] The calculation process of the fusion weight coefficient corresponding to the meteorological station observed temperature is as follows: w2 = cloud_fraction / normalization_factor;

[0098] The calculation process of the fusion weight coefficient corresponding to the reanalysis temperature is as follows: w3 = 0.5 * cloud_fraction / normalization_factor.

[0099] In practical applications, the data can be divided according to time windows. For example: input time window: temperature characteristics from the previous moment to the current moment (1 day), future meteorological variables predicted by the numerical model from the previous moment to the current moment (1 day). Target time window: probability and intensity distribution of high temperature and heat wave events from 1 day to 3 days in the future.

[0100] Furthermore, a real-time update strategy can be adopted to add the latest satellite remote sensing data to the sample dataset every day to dynamically optimize the model. Among them, the training set is used for model training, accounting for 70% of the data ratio; the validation set is used for parameter tuning and model validation, accounting for 20% of the data ratio; the test set is used for final model performance evaluation, accounting for 10% of the data ratio.

[0101] Furthermore, the hyperparameters for building the support vector machine high temperature and heat wave prediction model are as follows: penalty coefficient (C): 1, ~100; loss function tolerance (∈): 0.01 ~ 1; width of the RBF kernel function (γ): 0.01 ~ 0.1; kernel function type (kernel): 'rbf'; maximum number of iterations (max_iter): -1.

[0102] Furthermore, grid search (GridSearchCV) can be used to tune the hyperparameters of the SVR model to find the best model configuration. The code is as follows:

[0103] param_grid = {'C': 1 to 100, 'epsilon': 0.01 to 1}

[0104] Furthermore, the embodiment of the present invention provides a dynamic update function: regularly retrain the model to incorporate the latest data to ensure adaptability to changes in meteorological conditions.

[0105] (3.3) In the case of cloudy coverage, the high-temperature heatwave risk warning model uses the XGBoost model. The training steps of the high-temperature heatwave risk warning model include: using the reanalysis data and numerical model forecast data of the study area at historical times as the model input, and using the high-temperature heatwave event data of the study area at historical times as the training label to construct a third sample dataset, and training the XGBoost model using the third sample dataset.

[0106] In the case of clouds, due to the strong shielding effect of clouds on the surface temperature, it is difficult to directly rely on remote sensing observation data for the monitoring of high-temperature heatwaves. Therefore, a short-term forecasting model is constructed based on reanalysis data and numerical model forecast data. By using meteorological variables and historical high-temperature heatwave event data, a statistical model is established to predict the high-temperature heatwave risk in the next 1 to 3 days. Combining reanalysis data to construct features, including daily maximum temperature, temperature-humidity index, temperature change rate, water vapor pressure, radiation energy, wind speed and direction, and precipitation intensity. After standardizing the data, the training set and the test set are divided, and finally the XGBoost is used to construct a high-temperature heatwave risk warning model for the next 1-3 days.

[0107] Furthermore, the data can be divided according to time windows. For example: input time window: meteorological variables from the previous moment to the current moment (1 day), future meteorological variables predicted by the numerical model from the previous moment to the current moment (1 day). Target time window: the probability and intensity distribution of high-temperature heatwave events in the next 1 to 3 days.

[0108] Furthermore, a real-time update strategy is adopted. Every day, the latest satellite remote sensing data and numerical model forecast data are added to the sample dataset to dynamically optimize the model. Among them, the training set is used for model training, accounting for 70% of the data ratio; the validation set is used for parameter tuning and model validation, accounting for 20% of the data ratio; the test set is used for the final model performance evaluation, accounting for 10% of the data ratio.

[0109] Furthermore, the hyperparameters of the XGBoost short-term prediction model are as follows: the number of weak learners (n_estimators): 300; the maximum depth of the tree (max_depth): 6; the learning rate (learning_rate): 0.1; the minimum number of samples in a leaf node (min_child_weight): 3; the minimum loss reduction required to split a node (gamma): 0.1; the sample sampling ratio (subsample): 0.8; the feature sampling ratio (colsample_bytree): 0.8.

[0110] Furthermore, the early stopping method can be used to dynamically stop the training according to the performance of the validation set, and obtain the XGBoost machine learning model.

[0111] Furthermore, the embodiment of the present invention provides a dynamic update function: retrain or fine-tune the model according to the latest satellite remote sensing data and numerical model prediction data to adapt to short-term extreme weather changes.

[0112] In order to improve the accuracy and adaptability of the high temperature and heat wave risk warning model for the next 1-3 days, the embodiment of the present invention combines the above three individual models to construct a multi-mode ensemble dynamic adaptation model. By analyzing the spatio-temporal distribution characteristics of cloud cover, according to the research area division results, dynamically allocate model resources, intelligently select the optimal warning strategy, and comprehensively adapt to different meteorological conditions and spatial characteristics.

[0113] After completing the training of the high temperature and heat wave risk warning model, first, according to the model dynamic allocation mechanism, dynamically allocate the corresponding high temperature and heat wave risk warning model for each grid in the research area according to the cloud cover data. For details, please refer to the foregoing embodiments, and the embodiments of the present invention will not elaborate herein; then, through the high temperature and heat wave risk warning model corresponding to each grid, based on satellite remote sensing data and numerical model prediction data, determine the high temperature and heat wave warning results corresponding to each grid. The specific implementation process is as follows:

[0114] In the case of few clouds, the short-term prediction mechanism of the random forest model under cloud-free conditions: real-time data input: input the latest observed satellite remote sensing data into the trained random forest model; model prediction data input: input the latest model prediction data for the next 1-3 days into the trained random forest model; high temperature and heat wave prediction: output the probability and intensity distribution of high temperature and heat wave within 1 to 3 days in the future.

[0115] In the case of partial clouds, the short-term prediction mechanism of the support vector machine model under partial cloud conditions: real-time data input: input the latest observed satellite remote sensing data and numerical model prediction data into the trained support vector machine model; high temperature and heat wave prediction: output the probability and intensity distribution of high temperature and heat wave within 1 to 3 days in the future.

[0116] In cloudy conditions, the forecasting mechanism of the XGBoost model under cloudy conditions: Real-time data input: Input the latest observed satellite remote sensing data and numerical model forecasting data into the trained XGBoost forecasting model; High temperature and heat wave forecasting: The risk of high temperature and heat wave events in the next 1 to 3 days.

[0117] For the aforementioned step S106, the embodiments of the present invention also provide a specific implementation manner of determining the susceptibility weight corresponding to each grid according to the land use type analysis data, and combining the high temperature and heat wave warning results corresponding to each grid to determine the high temperature and heat wave risk level corresponding to each grid. The following operations (a) to (d) are performed for any grid in the study area:

[0118] Operation (a), determine the land use type to which the grid belongs according to the land use type analysis data.

[0119] Operation (b), according to the mapping relationship between the land use type, seasonal factors, and susceptibility weight, based on the land use type to which the grid belongs and the season in which it is located, determine the susceptibility weight corresponding to the grid.

[0120] (b1) Land use type data: Use satellite remote sensing data (such as MODIS, Landsat), high-resolution land cover maps, or local land use statistical data. According to the study area, the land use is divided into urban areas, agricultural land, forest areas, water bodies, and bare land, and the land use type data is regularly updated and optimized. The input is the high temperature and heat wave warning results obtained through satellite remote sensing data, ground station observation data, numerical model forecasting data, and high temperature and heat wave risk warning models in the aforementioned embodiments. Since the summer half-year is a season with severe and frequent high temperature and heat wave risks, higher-frequency high temperature risk forecasts are made in the summer half-year according to actual needs. (b2) According to the response characteristics of different land use types to high temperature and heat waves, assign specific susceptibility weights to each type as the basis for optimized evaluation. The characteristics of high temperature and heat waves in different seasons and the attributes of land use types result in different sensitivities for different regions, so the weights will be divided by quarter.

[0121] Table 1 Susceptibility weight

[0122]

[0123]

[0124] Operation (c), based on the susceptibility weight corresponding to the grid and the high temperature and heat wave warning result, determine the refined high temperature risk index corresponding to the grid. Specifically, the product of the susceptibility weight corresponding to the grid and the high temperature and heat wave warning result is used as the refined high temperature risk index corresponding to the grid. The formula is as follows:

[0125] R final (x, y) = R initial (x, y) · W landuse (x, y);

[0126] Among them, R final (x, y) is the refined high - temperature risk index, R initial (x, y) is the high - temperature heatwave warning result, W landuse (x, y) is the susceptibility weight of land use types. To avoid sudden changes at the data boundaries, a spatial smoothing algorithm is used to optimize the results.

[0127] Operation (d), according to the refined high - temperature risk index corresponding to the grid, determine the high - temperature heatwave risk level corresponding to the grid.

[0128] Establish different grading criteria in different regions. Based on the idea of dynamically adjusting the percentile range and combined with the characteristics of the regional climate environment, adapt the risk classification of high - temperature events. The following is an example table of the grading criteria we provide. Different regions can adopt more refined grading criteria according to the actual situation. Generally, mainly considering the characteristics of six land use types: urban, agricultural, forest, grassland, bare land, and water body, the percentile range is set differently.

[0129] Table 2 High - temperature heatwave risk level

[0130]

[0131]

[0132] In summary, the method provided by the embodiments of the present invention has at least the following advantages:

[0133] (1) Through multi - source data fusion, including satellite remote - sensing data, re - analysis data, surface meteorological observation data, and numerical model forecast data, combined with land use type analysis, the embodiments of the present invention achieve high - precision forecasting of high - temperature heatwave risks for the next 1 - 3 days. Dynamically select suitable modeling methods and adjust the model in real - time according to cloud cover, improving the applicability and accuracy of early warnings under different weather conditions. Compared with traditional methods, it significantly improves the refinement level and spatial resolution of high - temperature heatwave risk early warnings, providing reliable support for precise disaster prevention and mitigation.

[0134] (2) The technology of the embodiments of the present invention has more flexible adaptability and regional optimization compared with the traditional method. According to the cloud coverage and regional characteristics, the embodiments of the present invention dynamically adjust the data fusion strategy and model parameters to flexibly respond to various scenarios such as cloudless, partially cloudy, and cloudy, ensuring the scientific nature of the evaluation results. Combining the differential risk characteristics of land use types (such as urban, agricultural, forest, etc.), the high-temperature heatwave risk is refined and divided regionally. By optimizing the specific susceptibilities of regions such as urban, agricultural, forest, grassland, bare land, and water bodies, the pertinence and regional adaptability of the high-temperature heatwave risk warning are further improved.

[0135] Based on the foregoing embodiments, the embodiments of the present invention provide a satellite remote sensing high-temperature heatwave risk warning device. Refer to Figure 3 the structural schematic diagram of a satellite remote sensing high-temperature heatwave risk warning device shown, and the device mainly includes the following parts:

[0136] The data acquisition module 302 is used to acquire satellite remote sensing data, numerical model prediction data, cloud coverage data, and land use type analysis data corresponding to the study area;

[0137] The model allocation and intensity prediction module 304 is used to dynamically allocate corresponding high-temperature heatwave risk warning models for each grid in the study area according to the cloud coverage data according to the model dynamic allocation mechanism, so as to determine the high-temperature heatwave warning result corresponding to each grid based on the satellite remote sensing data and the numerical model prediction data through the high-temperature heatwave risk warning model corresponding to each grid;

[0138] The refined level determination module 306 is used to determine the susceptibility weight corresponding to each grid according to the land use type analysis data, and combine the high-temperature heatwave warning result corresponding to each grid to determine the high-temperature heatwave risk level corresponding to each grid. The susceptibility weight is used to optimize the high-temperature heatwave warning result corresponding to each grid.

[0139] The satellite remote sensing high temperature heat wave risk early warning device provided by the embodiment of the present invention constructs a high temperature heat wave risk early warning model for predicting the high temperature heat wave risk in the next 1-3 days based on multi-source data fusion (remote sensing satellite data, reanalysis data, ground station observation data, and numerical model forecast data), and realizes the high temperature heat wave risk early warning that dynamically adapts to different meteorological conditions and underlying surface areas with different characteristics through cloud cover classification (cloudless, partly cloudy, cloudy) and land use type analysis. This method takes the rapid perception, forecasting, and early warning of high temperature heat wave risks on a short time scale as the core, combines the refined analysis of land use types, and provides high-precision risk division and level estimation for different regions. Its advantage lies in being able to maintain high early warning accuracy under complex cloud conditions and conduct refined risk level division for different land use types. It solves the problems of insufficient accuracy and poor adaptability in cloud cover, data missing, and regional difference processing in traditional methods, thereby improving the pertinence and accuracy of high temperature heat wave risk early warning.

[0140] In one implementation manner, the model allocation and intensity prediction module 304 is specifically configured to:

[0141] Perform the following operations on any grid within the research area:

[0142] Extract the cloud cover ratio value corresponding to the grid according to the cloud cover data;

[0143] Based on the threshold interval where the cloud cover ratio value is located, determine the cloud cover degree corresponding to the grid. The cloud cover degree is divided into light cloud cover, partly cloud cover, and heavy cloud cover;

[0144] According to the mapping relationship between the cloud cover degree and the high temperature heat wave risk early warning model, allocate the corresponding high temperature heat wave risk early warning model to the grid;

[0145] Among them, different cloud cover degrees correspond to different high temperature heat wave risk early warning models, and different high temperature heat wave risk early warning models have different model types, and different sample data sets are used for training different high temperature heat wave risk early warning models.

[0146] In one implementation manner, when the cloud cover degree is light cloud cover, the high temperature heat wave risk early warning model adopts a random forest model, and further includes a first model training module, which is used for: extracting the surface temperature, vegetation index, and radiation intensity based on the satellite remote sensing data of the research area at historical moments, using the surface temperature, vegetation index, radiation intensity, and numerical model forecast data of the research area at historical moments as model inputs, using the high temperature heat wave event data of the research area at historical moments as training labels, constructing a first sample data set, and using the first sample data set to train the random forest model;

[0147] In the case of partial cloud cover in terms of cloud cover degree, the high-temperature heatwave risk warning model adopts a support vector machine model, and further includes a second model training module for: based on the cloud cover data of the study area at historical moments, performing weighted fusion on the satellite remote sensing data, ground station observation data, and reanalysis data of the study area at historical moments to obtain temperature features, using the temperature features and numerical model prediction data of the study area at historical moments as model inputs, using the high-temperature heatwave event data of the study area at historical moments as training labels, constructing a second sample dataset, and using the second sample dataset to train the support vector machine model;

[0148] In the case of mostly cloudy in terms of cloud cover degree, the high-temperature heatwave risk warning model adopts an XGBoost model, and further includes a third model training module for: using the reanalysis data and numerical model prediction data of the study area at historical moments as model inputs, using the high-temperature heatwave event data of the study area at historical moments as training labels, constructing a third sample dataset, and using the third sample dataset to train the XGBoost model.

[0149] In one implementation, the second model training module is specifically used for:

[0150] Performing the following operations for any grid within the study area:

[0151] Based on the satellite remote sensing data, ground station observation data, and reanalysis data of the study area at historical moments, respectively determine the land surface temperature, meteorological station observation temperature, and reanalysis temperature of the study area at historical moments;

[0152] According to the cloud cover data of the study area at historical moments, extract the cloud cover ratio value of this grid at historical moments, and determine the standard weight value corresponding to this grid based on the cloud cover ratio value;

[0153] According to the cloud cover ratio value and the standard weight value, respectively determine the fusion weight coefficients corresponding to the land surface temperature, meteorological station observation temperature, and reanalysis temperature, so as to perform weighted fusion on the land surface temperature, meteorological station observation temperature, and reanalysis temperature using the fusion weight coefficients to obtain temperature features.

[0154] In one implementation, the refined level determination module 306 is specifically used for:

[0155] Performing the following operations for any grid within the study area:

[0156] Determine the land use type to which this grid belongs according to the land use type analysis data;

[0157] Based on the mapping relationship among land use types, seasonal factors, and susceptibility weights, determine the susceptibility weight corresponding to the grid based on the land use type to which the grid belongs and the season in which it is located.

[0158] In one implementation, the refinement level determination module 306 is specifically configured to:

[0159] Perform the following operations on any grid within the study area:

[0160] Based on the susceptibility weight corresponding to the grid and the high temperature and heat wave warning result, determine the refined high temperature risk index corresponding to the grid;

[0161] According to the refined high temperature risk index corresponding to the grid, determine the high temperature and heat wave risk level corresponding to the grid.

[0162] In one implementation, the refinement level determination module 306 is specifically configured to:

[0163] Take the product of the susceptibility weight corresponding to the grid and the high temperature and heat wave warning result as the refined high temperature risk index corresponding to the grid.

[0164] The device provided by the embodiments of the present invention has the same implementation principle and the same technical effects as those of the foregoing method embodiments. For the sake of brief description, for the parts not mentioned in the device embodiments, reference may be made to the corresponding content in the foregoing method embodiments.

[0165] The embodiments of the present invention provide an electronic device. Specifically, the electronic device includes a processor and a storage device; a computer program is stored on the storage device, and the computer program, when run by the processor, executes the method according to any one of the foregoing embodiments.

[0166] Figure 4 FIG. is a schematic structural diagram of an electronic device provided by an embodiment of the present invention. The electronic device 100 includes: a processor 40, a memory 41, a bus 42, and a communication interface 43. The processor 40, the communication interface 43, and the memory 41 are connected through the bus 42; the processor 40 is used to execute an executable module stored in the memory 41, such as a computer program.

[0167] Among them, the memory 41 may include a high-speed random access memory (RAM, Random Access Memory), and may also include a non-volatile memory, such as at least one disk memory. Through at least one communication interface 43 (which can be wired or wireless), a communication connection is realized between the system network element and at least one other network element, and the Internet, wide area network, local area network, metropolitan area network, etc. can be used.

[0168] The bus 42 can be an ISA bus, a PCI bus, an EISA bus, or the like. The bus can be divided into an address bus, a data bus, a control bus, etc. For the sake of convenience of representation, Figure 4 only a bidirectional arrow is used in Figure 4 , but it does not mean that there is only one bus or one type of bus.

[0169] Among them, the memory 41 is used to store a program. After receiving an execution instruction, the processor 40 executes the program. The method executed by the device defined by the flow process disclosed in any embodiment of the foregoing embodiments of the present invention can be applied to the processor 40 or implemented by the processor 40.

[0170] The processor 40 may be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above method can be completed by the integrated logic circuit in the hardware of the processor 40 or the instructions in the form of software. The above-mentioned processor 40 can be a general-purpose processor, including a central processing unit (CPU for short), a network processor (NP for short), etc.; it can also be a digital signal processor (DSP for short), an application specific integrated circuit (ASIC for short), a field-programmable gate array (FPGA for short), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components. It can implement or execute the various methods, steps and logic block diagrams disclosed in the embodiments of the present invention. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor, etc. The steps of the method disclosed in combination with the embodiments of the present invention can be directly embodied as being executed and completed by the hardware decoding processor, or executed and completed by the combination of the hardware and software modules in the decoding processor. The software module can be located in a mature storage medium in the art such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory, or an electrically erasable programmable memory, a register, etc. This storage medium is located in the memory 41, and the processor 40 reads the information in the memory 41 and combines its hardware to complete the steps of the above method.

[0171] The computer program product of the readable storage medium provided by the embodiments of the present invention includes a computer-readable storage medium storing program codes. The instructions included in the program codes can be used to execute the method described in the foregoing method embodiments. For the specific implementation, reference can be made to the foregoing method embodiments, which will not be elaborated here.

[0172] When the above-mentioned functions are implemented in the form of software function units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage medium includes: various media such as USB flash drives, mobile hard disks, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), magnetic disks, or optical discs that can store program codes.

[0173] Finally, it should be noted that the above-described embodiments are only specific embodiments of the present invention, used to illustrate the technical solutions of the present invention, rather than limiting it. The protection scope of the present invention is not limited thereto. Although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that any person skilled in the art within the technical scope disclosed by the present invention can still modify the technical solutions described in the foregoing embodiments, or can easily conceive of changes, or make equivalent replacements for some of the technical features; and these modifications, changes, or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be covered by the protection scope of the present invention. Therefore, the protection scope of the present invention should be determined by the protection scope of the claims.

Claims

1. A satellite remote sensing high-temperature heat wave risk early warning method, characterized in that Including: Obtaining satellite remote sensing data, numerical model prediction data, cloud cover data, and land use type analysis data corresponding to the study area; According to the model dynamic allocation mechanism, dynamically allocate corresponding high temperature and heat wave risk warning models for each grid in the study area according to the cloud cover data, so as to determine the high temperature and heat wave warning results corresponding to each grid through the high temperature and heat wave risk warning models corresponding to each grid, based on the satellite remote sensing data and the numerical model prediction data; Determine the susceptibility weight corresponding to each grid according to the land use type analysis data, and combine the high temperature and heat wave warning results corresponding to each grid to determine the high temperature and heat wave risk level corresponding to each grid, where the susceptibility weight is used to optimize the high temperature and heat wave warning results corresponding to each grid.

2. The satellite remote sensing high temperature heat wave risk warning method according to claim 1, wherein According to the model dynamic allocation mechanism, dynamically allocate corresponding high temperature and heat wave risk warning models for each grid in the study area according to the cloud cover data, including: Perform the following operations on any grid in the study area: Extract the cloud cover ratio value corresponding to the grid according to the cloud cover data; Based on the threshold interval where the cloud cover ratio value is located, determine the cloud cover degree corresponding to the grid, and the cloud cover degree is divided into few cloud cover, partial cloud cover, and multi-cloud cover; Allocate the corresponding high temperature and heat wave risk warning model for the grid according to the mapping relationship between the cloud cover degree and the high temperature and heat wave risk warning model; Among them, different cloud cover degrees correspond to different high temperature and heat wave risk warning models, and different high temperature and heat wave risk warning models have different model types, and different sample data sets are used for training different high temperature and heat wave risk warning models.

3. The satellite remote sensing high temperature heat wave risk early warning method according to claim 2, wherein In the case where the cloud cover degree is few cloud cover, the high temperature and heat wave risk warning model adopts a random forest model, and the training steps of the random forest model include: extracting the surface temperature, vegetation index, and radiation intensity based on the satellite remote sensing data of the study area at historical times, using the surface temperature, the vegetation index, the radiation intensity, and the numerical model prediction data of the study area at historical times as model inputs, using the high temperature and heat wave event data of the study area at historical times as training labels, constructing a first sample data set, and training the random forest model using the first sample data set; In the case where the cloud cover degree is partial cloud cover, the high temperature and heat wave risk warning model adopts a support vector machine model, and the training steps of the support vector machine model include: based on the cloud cover data of the study area at historical times, performing weighted fusion on the satellite remote sensing data, ground station observation data, and reanalysis data of the study area at historical times to obtain temperature characteristics, using the temperature characteristics and numerical model prediction data of the study area at historical times as model inputs, using the high temperature and heat wave event data of the study area at historical times as training labels, constructing a second sample data set, and training the support vector machine model using the second sample data set; When the cloud coverage degree is the cloudy coverage, the high temperature and heat wave risk early warning model adopts the XGBoost model, and the training steps of the high temperature and heat wave risk early warning model include: using the reanalysis data and numerical model prediction data of the study area at historical moments as model inputs, using the high temperature and heat wave event data of the study area at historical moments as training labels, constructing a third sample data set, and using the third sample data set to train the XGBoost model.

4. The satellite remote sensing high-temperature heat wave risk early warning method according to claim 3, wherein Based on the cloud coverage data of the study area at historical moments, weighted fusion of the satellite remote sensing data, ground station observation data, and reanalysis data of the study area at the historical moments is performed to obtain temperature features, including: Perform the following operations for any grid within the study area: Based on the satellite remote sensing data, ground station observation data, and reanalysis data of the study area at the historical moments, respectively determine the land surface temperature, meteorological station observation temperature, and reanalysis temperature of the study area at the historical moments; According to the cloud coverage data of the study area at historical moments, extract the cloud coverage ratio value of this grid at historical moments, and determine the standard weight value corresponding to this grid based on the cloud coverage ratio value; According to the cloud coverage ratio value and the standard weight value, respectively determine the fusion weight coefficients corresponding to the land surface temperature, the meteorological station observation temperature, and the reanalysis temperature, so as to perform weighted fusion on the land surface temperature, the meteorological station observation temperature, and the reanalysis temperature using the fusion weight coefficients to obtain temperature features.

5. The satellite remote sensing high-temperature heat wave risk early warning method according to claim 1, wherein Determine the susceptibility weight corresponding to each grid according to the land use type analysis data, including: Perform the following operations for any grid within the study area: Determine the land use type to which this grid belongs according to the land use type analysis data; According to the mapping relationship between the land use type, seasonal factors, and susceptibility weight, based on the land use type to which this grid belongs and the season in which it is located, determine the susceptibility weight corresponding to this grid.

6. The satellite remote sensing high temperature heat wave risk warning method according to claim 1, wherein Combining the high temperature and heat wave early warning results corresponding to each grid, determine the high temperature and heat wave risk level corresponding to each grid, including: Perform the following operations for any grid within the study area: Based on the susceptibility weight and the high temperature and heat wave early warning result corresponding to this grid, determine the refined high temperature risk index corresponding to this grid; According to the refined high temperature risk index corresponding to this grid, determine the high temperature and heat wave risk level corresponding to this grid.

7. The satellite remote sensing high temperature heat wave risk warning method according to claim 6, wherein Based on the susceptibility weight and the high temperature and heat wave early warning result corresponding to this grid, determine the refined high temperature risk index corresponding to this grid, including: Take the product between the susceptibility weight and the high temperature and heat wave early warning result corresponding to this grid as the refined high temperature risk index corresponding to this grid.

8. A satellite remote sensing high-temperature heat wave risk early warning device, characterized in that, Including: A data acquisition module for acquiring satellite remote sensing data, numerical model prediction data, cloud coverage data, and land use type analysis data corresponding to the study area; A model allocation and intensity prediction module, which is used to dynamically allocate corresponding high-temperature heatwave risk warning models for each grid in the research area according to the cloud coverage data according to the model dynamic allocation mechanism, so as to determine the high-temperature heatwave warning results corresponding to each grid through the high-temperature heatwave risk warning models corresponding to each grid, based on the satellite remote sensing data and the numerical model prediction data. A refined level determination module, which is used to determine the susceptibility weight corresponding to each grid according to the land use type analysis data, and combine the high-temperature heatwave warning results corresponding to each grid to determine the high-temperature heatwave risk level corresponding to each grid, and the susceptibility weight is used to optimize the high-temperature heatwave warning results corresponding to each grid.

9. An electronic device, characterized in that, It includes a processor and a memory. The memory stores computer-executable instructions that can be executed by the processor, and the processor executes the computer-executable instructions to implement the method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-executable instructions. When the computer-executable instructions are called and executed by the processor, the computer-executable instructions prompt the processor to implement the method according to any one of claims 1 to 7.

Images