Rainfall forecasting method and device combining meteorological satellite brightness temperature data and GNSS-PWV (Global Navigation Satellite System-Pulse Width Value)

Abstract

The invention discloses a rainfall forecasting method and device combining meteorological satellite brightness temperature data and GNSS-PWV. The method comprises the following steps: acquiring channel brightness temperature, brightness temperature difference, PWV, PWV increment and PWV increment rate of a rainfall position to be forecasted in a plurality of previous time periods; and matching the channel brightness temperature, the brightness temperature difference, the PWV, the PWV increment and the PWV increment rate according to time, inputting into the trained multilayer neural network model, and outputting a predicted rainfall level in a future preset time period. According to the invention, the brightness temperature data and the GNSS-PWV data are combined, so that the empty report rate and the missing report rate of rainfall forecasting based on single brightness temperature data of a meteorological satellite are reduced, and the precision of rainfall forecasting is improved.

Description

Technical Field

[0001] The present invention relates to the field of rainfall prediction, and in particular to a rainfall forecasting method and device combining meteorological satellite brightness temperature data and GNSS-PWV. Background Art

[0002] In traditional rainfall forecasting methods, the brightness temperature (BT) data provided by meteorological satellites play an important role. These data reflect the response of clouds to microwave radiation emitted by satellite sensors and can be used to identify cloud structure and estimate rainfall intensity. However, brightness temperature reflects more of the surface characteristics of clouds and lacks detailed information on the total amount of water vapor in the atmosphere, resulting in false and missed reports in rainfall forecasts based on single brightness temperature data from meteorological satellites, which reduces the accuracy of rainfall forecasts. Summary of the invention

[0003] In order to overcome the deficiencies of the above-mentioned prior art, the present invention provides a rainfall forecast method and device combining brightness temperature data of meteorological satellite and GNSS-PWV. By combining brightness temperature data and GNSS-PWV data, the false alarm rate and missed alarm rate of rainfall forecast based on single brightness temperature data of meteorological satellite are reduced, and the accuracy of rainfall forecast is improved.

[0004] According to one aspect of the present invention, a rainfall forecasting method combining meteorological satellite brightness temperature data and GNSS-PWV is provided, comprising: Obtain the channel brightness temperature, brightness temperature difference, PWV, PWV increment and PWV increase rate for the previous several time periods at the location of the rainfall to be predicted; After matching the channel brightness temperature, brightness temperature difference, PWV, PWV increment and PWV increase rate according to time, the results are input into the trained multi-layer neural network model to output the predicted rainfall level for a preset period in the future; the training of the multi-layer neural network model includes: Acquire remote sensing data from meteorological satellites in real time, and extract long-term series of channel brightness temperature and brightness temperature difference at the location of rainfall to be predicted from the remote sensing data; Acquire GNSS data, and extract the PWV, PWV increment and PWV increase rate of the long time series at the location of the rainfall to be predicted from the GNSS data; Obtaining long-term rainfall data at the location where rainfall is to be forecasted, and dividing the rainfall data into several rainfall levels; The rainfall level during the rainfall period is matched with the channel brightness temperature, brightness temperature difference, PWV, PWV increment and PWV increase rate of several periods before the rainfall period in time to obtain a sample data set; Construct a multi-layer neural network model, wherein the input layer of the multi-layer neural network model is the channel brightness temperature, brightness temperature difference, PWV, PWV increment and PWV increase rate of several time periods before different times, and the output layer is the rainfall level of the preset time period in the future corresponding to the time; The constructed multi-layer neural network model is trained using the sample data set to obtain a trained multi-layer neural network model.

[0005] As a further technical solution, after obtaining the sample data set, it also includes: The sample data set is divided into a rainfall data set and a non-rainfall data set according to the standard of whether there is rainfall or not; The random downsampling method is used to select the same number of samples from the non-rainfall dataset as the rainfall dataset to form the modeling dataset.

[0006] As a further technical solution, after obtaining the modeling data set, it also includes: Randomly select a part of the modeling data set as the training set, and the remaining data as the test set; The training set and the test set are input into the constructed multi-layer neural network model, the model is trained until convergence, and the model parameters are saved.

[0007] As a further technical solution, the rainfall levels are divided into no rain, light rain, moderate rain, heavy rain and extremely heavy rain.

[0008] As a further technical solution, the multi-layer neural network model is constructed using a back-propagation neural network including three hidden layers.

[0009] As a further technical solution, the channels used to extract brightness temperature include BT11.2 and BT13.3, and the channels used to calculate brightness temperature differences include BTD6.2-11.2, BTD7.3-12.4, BTD10.4-12.4, BTD7.3-10.4, BTD8.6-11.2 and BTD11.2-12.4; among them, BT11.2 and BT13.3 represent the channel brightness temperatures of 11.2μm and 13.3μm respectively; BTD6.2-11.2 represents the brightness temperature difference between BT6.2-BT11.2, and the others are the same.

[0010] According to one aspect of the present invention, there is provided a rainfall forecasting device combining meteorological satellite brightness temperature data and GNSS-PWV, comprising: A data acquisition module is used to obtain the channel brightness temperature, brightness temperature difference, PWV, PWV increment and PWV increase rate of the previous several time periods at the location of the rainfall to be predicted; The rainfall forecast module is used to input the channel brightness temperature, brightness temperature difference, PWV, PWV increment and PWV increase rate into the trained multi-layer neural network model after matching according to time, and output the predicted rainfall level for a preset period in the future; the training of the multi-layer neural network model includes: Acquire remote sensing data from meteorological satellites in real time, and extract long-term series of channel brightness temperature and brightness temperature difference at the location of rainfall to be predicted from the remote sensing data; Acquire GNSS data, and extract the PWV, PWV increment and PWV increase rate of the long time series at the location of the rainfall to be predicted from the GNSS data; Obtaining long-term rainfall data at the location where rainfall is to be forecasted, and dividing the rainfall data into several rainfall levels; The rainfall level during the rainfall period is matched with the channel brightness temperature, brightness temperature difference, PWV, PWV increment and PWV increase rate of several periods before the rainfall period in time to obtain a sample data set; Construct a multi-layer neural network model, wherein the input layer of the multi-layer neural network model is the channel brightness temperature, brightness temperature difference, PWV, PWV increment and PWV increase rate of several time periods before different times, and the output layer is the rainfall level of the preset time period in the future corresponding to the time; The constructed multi-layer neural network model is trained using the sample data set to obtain a trained multi-layer neural network model.

[0011] According to one aspect of the present invention, there is provided a ground observation device, comprising a processor and a memory, wherein the memory stores program instructions executed by the processor, and the processor calls the program instructions to execute the steps of the rainfall forecasting method combining meteorological satellite brightness temperature data and GNSS-PWV.

[0012] According to one aspect of the present specification, a non-transitory computer-readable storage medium is provided, wherein the non-transitory computer-readable storage medium stores computer instructions, wherein the computer instructions enable the computer to execute the steps of the rainfall forecasting method combining meteorological satellite brightness temperature data and GNSS-PWV.

[0013] Compared with the prior art, the present invention has the following beneficial effects: 1. The present invention utilizes the brightness temperature data of meteorological satellites, combined with the GNSS-PWV correlation factor, and uses a model constructed by a neural network (NN) to predict rainfall, thereby reducing the false alarm rate and missed alarm rate of rainfall forecast based on single brightness temperature data of meteorological satellites and improving the accuracy of rainfall forecast.

[0014] 2. The present invention constructs a multi-layer neural network model, which uses data fusion technology to combine brightness temperature data and GNSS-PWV data, thereby reducing the false alarm rate and missed alarm rate of rainfall forecasts based on single brightness temperature data of meteorological satellites, and improving the overall accuracy of rainfall forecasts. Especially in areas with complex terrain or scarce observation data, the model can more effectively capture and analyze the dynamics of water vapor in the atmosphere, thereby providing more accurate single-point short-term rainfall forecasts, which can cope with possible extreme rainfall events. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, a brief introduction is given below to the drawings used in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0016] Figure 1 A schematic flow chart of a rainfall forecasting method combining meteorological satellite brightness temperature data and GNSS-PWV provided in an embodiment of the present invention.

[0017] Figure 2 A schematic diagram of the structure of a rainfall forecasting device combining meteorological satellite brightness temperature data and GNSS-PWV provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0018] The present invention aims at the problem of false alarm and under-reporting in the existing rainfall forecast based on single brightness temperature data of meteorological satellite, and considering that the atmospheric precipitable water volume (PWV) of GNSS is an important meteorological parameter that can be used for rainfall forecast. Therefore, a rainfall forecast method combining brightness temperature data of meteorological satellite and GNSS-PWV is provided. By combining brightness temperature data and GNSS-PWV data, the false alarm rate and under-reporting rate of rainfall forecast based on single brightness temperature data of meteorological satellite are reduced, thereby improving the accuracy of rainfall forecast.

[0019] It should be noted that the technologies involved in the specification of the present invention are all well-known technologies unless otherwise specified.

[0020] The terms "including" and "having" and any variations thereof in the specification and claims of the present invention and the above-mentioned drawings are intended to cover non-exclusive inclusions, for example, a process, method, system, product or apparatus comprising a series of steps or units is not necessarily limited to the steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to these processes, methods, products or apparatuses.

[0021] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments in 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. In addition, the technical features in the various embodiments or single embodiments provided by the present invention are arbitrarily combined with each other to form a new technical solution. This combination is not restricted by the sequence of steps and / or the structural composition mode, but must be based on the ability of ordinary technicians in this field to achieve it. When the combination of technical solutions is contradictory or cannot be achieved, it should be considered that this combination of technical solutions does not exist and is not within the scope of protection required by the present invention.

[0022] The embodiment of the present invention provides a rainfall forecasting method combining meteorological satellite brightness temperature data and GNSS-PWV, such as Figure 1 As shown, it includes: obtaining the channel brightness temperature, brightness temperature difference, PWV, PWV increment and PWV increase rate of the previous several time periods at the location of the rainfall to be predicted; matching the channel brightness temperature, brightness temperature difference, PWV, PWV increment and PWV increase rate according to time, inputting the trained multi-layer neural network model, and outputting the predicted rainfall level of the future preset time period. The embodiment of the present invention reduces the false alarm rate and missed alarm rate of rainfall forecast based on single brightness temperature data of meteorological satellite by combining brightness temperature data and GNSS-PWV data, thereby improving the accuracy of rainfall forecast.

[0023] The training of the multi-layer neural network model described in the embodiment of the present invention includes data acquisition, data processing, and building and training the multi-layer neural network model.

[0024] The data acquisition steps include: (1) Real-time acquisition of remote sensing data from meteorological satellites, usually with a time resolution of 10 minutes. Extract the long-term channel brightness temperature (BT) at the location of the rainfall forecast from the remote sensing data from meteorological satellites, and calculate the corresponding brightness temperature difference (BTD).

[0025] (2) The long-term PWV, PWV increment and PWV rate data are collected at the GNSS station at the location where the rainfall is to be predicted, usually with a time resolution of 10 min.

[0026] (3) Obtain long-term rainfall data at the location where rainfall is to be forecast from ERA5 data, with an original time resolution of 1 hour.

[0027] The steps of data processing include: (1) The rainfall is divided into the following rainfall levels according to the World Meteorological Organization's classification standards for different rainfall levels: no rain (<0.1 mm / h), light rain (0.1-2.5 mm / h), moderate rain (2.5-9.9 mm / h), heavy rain (10.0-49.9 mm / h), and extremely heavy rain (≧50.0 mm / h).

[0028] (2) The rainfall level during the rainfall period is matched with the forecast factors (PWV, PWV increment, PWV increase rate, BT, BTD) of a certain period of time (e.g., 6 hours) before the rainfall period to obtain a sample data set.

[0029] (3) The sample data set is divided into a rainfall data set and a non-rainfall data set according to the presence or absence of rainfall.

[0030] (4) The random downsampling method is used to select the same number of samples as the rainfall data from the non-rainfall data to form the modeling data set.

[0031] The steps of constructing and training a multi-layer neural network model include: (1) The prediction factors (PWV, PWV increment, PWV increase rate, BT, BTD) of a period of time before different moments (e.g., 6 hours) are used as input features, and the rainfall level of a period of time in the future (e.g., 1 hour) of the corresponding moment is used as the output result.

[0032] (2) During the model training process, a portion of the modeling data set (for example, 70% or 80%) is randomly selected as the training set, and the remaining data (for example, 30% or 20%) is used as the test set.

[0033] (3) Input the training set and test set into the model, train the model until convergence, and save the model parameters.

[0034] After completing the model training, the trained model is used for rainfall forecasting. First, the forecast factors (PWV, PWV increment, PWV increase rate, BT, BTD) of the previous hours at the location of the rainfall to be forecasted are collected according to the data acquisition step. Then, the data is processed according to the data processing step and input into the trained model to obtain the predicted rainfall level for a period of time in the future (such as 1 hour).

[0035] As a preferred embodiment, the embodiment of the present invention introduces an embodiment of the Wuhan (WUH2) GNSS station rainfall forecast based on the Japanese meteorological remote sensing satellite Himawari-8 and the back-propagation neural network model.

[0036] (1) Data acquisition: i. Download the multi-band data of Himawari-8 in 2020-2021 with a time resolution of 10min. Extract the long-term channel brightness temperature data at the WUH2 station from the remote sensing data of Himawari-8, and calculate the corresponding brightness temperature difference. Including BT11.2, BT13.3, BTD6.2-11.2, BTD7.3-12.4, BTD10.4-12.4, BTD7.3-10.4, BTD8.6-11.2, BTD11.2-12.4 (BT11.2, represents the brightness temperature of the 11.2μm channel; BTD6.2-11.2, represents the brightness temperature difference between BT6.2-BT11.2, and the others are the same).

[0037] ii. Calculate the PWV, PWV increment, and PWV increase rate of the WUH2 station from 2020 to 2021 with a time resolution of 10 min.

[0038] iii. Download the ERA5 data for 2020-2021 (the fifth-generation global atmospheric reanalysis dataset launched by the European Centre for Medium-Range Weather Forecasts) and obtain the rainfall data at WUH2 from it. The original time resolution is 1 hour.

[0039] (2) Data processing: i. The rainfall is divided into the following rainfall levels according to the World Meteorological Organization's standards for different rainfall levels: no rain (<0.1 mm / h), light rain (0.1-2.5 mm / h), moderate rain (2.5-9.9 mm / h), heavy rain (10.0-49.9 mm / h), and extremely heavy rain (≧50.0 mm / h).

[0040] ii. Match the rainfall level during the rainfall period with the forecast factors (PWV, PWV increment, PWV increase rate, the brightness temperature BT, the brightness temperature difference BTD) 6 hours before the rainfall period to obtain a sample data set.

[0041] iii. The sample data set is divided into a rainfall data set and a non-rainfall data set according to the standard of whether there is rainfall or not. The random downsampling method is used to select the same number of samples from the non-rainfall data as the rainfall data to form the modeling data set.

[0042] (3) Construct a backward neural network (BPNN) model with 3 hidden layers: i. Use the forecast factors (PWV, PWV increment, PWV increase rate, the above BT, the above BTD) of the previous 6 hours at different times as the input layer, and the rainfall level in the next hour at the corresponding time as the output layer, and output no rain (0), light rain (1), moderate rain (2), heavy rain (3), and extremely heavy rain (4).

[0043] ii. During the model training process, 80% of the modeling data set is randomly selected as the training set, and the remaining 20% ​​is used as the test set.

[0044] iii. Input the training set and test set into BPNN, train the model until convergence, and save the model parameters.

[0045] (4) Use the trained model to make rainfall forecasts. First, the forecast factors (PWV, PWV increment, PWV increase rate, the above BT, the above BTD) of the previous 6 hours at the WUH2 station are collected according to the data acquisition steps. Then, the data is processed according to the data processing steps and input into the trained model to obtain the forecast rainfall level within 1 hour at that moment.

[0046] The implementation basis of each embodiment of the present invention is to implement programmed processing through a device with a processor function. Therefore, in engineering practice, the technical solutions and functions of each embodiment of the present invention are encapsulated into various modules. Based on this reality, on the basis of the above embodiments, an embodiment of the present invention provides a rainfall forecasting device combining meteorological satellite brightness temperature data and GNSS-PWV, which is used to execute the rainfall forecasting method combining meteorological satellite brightness temperature data and GNSS-PWV in the above method embodiment.

[0047] See also Figure 2 The device comprises: a data acquisition module, which is used to acquire the channel brightness temperature, brightness temperature difference, PWV, PWV increment and PWV increase rate of the previous several time periods at the rainfall location to be predicted; a rainfall forecasting module, which is used to input the channel brightness temperature, brightness temperature difference, PWV, PWV increment and PWV increase rate into a trained multi-layer neural network model after matching them according to time, and output the predicted rainfall level for a preset time period in the future; the training of the multi-layer neural network model comprises: acquiring remote sensing data of meteorological satellites in real time, and extracting the channel brightness temperature and brightness temperature difference of the long time series at the rainfall location to be predicted from the remote sensing data; acquiring GNSS data, and extracting the PWV of the long time series at the rainfall location to be predicted from the GNSS data , PWV increment and PWV increase rate; obtain long-term series rainfall data at the rainfall location to be predicted, and divide the rainfall data into several rainfall levels; match the rainfall level of a rainfall period with the channel brightness temperature, brightness temperature difference, PWV, PWV increment and PWV increase rate of several time periods before the rainfall period in time to obtain a sample data set; construct a multi-layer neural network model, the input layer of the multi-layer neural network model is the channel brightness temperature, brightness temperature difference, PWV, PWV increment and PWV increase rate of several time periods before different moments, and the output layer is the rainfall level of a preset time period in the future corresponding to the moment; use the sample data set to train the constructed multi-layer neural network model to obtain a trained multi-layer neural network model.

[0048] The rainfall forecasting device combining meteorological satellite brightness temperature data and GNSS-PWV provided in the embodiment of the present invention adopts Figure 2 By combining brightness temperature data and GNSS-PWV data, the false alarm rate and missed alarm rate of rainfall forecast based on single brightness temperature data of meteorological satellite are reduced, and the accuracy of rainfall forecast is improved.

[0049] It should be noted that the device embodiments provided by the present invention are used to implement the methods in the above method embodiments as well as the methods in other method embodiments provided by the present invention. The only difference is that the corresponding functional modules are set, and the principle is basically the same as the principle of the above device embodiments provided by the present invention. As long as the technical personnel in the field refer to the specific technical solutions in other method embodiments on the basis of the above device embodiments, obtain the corresponding technical means and the technical solutions composed of these technical means by combining the technical features, and on the premise of ensuring the practicality of the technical solutions, improve the modules in the above device embodiments to obtain the corresponding device class embodiments, which are used to implement the methods in other method class embodiments. For example: Based on the content of the above device embodiment, as a preferred embodiment, the rainfall forecasting device combining meteorological satellite brightness temperature data and GNSS-PWV provided in the embodiment of the present invention, after obtaining the sample data set, further includes: The sample data set is divided into a rainfall data set and a non-rainfall data set according to the standard of whether there is rainfall or not; The random downsampling method is used to select the same number of samples from the non-rainfall dataset as the rainfall dataset to form the modeling dataset.

[0050] Based on the content of the above device embodiment, as a preferred embodiment, the rainfall forecasting device combining meteorological satellite brightness temperature data and GNSS-PWV provided in the embodiment of the present invention, after obtaining the modeling data set, further includes: Randomly select a part of the modeling data set as the training set, and the remaining data as the test set; The training set and the test set are input into the constructed multi-layer neural network model, the model is trained until convergence, and the model parameters are saved.

[0051] Based on the contents of the above-mentioned device embodiments, as a preferred embodiment, the rainfall forecasting device combining meteorological satellite brightness temperature data and GNSS-PWV provided in the embodiments of the present invention further includes: dividing the rainfall levels into no rain, light rain, moderate rain, heavy rain and extremely heavy rain in turn.

[0052] Based on the contents of the above-mentioned device embodiments, as a preferred embodiment, in the rainfall forecasting device combining meteorological satellite brightness temperature data and GNSS-PWV provided in the embodiments of the present invention, the multi-layer neural network model is constructed using a back-propagation neural network including three hidden layers.

[0053] Based on the contents of the above-mentioned device embodiments, as a preferred embodiment, in the rainfall forecasting device combining meteorological satellite brightness temperature data and GNSS-PWV provided in the embodiments of the present invention, the channels for extracting brightness temperature include BT11.2 and BT13.3, and the channels for calculating brightness temperature differences include BTD6.2-11.2, BTD7.3-12.4, BTD10.4-12.4, BTD7.3-10.4, BTD8.6-11.2 and BTD11.2-12.4 (BT11.2 represents the brightness temperature of the 11.2μm channel; BTD6.2-11.2 represents the brightness temperature difference between BT6.2-BT11.2, and the others are the same).

[0054] Based on the same inventive concept as the above-mentioned embodiment, an embodiment of the present invention also provides a ground observation device, including a processor and a memory, wherein the memory stores program instructions executed by the processor, and the processor calls the program instructions to execute the steps of the rainfall forecasting method combining meteorological satellite brightness temperature data and GNSS-PWV.

[0055] When using the ground observation equipment provided by the embodiment of the present invention to make rainfall forecasts, the false alarm rate and missed alarm rate of rainfall forecasts based on single brightness temperature data of meteorological satellites can be reduced by combining brightness temperature data and GNSS-PWV data, thereby improving the accuracy of rainfall forecasts.

[0056] Based on the same inventive concept as the above embodiment, the embodiment of the present invention further provides a non-transitory computer-readable storage medium, wherein the non-transitory computer-readable storage medium stores computer instructions, wherein the computer instructions enable the computer to execute the steps of the rainfall forecasting method combining meteorological satellite brightness temperature data and GNSS-PWV, including: Obtain the channel brightness temperature, brightness temperature difference, PWV, PWV increment and PWV increase rate for the previous several time periods at the location of the rainfall to be predicted; After matching the channel brightness temperature, brightness temperature difference, PWV, PWV increment and PWV increase rate according to time, they are input into the trained multi-layer neural network model to output the predicted rainfall level for a preset period of time in the future; wherein the training of the multi-layer neural network model includes: acquiring remote sensing data from meteorological satellites in real time, and extracting the long-term series of channel brightness temperature and brightness temperature difference at the location of rainfall to be predicted from the remote sensing data; acquiring GNSS data, and extracting the long-term series of PWV, PWV increment and PWV increase rate at the location of rainfall to be predicted from the GNSS data; acquiring the long-term series of rainfall at the location of rainfall to be predicted The method comprises the following steps: obtaining rainfall data, dividing the rainfall data into several rainfall levels; matching the rainfall level in a rainfall period with the channel brightness temperature, brightness temperature difference, PWV, PWV increment and PWV increase rate in several time periods before the rainfall period in time, and obtaining a sample data set; constructing a multi-layer neural network model, wherein the input layer of the multi-layer neural network model is the channel brightness temperature, brightness temperature difference, PWV, PWV increment and PWV increase rate in several time periods before different moments, and the output layer is the rainfall level in a preset time period in the future corresponding to the moment; and training the constructed multi-layer neural network model using the sample data set to obtain a trained multi-layer neural network model.

[0057] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the technical solutions of the embodiments of the present invention.

Claims

1. A rainfall forecast method combining meteorological satellite brightness temperature data and GNSS-PWV, characterized in that: include: Obtain the channel brightness temperature, brightness temperature difference, PWV, PWV increment and PWV increase rate for the previous several time periods at the location of the rainfall to be predicted; After matching the channel brightness temperature, brightness temperature difference, PWV, PWV increment and PWV increase rate according to time, the results are input into the trained multi-layer neural network model to output the predicted rainfall level for a preset period in the future; The training of the multi-layer neural network model includes: Acquire remote sensing data from meteorological satellites in real time, and extract long-term series of channel brightness temperature and brightness temperature difference at the location of rainfall to be predicted from the remote sensing data; Acquire GNSS data, and extract the PWV, PWV increment and PWV increase rate of the long time series at the location of the rainfall to be predicted from the GNSS data; Obtaining long-term rainfall data at the location where rainfall is to be forecasted, and dividing the rainfall data into several rainfall levels; The rainfall level during the rainfall period is matched with the channel brightness temperature, brightness temperature difference, PWV, PWV increment and PWV increase rate of several periods before the rainfall period in time to obtain a sample data set; Construct a multi-layer neural network model, wherein the input layer of the multi-layer neural network model is the channel brightness temperature, brightness temperature difference, PWV, PWV increment and PWV increase rate of several time periods before different times, and the output layer is the rainfall level of the preset time period in the future corresponding to the time; The constructed multi-layer neural network model is trained using the sample data set to obtain a trained multi-layer neural network model.

2. The rainfall forecasting method combining meteorological satellite brightness temperature data and GNSS-PWV according to claim 1 is characterized in that: After getting the sample data set, it also includes: The sample data set is divided into a rainfall data set and a non-rainfall data set according to the standard of whether there is rainfall or not; The random downsampling method is used to select the same number of samples from the non-rainfall dataset as the rainfall dataset to form the modeling dataset.

3. The rainfall forecasting method combining meteorological satellite brightness temperature data and GNSS-PWV according to claim 1 is characterized in that: After obtaining the modeling data set, it also includes: Randomly select a part of the modeling data set as the training set, and the remaining data as the test set; The training set and the test set are input into the constructed multi-layer neural network model, the model is trained until convergence, and the model parameters are saved.

4. The rainfall forecasting method combining meteorological satellite brightness temperature data and GNSS-PWV according to claim 1 is characterized in that: The rainfall levels are divided into no rain, light rain, moderate rain, heavy rain and extremely heavy rain.

5. The rainfall forecasting method combining meteorological satellite brightness temperature data and GNSS-PWV according to claim 1 is characterized in that: The multi-layer neural network model is constructed using a back-propagation neural network including three hidden layers.

6. The rainfall forecasting method combining meteorological satellite brightness temperature data and GNSS-PWV according to claim 1 is characterized in that: The channels used to extract brightness temperature include BT11.2 and BT13.3, and the channels used to calculate brightness temperature difference include BTD6.2-11.2, BTD7.3-12.4, BTD10.4-12.4, BTD7.3-10.4, BTD8.6-11.2 and BTD11.2-12.4; among them, BT11.2 and BT13.3 represent the channel brightness temperatures of 11.2μm and 13.3μm respectively; BTD6.2-11.2 represents the brightness temperature difference between BT6.2-BT11.2, and the others are the same.

7. A rainfall forecasting device combining meteorological satellite brightness temperature data and GNSS-PWV, characterized in that: include: A data acquisition module is used to obtain the channel brightness temperature, brightness temperature difference, PWV, PWV increment and PWV increase rate of the previous several time periods at the location of the rainfall to be predicted; A rainfall forecast module is used to match the channel brightness temperature, brightness temperature difference, PWV, PWV increment and PWV increase rate according to time, input them into the trained multi-layer neural network model, and output the predicted rainfall level for a preset period in the future; The training of the multi-layer neural network model includes: Acquire remote sensing data from meteorological satellites in real time, and extract long-term series of channel brightness temperature and brightness temperature difference at the location of rainfall to be predicted from the remote sensing data; Acquire GNSS data, and extract the PWV, PWV increment and PWV increase rate of the long time series at the location of the rainfall to be predicted from the GNSS data; Obtaining long-term rainfall data at the location where rainfall is to be forecasted, and dividing the rainfall data into several rainfall levels; The rainfall level during the rainfall period is matched with the channel brightness temperature, brightness temperature difference, PWV, PWV increment and PWV increase rate of several periods before the rainfall period in time to obtain a sample data set; Construct a multi-layer neural network model, wherein the input layer of the multi-layer neural network model is the channel brightness temperature, brightness temperature difference, PWV, PWV increment and PWV increase rate of several time periods before different times, and the output layer is the rainfall level of the preset time period in the future corresponding to the time; The constructed multi-layer neural network model is trained using the sample data set to obtain a trained multi-layer neural network model.

8. A ground observation device, characterized in that: It comprises a processor and a memory, wherein the memory stores program instructions executed by the processor, and the processor calls the program instructions to execute the steps of the rainfall forecasting method combining meteorological satellite brightness temperature data and GNSS-PWV as described in any one of claims 1 to 6.

9. A non-transitory computer readable storage medium, characterized in that: The non-transitory computer-readable storage medium stores computer instructions, and the computer instructions enable the computer to execute the steps of the rainfall forecasting method combining meteorological satellite brightness temperature data and GNSS-PWV as claimed in any one of claims 1 to 6.

Images