Construction method of high-temperature heat wave approaching early warning model and early warning method

Abstract

The invention provides a construction method of a high-temperature heat wave approaching early warning model and an early warning method. The construction method comprises the following steps: acquiring historical single-station high-temperature heat wave event data, ground meteorological station observation data and meteorological satellite land surface temperature product data recorded by a meteorological observation station in a target area; performing time-space matching and data quality control processing on historical single-station high-temperature heat wave event data, ground meteorological station observation data and meteorological satellite land surface temperature product data, and constructing a high-temperature heat wave approaching early warning sample set; and based on the high-temperature heat wave approaching early-warning sample set, performing high-temperature heat wave approaching early-warning model training by adopting a bidirectional long-short-term memory neural network and a time sequence weight method to obtain a trained high-temperature heat wave approaching early-warning model. The accuracy of high-temperature heat wave approaching early warning is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of high-temperature heat wave disaster early warning, and in particular to a method for constructing a high-temperature heat wave approaching early warning model and an early warning method. Background Art

[0002] Traditional high temperature heat wave warnings usually rely on weather factors such as temperature predicted by numerical models, but the results of numerical forecasts are inevitably uncertain and cannot fully reflect the actual high temperature heat wave situation. The data that can truly reflect the high temperature heat wave situation comes from observations at ground meteorological stations and monitoring results from satellite remote sensing. Among them, ground meteorological station observations have higher accuracy and better temporal continuity, but poor spatial continuity and cannot fully invert the high temperature heat wave situation in a large area. Summary of the invention

[0003] In view of this, an object of the present invention is to provide a method for constructing a high-temperature heat wave approaching warning model and a warning method, so as to improve the accuracy of the high-temperature heat wave approaching warning.

[0004] In order to achieve the above purpose, the technical solution adopted by the embodiment of the present invention is as follows:

[0005] In a first aspect, an embodiment of the present invention provides a method for constructing a high-temperature heat wave imminent warning model, including: obtaining historical single-station high-temperature heat wave event data, ground-based meteorological station observation data, and meteorological satellite land surface temperature product data recorded by meteorological observation stations in a target area; performing spatiotemporal matching and data quality control processing on the historical single-station high-temperature heat wave event data, ground-based meteorological station observation data, and meteorological satellite land surface temperature product data to construct a high-temperature heat wave imminent warning sample set; based on the high-temperature heat wave imminent warning sample set, a bidirectional long short-term memory neural network and a temporal weight method are used to train a high-temperature heat wave imminent warning model to obtain a trained high-temperature heat wave imminent warning model; wherein the high-temperature heat wave imminent warning model is used for high-temperature heat wave imminent warning in a first preset time period in the future.

[0006] In one embodiment, historical single-station high temperature heat wave event data, ground meteorological station observation data, and meteorological satellite land surface temperature product data are subjected to spatiotemporal matching and data quality control processing to construct a high temperature heat wave imminent warning sample set, including: data cleaning of the historical single-station high temperature heat wave event data, and determining the high temperature heat wave event warning level of each high temperature heat wave event based on the event intensity index in the historical single-station high temperature heat wave event data; spatiotemporal matching and data quality control processing of the ground meteorological station observation data to obtain the target ground meteorological station observation data within a second preset time period; spatiotemporal matching and data quality control processing of the meteorological satellite land surface temperature product data to obtain the target meteorological satellite land surface temperature product data within the second preset time period; spatially interpolating the event intensity index of each high temperature heat wave event or the high temperature heat wave event warning level and the target ground meteorological station observation data based on the spatial resolution of the target meteorological satellite land surface temperature product data, and normalizing the target ground meteorological station observation data and the target meteorological satellite land surface temperature product data to obtain the high temperature heat wave imminent warning sample set.

[0007] In one embodiment, the high temperature heat wave event warning level of each historical single-station high temperature heat wave event is determined based on the event intensity index in the historical single-station high temperature heat wave event data, including: if the event intensity index is less than a first threshold, then the high temperature heat wave event warning level of the historical single-station high temperature heat wave event is determined to be the first level; if the event intensity index is greater than the first threshold and less than the second threshold, then the high temperature heat wave event warning level of the historical single-station high temperature heat wave event is determined to be the second level; if the event intensity index is greater than the second threshold and less than the third threshold, then the high temperature heat wave event warning level of the historical single-station high temperature heat wave event is determined to be the third level; if the event intensity index is greater than the third threshold, then the high temperature heat wave event warning level of the historical single-station high temperature heat wave event is determined to be the fourth level.

[0008] In one embodiment, the ground meteorological station observation data is subjected to spatiotemporal matching and data quality control processing to obtain the target ground meteorological station observation data within a second preset time period, including: based on the observation time of the ground meteorological station, the ground meteorological station observation data is matched with the time information in the historical single-station high temperature heat wave event data to obtain the ground meteorological station observation data within the second preset time period; wherein the time information includes the start date and end date of the high temperature heat wave event; the second preset time period is from the preset number of days before the start date of the high temperature heat wave event to the end date of the high temperature heat wave event; the ground meteorological station observation data within the second preset time period is subjected to data quality control processing to obtain the target ground meteorological station observation data; wherein the data quality control processing at least includes: climatological limit value check, climate extreme value check, internal consistency check, spatial consistency check and temporal consistency check.

[0009] In one embodiment, the meteorological satellite land surface temperature product data is subjected to spatiotemporal matching and data quality control processing to obtain target meteorological satellite land surface temperature product data within a second preset time period, including: performing data analysis on the meteorological satellite land surface temperature product data to obtain the meteorological satellite scanning time and scanning longitude and latitude; based on the meteorological satellite scanning time, matching the meteorological satellite land surface temperature product data with the time information in the historical single-station high temperature heat wave event data to obtain the meteorological satellite land surface temperature product data within the second preset time period; based on the scanning longitude and latitude of the meteorological satellite land surface temperature product data within the second preset time period, searching for the surface temperature data of the ground meteorological observation station within a preset range; matching and comparing the meteorological satellite land surface temperature product data within the second preset time period with the surface temperature data of the ground meteorological observation station in the corresponding time period, and smoothing the meteorological satellite land surface temperature product data within the second preset time period to obtain the target meteorological satellite land surface temperature product data.

[0010] In one embodiment, based on a high-temperature heat wave approaching warning sample set, a bidirectional long short-term memory neural network and a temporal weight method are used to train a high-temperature heat wave approaching warning model, including: setting network parameters of the bidirectional long short-term memory neural network, and determining the weight of each time step of the bidirectional long short-term memory neural network through an attention mechanism to obtain a high-temperature heat wave approaching warning model; dividing the high-temperature heat wave approaching warning sample set into a training sample set and a test sample set; iteratively training the high-temperature heat wave approaching warning model based on the training sample set, and evaluating the trained high-temperature heat wave approaching warning model based on the test sample set; if the evaluation index of the high-temperature heat wave approaching warning model meets the preset requirements, a trained high-temperature heat wave approaching warning model is obtained, otherwise, the high-temperature heat wave approaching warning model is iteratively trained again based on the training sample set until the evaluation index meets the preset requirements; wherein the evaluation index includes at least: mean absolute error and root mean square error.

[0011] In a second aspect, an embodiment of the present invention provides a method for warning of an approaching high-temperature heat wave, comprising: obtaining ground meteorological station observation data and meteorological satellite land surface temperature product data in a target area; inputting the ground meteorological station observation data and the meteorological satellite land surface temperature product data into a pre-trained high-temperature heat wave approach warning model to obtain a high-temperature heat wave approach warning result for a first preset time period in the future of the target area; wherein the high-temperature heat wave approach warning model is obtained by using any method for constructing a high-temperature heat wave approach warning model provided in the first aspect above, and the high-temperature heat wave approach warning result includes: an event intensity index of a high-temperature heat wave event or a high-temperature heat wave event warning level.

[0012] In a third aspect, an embodiment of the present invention provides a device for constructing a high-temperature heat wave imminent warning model, including: a first data acquisition module, used to obtain historical single-station high-temperature heat wave event data, ground meteorological station observation data, and meteorological satellite land surface temperature product data recorded by meteorological observation stations in a target area; a sample set construction module, used to perform spatiotemporal matching and data quality control processing on historical single-station high-temperature heat wave event data, ground meteorological station observation data, and meteorological satellite land surface temperature product data, and construct a high-temperature heat wave imminent warning sample set; a model training module, used to train a high-temperature heat wave imminent warning model based on the high-temperature heat wave imminent warning sample set, using a bidirectional long short-term memory neural network and a temporal weight method, to obtain a trained high-temperature heat wave imminent warning model; wherein the high-temperature heat wave imminent warning model is used for high-temperature heat wave imminent warning in the first preset time period in the future.

[0013] In a fourth aspect, an embodiment of the present invention provides a high-temperature heat wave imminent warning device, comprising: a second data acquisition module, used to acquire ground meteorological station observation data and meteorological satellite land surface temperature product data of a target area; a high-temperature heat wave imminent warning module, used to input the ground meteorological station observation data and the meteorological satellite land surface temperature product data into a pre-trained high-temperature heat wave imminent warning model, and obtain a high-temperature heat wave imminent warning result for a first preset time period in the future of the target area; wherein the high-temperature heat wave imminent warning model is obtained by adopting any method for constructing a high-temperature heat wave imminent warning model provided in the first aspect above, and the high-temperature heat wave imminent warning result includes: an event intensity index of a high-temperature heat wave event or a high-temperature heat wave event warning level.

[0014] In a fifth aspect, an embodiment of the present invention provides an electronic device, comprising a processor and a memory, wherein the memory stores computer-executable instructions that can be executed by the processor, and the processor executes the computer-executable instructions to implement the steps of any one of the methods provided in the first aspect or the steps of any one of the methods provided in the second aspect.

[0015] In a sixth aspect, an embodiment of the present invention provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of any one of the methods provided in the first aspect or the steps of any one of the methods provided in the second aspect are executed.

[0016] The embodiments of the present invention bring the following beneficial effects:

[0017] The method for constructing the above-mentioned high-temperature heat wave approaching warning model and the warning method provided by the embodiments of the present invention first obtain the historical single-station high-temperature heat wave event data, ground meteorological station observation data and meteorological satellite land surface temperature product data recorded by the meteorological observation station in the target area; then perform spatiotemporal matching and data quality control processing on the historical single-station high-temperature heat wave event data, ground meteorological station observation data and meteorological satellite land surface temperature product data to construct a high-temperature heat wave approaching warning sample set; finally, based on the high-temperature heat wave approaching warning sample set, a bidirectional long short-term memory neural network and a temporal weight method are used to train the high-temperature heat wave approaching warning model to obtain a trained high-temperature heat wave approaching warning model; wherein the high-temperature heat wave approaching warning model is used for the high-temperature heat wave approaching warning in the first preset time period in the future. The above method is based on ground meteorological station observation data and meteorological satellite land surface temperature product data, combined with a bidirectional long short-term memory neural network and a time series weight method, to construct an approaching high temperature heat wave warning for the first preset time period in the future. The bidirectional long short-term memory neural network can be used to obtain the historical and future characteristics of the high temperature heat wave time series for the model. On this basis, the time series weight method can provide different weights for the characteristic factors at different time points, highlighting the characteristic factors that contribute more at different time points, thereby improving the accuracy of the approaching high temperature heat wave warning.

[0018] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or understood by practicing the present invention. The purpose and other advantages of the present invention are realized and obtained by the structures particularly pointed out in the description, claims and drawings.

[0019] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, preferred embodiments are given below and described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0021] Figure 1 A flowchart of a method for constructing a high temperature heat wave approaching warning model provided by an embodiment of the present invention;

[0022] Figure 2 A flowchart of a method for constructing a high temperature heat wave approaching warning model combining a bidirectional long short-term memory neural network and time series weights provided in an embodiment of the present invention;

[0023] Figure 3 A flowchart of a method for early warning of an approaching high temperature heat wave provided by an embodiment of the present invention;

[0024] Figure 4 A schematic diagram of the structure of a device for constructing a high-temperature heat wave approaching warning model provided by an embodiment of the present invention;

[0025] Figure 5 A schematic structural diagram of a high temperature heat wave approaching warning device provided by an embodiment of the present invention;

[0026] Figure 6 A schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0027] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0028] At present, traditional high temperature heat wave warnings cannot fully reflect the actual high temperature heat wave conditions, and cannot comprehensively reflect the high temperature heat wave conditions in large areas.

[0029] Based on this, the embodiments of the present invention provide a method for constructing a high-temperature heat wave approaching warning model and a warning method, which can improve the accuracy of the high-temperature heat wave approaching warning.

[0030] To facilitate understanding of this embodiment, a method for constructing a high temperature heat wave approaching warning model disclosed in an embodiment of the present invention is first introduced in detail. The method can be executed by an electronic device, such as a smart phone, a computer, a tablet computer, etc. Figure 1 The flowchart of a method for constructing a high temperature heat wave approaching warning model is shown, which illustrates that the method mainly includes the following steps S101 to S103:

[0031] Step S101: Acquire historical single-station high temperature heat wave event data recorded by meteorological observation stations in the target area, ground meteorological station observation data, and meteorological satellite land surface temperature product data.

[0032] In one embodiment, historical single-station high temperature heat wave events recorded by meteorological observation stations in the target area are obtained, and original information such as the station name, longitude, latitude, start date, end date, event intensity index, etc. of each high temperature heat wave event is recorded; ground meteorological station observation data in the target area are obtained, including regional stations and automatic stations, etc., and the observation data at least include: average temperature, maximum temperature, average surface temperature, relative humidity, minimum temperature, sea level pressure, perceived temperature, sunshine hours and average wind speed, etc.; meteorological satellite (such as FY4B) land surface temperature product data is obtained, and data analysis is performed to read the meteorological satellite scanning time and scanning longitude and latitude.

[0033] Step S102: Performing spatiotemporal matching and data quality control processing on historical single-station high temperature heat wave event data, ground meteorological station observation data, and meteorological satellite land surface temperature product data to construct a high temperature heat wave approaching warning sample set.

[0034] In one implementation, historical single-station high temperature heat wave event data, ground meteorological station observation data, and meteorological satellite land surface temperature product data are respectively subjected to spatiotemporal matching and data quality control processing, and a high temperature heat wave imminent warning sample set is constructed based on the processed data.

[0035] Step S103: Based on the high temperature heat wave approach warning sample set, a high temperature heat wave approach warning model is trained using a bidirectional long short-term memory neural network and a temporal weight method to obtain a trained high temperature heat wave approach warning model.

[0036] In one embodiment, the high temperature heat wave approaching warning model is used for the high temperature heat wave approaching warning in the first preset time period in the future. The high temperature heat wave approaching warning model trained in this embodiment can provide high temperature heat wave approaching warning in the next 1-3 days.

[0037] In the specific implementation, a bidirectional long short-term memory neural network Bi-LSTM model is selected as a high temperature heat wave approaching warning model, which is composed of a forward LSTM and a backward LSTM. The Bi-LSTM network can better capture bidirectional time series information, and therefore can solve the computational requirements of time series prediction determined by both the front and back parties. On this basis, a weight is introduced for each time step of the Bi-LSTM model based on the attention mechanism to constrain the similarity between the current time step and the historical time step, so that the Bi-LSTM model can focus on the key information in the time series and improve the accuracy of time series prediction. Based on this, a high temperature heat wave approaching warning model is constructed in the embodiment of the present invention, and the model is trained using the data in the high temperature heat wave approaching warning sample set to obtain a trained high temperature heat wave approaching warning model, which is used for the high temperature heat wave approaching warning in the next 1-3 days.

[0038] The method for constructing the above-mentioned high-temperature heat wave imminent warning model provided by an embodiment of the present invention is based on ground meteorological station observation data and meteorological satellite land surface temperature product data, combined with a bidirectional long short-term memory neural network and a time series weight method, to construct a high-temperature heat wave imminent warning for the first preset time period in the future. The bidirectional long short-term memory neural network can be used to obtain the historical and future characteristics of the high-temperature heat wave time series for the model. On this basis, the time series weight method can provide different weights for the characteristic factors at different time points, highlighting the characteristic factors that contribute more at different time points, thereby improving the accuracy of the high-temperature heat wave imminent warning.

[0039] In one embodiment, for the aforementioned step S102, that is, when performing spatiotemporal matching and data quality control processing on the historical single-station high temperature heat wave event data, the ground meteorological station observation data, and the meteorological satellite land surface temperature product data to construct a high temperature heat wave imminent warning sample set, the following methods may be adopted, including but not limited to the following steps (1) to (4):

[0040] Step (1): Clean the historical single-station high temperature heat wave event data, and determine the high temperature heat wave event warning level for each high temperature heat wave event based on the event intensity index in the historical single-station high temperature heat wave event data.

[0041] In the specific implementation, for the number of historical single-station high temperature heat wave events obtained, first, check the original information such as the site name, longitude, latitude, start date, end date, event intensity index, etc. of each high temperature heat wave event, and remove the information that lacks any field information or has obvious outliers to complete the cleaning of the historical single-station high temperature heat wave event data set; then, count each high temperature heat wave event, arrange each high temperature heat wave event in order from small to large according to the event intensity index, and select the first threshold, second threshold and third threshold to classify the high temperature heat wave events.

[0042] Specifically, if the event intensity index is less than the first threshold, the high temperature heat wave event warning level of the historical single-station high temperature heat wave event is determined to be the first level; if the event intensity index is greater than the first threshold and less than the second threshold, the high temperature heat wave event warning level of the historical single-station high temperature heat wave event is determined to be the second level; if the event intensity index is greater than the second threshold and less than the third threshold, the high temperature heat wave event warning level of the historical single-station high temperature heat wave event is determined to be the third level; if the event intensity index is greater than the third threshold, the high temperature heat wave event warning level of the historical single-station high temperature heat wave event is determined to be the fourth level.

[0043] In a specific implementation, the 50%, 70%, 90% percentiles A, B, and C can be selected as the first threshold, the second threshold, and the third threshold, respectively, and the corresponding events are graded to form a high temperature heat wave event warning level R. If the event intensity index is less than A, R is 1 (i.e., the first level, blue warning), if the event intensity index is greater than A and less than B, R is 2 (i.e., the second level, yellow warning), if the event intensity index is greater than B and less than C, R is 3 (i.e., the third level, orange warning), otherwise, R is 4 (i.e., the fourth level, red warning).

[0044] Step (2): Performing spatiotemporal matching and data quality control processing on the ground meteorological station observation data to obtain the target ground meteorological station observation data within a second preset time period.

[0045] In one embodiment, the spatiotemporal matching and data quality control processing of ground meteorological station observation data may be performed in the following ways, including but not limited to:

[0046] First, based on the observation time of the ground meteorological station, the ground meteorological station observation data is matched with the time information in the historical single-station high temperature heat wave event data to obtain the ground meteorological station observation data within a second preset time period; wherein the time information includes the start date and end date of the high temperature heat wave event; the second preset time period is from the preset number of days before the start date of the high temperature heat wave event to the end date of the high temperature heat wave event.

[0047] In the specific implementation, based on the observation time of the ground meteorological station, the ground meteorological station observation data is matched with the start date t1 and the end date t2 of the high temperature heat wave event in step (1), and the ground meteorological station observation data within the second preset time period is screened, that is, the ground meteorological station observation data from m days before t1 to t2, a total of (t2-t1+1+m) days. In the embodiment of the present invention, m can be 5-7 days.

[0048] Then, data quality control processing is performed on the ground meteorological station observation data within the second preset time period to obtain the target ground meteorological station observation data.

[0049] In the specific implementation, the quality control of the observation data of the screened ground meteorological stations is carried out, including: climatological limit value check, climate extreme value check, internal consistency check, spatial consistency check and time consistency check. Among them, the climatological limit value check is to check whether the observation data must be within the climatological limit value; the climate extreme value check is to check whether the observation data exceeds the climate extreme value; the internal consistency check is to check whether the relationship between the records of meteorological elements observed at the same time must conform to a certain law; the spatial consistency check is to compare the observation data of the meteorological station with the observation data of other nearby meteorological stations to determine whether the observation data is reasonable; the time consistency test is to check whether the changes of the observation data within a certain time range have a specific law.

[0050] Step (3): Performing spatiotemporal matching and data quality control processing on the meteorological satellite land surface temperature product data to obtain the target meteorological satellite land surface temperature product data within a second preset time period.

[0051] In one embodiment, the spatiotemporal matching and data quality control processing of meteorological satellite land surface temperature product data may be performed in the following ways, including but not limited to:

[0052] Firstly, the land surface temperature product data of meteorological satellites are analyzed to obtain the scanning time and longitude and latitude of meteorological satellites.

[0053] Secondly, based on the meteorological satellite scanning time, the meteorological satellite land surface temperature product data is matched with the time information in the historical single-station high temperature heat wave event data to obtain the meteorological satellite land surface temperature product data within the second preset time period.

[0054] In the specific implementation, based on the meteorological satellite scanning time, the meteorological satellite land surface temperature product data is matched with the start date t1 and the end date t2 of the high temperature heat wave event in step (1), and the meteorological satellite land surface temperature product data for a total of (t2-t1+1+m) days from m days before t1 to t2 is obtained by screening.

[0055] Then, based on the scanning longitude and latitude of the meteorological satellite land surface temperature product data within the second preset time period, the surface temperature data of the ground meteorological observation station within the preset range is searched.

[0056] In the specific implementation, based on the meteorological satellite land surface temperature product data obtained by the above screening, the surface temperature data of the ground meteorological observation station within a preset range is searched, that is, the surface temperature data of the ground meteorological observation station within a radius of N kilometers (such as 2 kilometers) of the scanning longitude and latitude center point is searched.

[0057] Finally, the meteorological satellite land surface temperature product data within the second preset time period is matched and compared with the surface temperature data of the ground meteorological observation station in the corresponding time period, and the meteorological satellite land surface temperature product data within the second preset time period is smoothed to obtain the target meteorological satellite land surface temperature product data.

[0058] In the specific implementation, the meteorological satellite land surface temperature product data within the second preset time period obtained by the above screening are matched and compared with the surface temperature data of the ground meteorological observation station on the corresponding date. If the meteorological satellite land surface temperature product data has obvious outliers or exceeds the preset threshold (such as 50°C), the meteorological satellite land surface temperature product data is subjected to sliding smoothing correction to obtain the target meteorological satellite land surface temperature product data.

[0059] Step (4): Based on the spatial resolution of the target meteorological satellite land surface temperature product data, the event intensity index or high temperature heat wave event warning level of each high temperature heat wave event and the target ground meteorological station observation data are spatially interpolated, and the target ground meteorological station observation data and the target meteorological satellite land surface temperature product data are normalized to obtain a high temperature heat wave imminent warning sample set.

[0060] In specific implementation, the event intensity index of each high temperature heat wave event or the high temperature heat wave event warning level can be used as sample labels, the target ground meteorological station observation data and the target meteorological satellite land surface temperature product data can be used as sample feature data, and the sample feature data can be normalized to obtain a high temperature heat wave imminent warning sample set.

[0061] Specifically, the event intensity index of the high temperature heat wave event or the high temperature heat wave event warning level obtained in step (1) and the target ground meteorological station observation data obtained in step (2) (such as: average temperature, maximum temperature, average surface temperature, relative humidity, minimum temperature, sea level pressure, body temperature, sunshine hours and average wind speed and other elements) are spatially interpolated, and the target spatial resolution of the interpolation is the spatial resolution of the meteorological satellite land surface temperature product data obtained in step (3) to form a sample set, wherein the event intensity index of the high temperature heat wave event or the high temperature heat wave event warning level is the sample label, and the rest are sample feature quantity data.

[0062] Furthermore, all sample feature data are normalized to the maximum and minimum values ​​and converted into dimensionless values, thereby eliminating the dimension differences of each indicator and avoiding the problems of model error reduction and training time extension caused by different data dimensions during model training. Assuming that the original value of a sample feature data is x, x min is the minimum value in the data, x max is the maximum value in the data, then the normalized x′ calculation formula is:

[0063] In one embodiment, for the aforementioned step S103, i.e., based on the high temperature heat wave approach warning sample set, the high temperature heat wave approach warning model training using the bidirectional long short-term memory neural network and the temporal weight method can be performed using the following methods, including but not limited to:

[0064] Firstly, the network parameters of the bidirectional long short-term memory neural network are set, and the weight of each time step of the bidirectional long short-term memory neural network is determined through the attention mechanism to obtain the approaching high temperature heat wave warning model.

[0065] In the specific implementation, the LSTM and Bi-LSTM network parameters are set as follows:

[0066] The number of LSTM input nodes is 10, which means that the model input factors are 9 ground meteorological station elements, including meteorological satellite land surface temperature and average temperature, maximum temperature, average surface temperature, relative humidity, minimum temperature, sea level pressure, body temperature, sunshine hours and average wind speed, totaling 10 feature quantities;

[0067] Number of LSTM output nodes: 1 to 3, indicating the predicted days;

[0068] LSTM initial time step: 15;

[0069] LSTM activation function: ReLU;

[0070] Number of LSTM hidden layers: 2;

[0071] LSTM hidden layer nodes: 64;

[0072] LSTM batch size: 16;

[0073] LSTM initial learning rate: 0.0005;

[0074] LSTM loss function: MSE;

[0075] LSTM optimizer: RMSprop or Adagrad;

[0076] LSTM iterations: 500

[0077] LSTM dropout: 0.5;

[0078] Bi-LSTM: bidirectional=True.

[0079] The method of introducing the attention mechanism is as follows:

[0080] 1) When Bi-LSTM decodes to S i-1 Time step, to proceed to the next time step S iWhen predicting, it is necessary to calculate each input position h j Impact on current position ij ,Right now:

[0081] ej=a(S-1,h)

[0082] Among them, S i represents the time step of the i-th day, S i-1 represents the time step of the i-1th day, h j Represents the position, that is, the position of the j-th spatial point.

[0083] 2) For e ij Normalize it and get the attention weight distribution a ij ,Right now:

[0084]

[0085] Where T represents the total number of spatial points.

[0086] 3) Using a ij Perform weighted summation to obtain the corresponding context vector c, namely:

[0087]

[0088] 4) Calculate the prediction S for the next time step i ,Right now:

[0089] S=f(S-1,hi-1,Ci)

[0090] Secondly, the high temperature heat wave approaching warning sample set is divided into a training sample set and a test sample set.

[0091] In the specific implementation, the high temperature heat wave approach warning sample set constructed in the above step (4) is divided into a training sample set (Train) and a test sample set (Test) in a ratio of 8:2, and the label data (Y) and the feature data (X) are distinguished.

[0092] Then, the high temperature heat wave approach warning model is iteratively trained based on the training sample set, and the trained high temperature heat wave approach warning model is evaluated based on the test sample set.

[0093] In the specific implementation, the label data (Train_Y) and feature data (Train_X) of the training sample set are substituted into the constructed Bi-LSTM model (i.e., the high temperature heat wave approaching warning model) for iterative training; then the feature value (Test_X) of the test sample set is substituted into the trained model to obtain the predicted value (Test_Y'), and Test_Y' is compared with Test_Y to evaluate the trained model. According to the characteristics of the model, the mean absolute error and the root mean square error can be used as evaluation indicators to evaluate the model accuracy in the embodiment of the present invention.

[0094] The mean absolute error (MAE) is used to measure the average absolute difference between the predicted value and the true value, indicating the average degree of deviation between the predicted value and the true value. The calculation formula is as follows:

[0095]

[0096] Among them, N represents the total number of samples, y i and x i represent the predicted value and the true value respectively.

[0097] The root mean square error (RMSE) is the square root of the ratio of the square of the deviation between the predicted value and the true value to the number of observations. The calculation formula is as follows:

[0098]

[0099] Finally, if the evaluation indicators of the high temperature heat wave approaching warning model meet the preset requirements, a trained high temperature heat wave approaching warning model is obtained; otherwise, the high temperature heat wave approaching warning model is iteratively trained again based on the training sample set until the evaluation indicators meet the preset requirements; among which, the evaluation indicators include at least: mean absolute error and root mean square error.

[0100] In the specific implementation, the model quality is judged according to the model evaluation indicators MAE and RMSE. The closer MAE and RMSE are to 0, the higher the model accuracy is. Otherwise, the above steps can be repeated to further tune the model parameters and retrain to continuously improve the model training effect.

[0101] The above method provided by the embodiment of the present invention obtains historical single-station high temperature heat wave event data, and performs data cleaning and level classification, obtains ground meteorological station data, and performs time-space matching and data quality control processing, obtains meteorological satellite land surface temperature product data, and performs time-space matching and data quality control processing, constructs a high temperature heat wave imminent warning sample set and training model, performs training and prediction based on the model, and improves the accuracy of the high temperature heat wave imminent warning.

[0102] For ease of understanding, the present invention also provides a specific method for constructing a high temperature heat wave approaching warning model combining a bidirectional long short-term memory neural network and a time series weight, see Figure 2 As shown, it mainly includes the following steps 1 to 5:

[0103] Step 1: Obtain historical single-station high temperature heat wave event data, and perform data cleaning and classification.

[0104] Step 2: Obtain ground meteorological station data, perform spatiotemporal matching and data quality control processing, and extract meteorological characteristic elements.

[0105] Step 3: Obtain land surface temperature product data from meteorological satellites and perform spatiotemporal matching and data quality control processing.

[0106] Step 4: Construct a sample set and training model for warning of approaching high temperature and heat waves.

[0107] Specifically, spatial interpolation is performed on the event intensity index (or level) of high temperature heat wave events and the ground meteorological station data (9 types of elements such as average temperature and maximum temperature), and the event intensity index (or level) of high temperature heat wave events is used as label data, and the ground meteorological station data (9 types of elements such as average temperature and maximum temperature) and meteorological satellite land surface temperature product data (land surface temperature) are used as feature quantities to construct a sample set of high temperature heat wave imminent warning.

[0108] Furthermore, a bidirectional long short-term memory neural network Bi-LSTM model is selected, the Bi-LSTM network parameters are set, and a weight is introduced for each time step of the Bi-LSTM network based on the attention mechanism.

[0109] Step 5: Perform training and prediction based on the model.

[0110] Specifically, the Bi-LSTM model is iteratively trained using the sample set data, and the trained Bi-LSTM model is evaluated using the test set data. The model quality is judged based on the model evaluation indicators MAE and RMSE. The closer MAE and RMSE are to 0, the higher the model accuracy. Otherwise, the model parameters are further tuned and retrained to continuously improve the model training effect.

[0111] When it is necessary to explain, the implementation principle of the method provided in this embodiment is the same as that of the aforementioned method embodiment, and will not be repeated again.

[0112] The above method provided by the embodiment of the present invention constructs a high temperature heat wave approaching warning model for the next 1-3 days based on satellite remote sensing and ground meteorological station observations based on artificial intelligence algorithms, which can give full play to the advantages of multi-source observation data and get rid of the dependence on numerical forecasting models to a certain extent. The above method is based on satellite remote sensing and ground meteorological station observation data, combined with a bidirectional long short-term memory neural network and a time series weight method, to construct a high temperature heat wave approaching warning model for the next 1-3 days. The long short-term memory neural network can obtain the historical and future characteristics of the high temperature heat wave time series for the model. On this basis, adding time series weights can provide different weights for characteristic factors at different time points, highlighting the characteristic factors that contribute more at different time points, thereby improving the accuracy of the high temperature heat wave approaching warning.

[0113] The present invention also provides a method for early warning of approaching high temperature heat waves. Figure 3 The flowchart of a high temperature heat wave approaching warning method shown in FIG. 1 illustrates that the method mainly includes the following steps S301 to S302:

[0114] Step S301: Acquire ground meteorological station observation data and meteorological satellite land surface temperature product data in the target area.

[0115] Step S320: inputting the ground meteorological station observation data and the meteorological satellite land surface temperature product data into a pre-trained high temperature heat wave approaching warning model to obtain the high temperature heat wave approaching warning result for the first preset time period in the future of the target area.

[0116] Among them, the high temperature heat wave imminent warning model is obtained by adopting the construction method of the high temperature heat wave imminent warning model provided in the aforementioned embodiment, and the high temperature heat wave imminent warning result includes: the event intensity index of the high temperature heat wave event or the high temperature heat wave event warning level.

[0117] The above-mentioned high-temperature heat wave approach warning method provided by an embodiment of the present invention uses a pre-constructed high-temperature heat wave approach warning model to perform high-temperature heat wave approach warning in the target area. The model uses a bidirectional long short-term memory neural network to obtain historical and future characteristics of the high-temperature heat wave time series for the model. On this basis, a time series weight method can provide different weights for characteristic factors at different time points, highlighting the characteristic factors that contribute more at different time points, thereby improving the accuracy of the high-temperature heat wave approach warning.

[0118] Regarding the method for constructing a high temperature heat wave approaching warning model provided in the above embodiment, the embodiment of the present invention further provides a device for constructing a high temperature heat wave approaching warning model, see Figure 4 The schematic diagram of the structure of a device for constructing a high temperature heat wave approaching warning model is shown, indicating that the device mainly includes the following parts:

[0119] The first data acquisition module 401 is used to acquire historical single-station high temperature heat wave event data recorded by meteorological observation stations in the target area, ground meteorological station observation data, and meteorological satellite land surface temperature product data;

[0120] The sample set construction module 402 is used to perform spatiotemporal matching and data quality control processing on historical single-station high temperature heat wave event data, ground meteorological station observation data, and meteorological satellite land surface temperature product data to construct a high temperature heat wave approach warning sample set;

[0121] The model training module 403 is used to train a high-temperature heat wave approaching warning model based on a high-temperature heat wave approaching warning sample set, using a bidirectional long short-term memory neural network and a temporal weight method to obtain a trained high-temperature heat wave approaching warning model; wherein the high-temperature heat wave approaching warning model is used for high-temperature heat wave approaching warning in the first preset time period in the future.

[0122] The device for constructing the above-mentioned high-temperature heat wave imminent warning model provided by an embodiment of the present invention is based on ground meteorological station observation data and meteorological satellite land surface temperature product data, combined with a bidirectional long short-term memory neural network and a time series weight method, to construct a high-temperature heat wave imminent warning for a first preset time period in the future. The bidirectional long short-term memory neural network can be used to obtain the historical and future characteristics of the high-temperature heat wave time series for the model. On this basis, the time series weight method can provide different weights for the characteristic factors at different time points, highlighting the characteristic factors that contribute more at different time points, thereby improving the accuracy of the high-temperature heat wave imminent warning.

[0123] In one embodiment, the sample set construction module 402 is specifically used to: perform data cleaning on historical single-station high temperature heat wave event data, and determine the high temperature heat wave event warning level of each high temperature heat wave event based on the event intensity index in the historical single-station high temperature heat wave event data; perform spatiotemporal matching and data quality control processing on ground meteorological station observation data to obtain target ground meteorological station observation data within a second preset time period; perform spatiotemporal matching and data quality control processing on meteorological satellite land surface temperature product data to obtain target meteorological satellite land surface temperature product data within the second preset time period; perform spatial interpolation on the event intensity index of each high temperature heat wave event or the high temperature heat wave event warning level and the target ground meteorological station observation data based on the spatial resolution of the target meteorological satellite land surface temperature product data, and perform normalization processing on the target ground meteorological station observation data and the target meteorological satellite land surface temperature product data to obtain a high temperature heat wave imminent warning sample set.

[0124] In one embodiment, the above-mentioned sample set construction module 402 is specifically used to: if the event intensity index is less than the first threshold, then determine the high temperature heat wave event warning level of the historical single-station high temperature heat wave event as the first level; if the event intensity index is greater than the first threshold and less than the second threshold, then determine the high temperature heat wave event warning level of the historical single-station high temperature heat wave event as the second level; if the event intensity index is greater than the second threshold and less than the third threshold, then determine the high temperature heat wave event warning level of the historical single-station high temperature heat wave event as the third level; if the event intensity index is greater than the third threshold, then determine the high temperature heat wave event warning level of the historical single-station high temperature heat wave event as the fourth level.

[0125] In one embodiment, the sample set construction module 402 is specifically used to: match the ground meteorological station observation data with the time information in the historical single-station high temperature heat wave event data based on the observation time of the ground meteorological station, and obtain the ground meteorological station observation data within a second preset time period; wherein the time information includes the start date and end date of the high temperature heat wave event; the second preset time period is the preset number of days before the start date of the high temperature heat wave event to the end date of the high temperature heat wave event; perform data quality control processing on the ground meteorological station observation data within the second preset time period to obtain the target ground meteorological station observation data; wherein the data quality control processing at least includes: climatological boundary value check, climate extreme value check, internal consistency check, spatial consistency check and time consistency check.

[0126] In one embodiment, the sample set construction module 402 is specifically used to: perform data analysis on meteorological satellite land surface temperature product data to obtain the meteorological satellite scanning time and scanning longitude and latitude; based on the meteorological satellite scanning time, match the meteorological satellite land surface temperature product data with the time information in the historical single-station high temperature heat wave event data to obtain the meteorological satellite land surface temperature product data within a second preset time period; based on the scanning longitude and latitude of the meteorological satellite land surface temperature product data within the second preset time period, search for the surface temperature data of the ground meteorological observation station within a preset range; match and compare the meteorological satellite land surface temperature product data within the second preset time period with the surface temperature data of the ground meteorological observation station in the corresponding time period, and smooth the meteorological satellite land surface temperature product data within the second preset time period to obtain the target meteorological satellite land surface temperature product data.

[0127] In one embodiment, the model training module 403 is specifically used to: set the network parameters of the bidirectional long short-term memory neural network, and determine the weight of each time step of the bidirectional long short-term memory neural network through the attention mechanism, so as to obtain a high-temperature heat wave approaching warning model, and divide the high-temperature heat wave approaching warning sample set into a training sample set and a test sample set; iteratively train the high-temperature heat wave approaching warning model based on the training sample set, and evaluate the trained high-temperature heat wave approaching warning model based on the test sample set; if the evaluation index of the high-temperature heat wave approaching warning model meets the preset requirements, the trained high-temperature heat wave approaching warning model is obtained, otherwise, the high-temperature heat wave approaching warning model is iteratively trained again based on the training sample set until the evaluation index meets the preset requirements; wherein the evaluation index includes at least: mean absolute error and root mean square error.

[0128] For the high temperature heat wave approaching warning method provided in the above embodiment, the embodiment of the present invention also provides a high temperature heat wave approaching warning device, see Figure 5 The schematic diagram of the structure of a high temperature heat wave approaching warning device is shown, indicating that the device mainly includes the following parts:

[0129] The second data acquisition module 501 is used to acquire ground meteorological station observation data and meteorological satellite land surface temperature product data of the target area;

[0130] The high temperature heat wave imminent warning module 502 is used to input the ground meteorological station observation data and the meteorological satellite land surface temperature product data into a pre-trained high temperature heat wave imminent warning model to obtain the high temperature heat wave imminent warning result for the first preset time period in the future of the target area; wherein the high temperature heat wave imminent warning model is obtained by adopting the construction method of the high temperature heat wave imminent warning model provided in the aforementioned embodiment, and the high temperature heat wave imminent warning result includes: the event intensity index of the high temperature heat wave event or the high temperature heat wave event warning level.

[0131] The above-mentioned high-temperature heat wave approach warning device provided by an embodiment of the present invention uses a pre-constructed high-temperature heat wave approach warning model to perform high-temperature heat wave approach warning in the target area. The model uses a bidirectional long short-term memory neural network to obtain historical and future characteristics of the high-temperature heat wave time series for the model. On this basis, a time series weight method can provide different weights for characteristic factors at different time points, highlighting the characteristic factors that contribute more at different time points, thereby improving the accuracy of the high-temperature heat wave approach warning.

[0132] It should be noted that the implementation principle and technical effects of the device provided in the embodiment of the present invention are the same as those of the aforementioned method embodiment. For the sake of brief description, for matters not mentioned in the device embodiment, reference can be made to the corresponding contents in the aforementioned method embodiment. The specific numerical values ​​provided in the implementation of the present invention are only exemplary and are not limited here.

[0133] An embodiment of the present invention further provides an electronic device. Specifically, the electronic device includes a processor and a storage device. The storage device stores a computer program, and when the computer program is executed by the processor, it executes the method described in any one of the above implementation methods.

[0134] Figure 6 A structural diagram of an electronic device provided in an embodiment of the present invention, the electronic device 100 includes: a processor 60, a memory 61, a bus 62 and a communication interface 63, wherein the processor 60, the communication interface 63 and the memory 61 are connected via the bus 62; the processor 60 is used to execute an executable module stored in the memory 61, such as a computer program.

[0135] The memory 61 may include a high-speed random access memory (RAM), and may also include a non-volatile memory, such as at least one disk memory. The communication connection between the system network element and at least one other network element is realized through at least one communication interface 63 (which may be wired or wireless), and the Internet, wide area network, local area network, metropolitan area network, etc. may be used.

[0136] The bus 62 may be an ISA bus, a PCI bus, or an EISA bus, etc. The bus may be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 6 Only one bidirectional arrow is used in the diagram, but this does not mean that there is only one bus or only one type of bus.

[0137] Among them, the memory 61 is used to store programs, and the processor 60 executes the program after receiving the execution instruction. The method executed by the device for flow process definition disclosed in any embodiment of the above-mentioned embodiment of the present invention can be applied to the processor 60 or implemented by the processor 60.

[0138] The processor 60 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 hardware integrated logic circuit or software instructions in the processor 60. The above processor 60 can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components. The methods, steps and logic block diagrams disclosed in the embodiments of the present invention can be implemented or executed. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor. The steps of the method disclosed in the embodiment of the present invention can be directly embodied as a hardware decoding processor for execution, or a combination of hardware and software modules in the decoding processor for execution. The software module can be located in a storage medium mature 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. The storage medium is located in the memory 61, and the processor 60 reads the information in the memory 61 and completes the steps of the above method in combination with its hardware.

[0139] The computer program product of the readable storage medium provided in the embodiment of the present invention includes a computer-readable storage medium storing program code, and the instructions included in the program code can be used to execute the methods described in the previous method embodiments. The specific implementation can be referred to the previous method embodiments, which will not be repeated here.

[0140] If the functions are implemented in the form of software functional 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, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc., which can store program codes.

[0141] Finally, it should be noted that the above-described embodiments are only specific implementations of the present invention, which are used to illustrate the technical solutions of the present invention, rather than to limit them. The protection scope of the present invention is not limited thereto. Although the present invention is described in detail with reference to the above-mentioned embodiments, ordinary technicians in the field should understand that any technician familiar with the technical field can still modify the technical solutions recorded in the above-mentioned embodiments within the technical scope disclosed by the present invention, or can easily think of changes, or make equivalent replacements for some of the technical features therein; and these modifications, changes or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be based on the protection scope of the claims.

Claims

1. A method for constructing a high temperature heat wave approaching warning model, characterized in that: include: Obtain historical single-station high temperature heat wave event data recorded by meteorological observation stations in the target area, ground meteorological station observation data, and meteorological satellite land surface temperature product data; Performing spatiotemporal matching and data quality control processing on the historical single-station high temperature heat wave event data, the ground meteorological station observation data, and the meteorological satellite land surface temperature product data to construct a high temperature heat wave imminent warning sample set; Based on the high-temperature heat wave approaching warning sample set, a high-temperature heat wave approaching warning model is trained by using a bidirectional long short-term memory neural network and a temporal weight method to obtain a trained high-temperature heat wave approaching warning model; wherein, the high-temperature heat wave approaching warning model is used for the high-temperature heat wave approaching warning in the first preset time period in the future.

2. The method according to claim 1, characterized in that: The historical single-station high temperature heat wave event data, the ground meteorological station observation data and the meteorological satellite land surface temperature product data are subjected to time-space matching and data quality control processing to construct a high temperature heat wave approaching warning sample set, including: Performing data cleaning on the historical single-station high-temperature heat wave event data, and determining a high-temperature heat wave event warning level for each high-temperature heat wave event based on an event intensity index in the historical single-station high-temperature heat wave event data; Performing spatiotemporal matching and data quality control processing on the ground meteorological station observation data to obtain the target ground meteorological station observation data within a second preset time period; Performing time-space matching and data quality control processing on the meteorological satellite land surface temperature product data to obtain target meteorological satellite land surface temperature product data within the second preset time period; Based on the spatial resolution of the target meteorological satellite land surface temperature product data, the event intensity index or high temperature heat wave event warning level of each high temperature heat wave event and the target ground meteorological station observation data are spatially interpolated, and the target ground meteorological station observation data and the target meteorological satellite land surface temperature product data are normalized to obtain a high temperature heat wave imminent warning sample set.

3. The method according to claim 2, characterized in that Determining the high temperature heat wave event warning level of each historical single-station high temperature heat wave event based on the event intensity index in the historical single-station high temperature heat wave event data includes: If the event intensity index is less than a first threshold, determining that the high temperature heat wave event warning level of the historical single-station high temperature heat wave event is the first level; If the event intensity index is greater than the first threshold and less than the second threshold, determining that the high temperature heat wave event warning level of the historical single-station high temperature heat wave event is the second level; If the event intensity index is greater than the second threshold and less than the third threshold, determining that the high temperature heat wave event warning level of the historical single-station high temperature heat wave event is the third level; If the event intensity index is greater than the third threshold, it is determined that the high temperature heat wave event warning level of the historical single-station high temperature heat wave event is the fourth level.

4. The method according to claim 2, characterized in that: Performing spatiotemporal matching and data quality control processing on the ground meteorological station observation data to obtain the target ground meteorological station observation data within a second preset time period includes: Based on the observation time of the ground meteorological station, the ground meteorological station observation data is matched with the time information in the historical single-station high temperature heat wave event data to obtain the ground meteorological station observation data within a second preset time period; wherein the time information includes the start date and end date of the high temperature heat wave event; the second preset time period is from the preset number of days before the start date of the high temperature heat wave event to the end date of the high temperature heat wave event; Data quality control processing is performed on the ground meteorological station observation data within the second preset time period to obtain the target ground meteorological station observation data; wherein the data quality control processing at least includes: climatological limit value check, climate extreme value check, internal consistency check, spatial consistency check and temporal consistency check.

5. The method according to claim 2, characterized in that: Performing time-space matching and data quality control processing on the meteorological satellite land surface temperature product data to obtain target meteorological satellite land surface temperature product data within the second preset time period includes: Performing data analysis on the meteorological satellite land surface temperature product data to obtain the meteorological satellite scanning time and scanning longitude and latitude; Based on the meteorological satellite scanning time, the meteorological satellite land surface temperature product data is matched with the time information in the historical single-station high temperature heat wave event data to obtain the meteorological satellite land surface temperature product data within the second preset time period; Based on the scanning longitude and latitude of the meteorological satellite land surface temperature product data within the second preset time period, searching for the surface temperature data of the ground meteorological observation station within a preset range; The meteorological satellite land surface temperature product data within the second preset time period are matched and compared with the surface temperature data of the ground meteorological observation station in the corresponding time period, and the meteorological satellite land surface temperature product data within the second preset time period are smoothed to obtain the target meteorological satellite land surface temperature product data.

6. The method according to claim 2, characterized in that Based on the high temperature heat wave approaching warning sample set, a high temperature heat wave approaching warning model training is performed using a bidirectional long short-term memory neural network and a temporal weight method, including: The network parameters of the bidirectional long short-term memory neural network are set, and the weight of each time step of the bidirectional long short-term memory neural network is determined through an attention mechanism to obtain an approaching high temperature heat wave warning model; Dividing the high temperature heat wave approach warning sample set into a training sample set and a test sample set; Iteratively training the high temperature heat wave approach warning model based on the training sample set, and evaluating the trained high temperature heat wave approach warning model based on the test sample set; If the evaluation index of the high temperature heat wave approaching warning model meets the preset requirements, a trained high temperature heat wave approaching warning model is obtained; otherwise, the high temperature heat wave approaching warning model is iteratively trained again based on the training sample set until the evaluation index meets the preset requirements; wherein the evaluation index includes at least: mean absolute error and root mean square error.

7. A high temperature heat wave approaching early warning method, characterized in that: include: Obtain ground meteorological station observation data and meteorological satellite land surface temperature product data in the target area; The ground meteorological station observation data and the meteorological satellite land surface temperature product data are input into a pre-trained high temperature heat wave imminent warning model to obtain a high temperature heat wave imminent warning result for the target area in the first preset time period in the future; wherein the high temperature heat wave imminent warning model is obtained by adopting the construction method of the high temperature heat wave imminent warning model as described in any one of claims 1 to 6, and the high temperature heat wave imminent warning result includes: an event intensity index of a high temperature heat wave event or a high temperature heat wave event warning level.

8. A device for constructing a high temperature heat wave approaching warning model, characterized in that: include: The first data acquisition module is used to obtain historical single-station high temperature heat wave event data recorded by meteorological observation stations in the target area, ground meteorological station observation data, and meteorological satellite land surface temperature product data; A sample set construction module is used to perform spatiotemporal matching and data quality control processing on the historical single-station high temperature heat wave event data, the ground meteorological station observation data, and the meteorological satellite land surface temperature product data to construct a high temperature heat wave imminent warning sample set; A model training module is used to train a high-temperature heat wave approaching warning model based on the high-temperature heat wave approaching warning sample set, using a bidirectional long short-term memory neural network and a temporal weight method to obtain a trained high-temperature heat wave approaching warning model; wherein the high-temperature heat wave approaching warning model is used for the high-temperature heat wave approaching warning in the first preset time period in the future.

9. A high temperature heat wave approaching warning device, characterized in that: include: The second data acquisition module is used to obtain ground meteorological station observation data and meteorological satellite land surface temperature product data in the target area; A high temperature heat wave imminent warning module is used to input the ground meteorological station observation data and the meteorological satellite land surface temperature product data into a pre-trained high temperature heat wave imminent warning model to obtain a high temperature heat wave imminent warning result for the first preset time period in the future of the target area; wherein the high temperature heat wave imminent warning model is obtained by adopting the construction method of the high temperature heat wave imminent warning model as described in any one of claims 1 to 6, and the high temperature heat wave imminent warning result includes: an event intensity index of a high temperature heat wave event or a high temperature heat wave event warning level.

10. An electronic device, characterized in that: The method comprises a processor and a memory, wherein the memory stores computer executable instructions that can be executed by the processor, and the processor executes the computer executable instructions to implement the steps of the method according to any one of claims 1 to 6 or the steps of the method according to claim 7.

Images