Rainfall forecasting method and device based on combined GNSS-PWV and meteorological satellite remote sensing

Abstract

The present invention discloses a rainfall forecasting method and device combining GNSS-PWV and meteorological satellite remote sensing, the method comprising: determining PWV, PWV increment and PWV increase rate in a certain time period before the rainfall station to be predicted, and extracting channel data in a certain time period before the rainfall station to be predicted from meteorological satellite remote sensing data; matching the PWV, PWV increment and PWV increase rate with the channel data by time, inputting the trained multi-layer neural network model, and outputting the rainfall level in a future preset time period. The present invention integrates GNSS-PWV data and meteorological satellite remote sensing data, uses a neural network to build a model for rainfall forecasting, reduces the false alarm rate of rainfall forecasting based on PWV, and thus improves the accuracy of rainfall forecasting.

Description

Technical Field

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

[0002] Precipitation forecasting is crucial for agriculture, urban planning, disaster prevention and other fields. There are two main existing rainfall forecasting methods: the first is based on numerical weather forecast models, which are suitable for large-area forecasts, but because the forecast area is divided into large grids, the forecast is not detailed enough, especially in mountainous areas with complex terrain, the accuracy is low. At the same time, the calculation consumes a lot of resources and has low calculation efficiency. The second method is the single-station forecasting method. For example, rainfall forecasting is based on PWV (Precipitable Water Vapor) and its change amount and change rate based on GNSS (Global Navigation Satellite System). Although this method is more effective in water vapor monitoring, it does not make full use of cloud observations because GNSS is not sensitive to cloud observations. Summary of the invention

[0003] In order to overcome the shortcomings of the above-mentioned prior art, the present invention provides a rainfall forecast method and device combining GNSS-PWV and meteorological satellite remote sensing. By integrating GNSS-PWV data and meteorological satellite remote sensing data, a neural network is used to build a model for rainfall forecasting, thereby reducing the false alarm rate of PWV-based rainfall forecasts and improving the accuracy of rainfall forecasts.

[0004] According to one aspect of the present invention, there is provided a rainfall forecasting method combining GNSS-PWV and meteorological satellite remote sensing, comprising:

[0005] Determine the PWV, PWV increment and PWV increase rate in a certain time period before the rainfall site to be predicted, and extract channel data in a certain time period before the rainfall site to be predicted from the meteorological satellite remote sensing data;

[0006] After matching the PWV, PWV increment and PWV increase rate with the channel data according to time, the PWV, PWV increment and PWV increase rate are input into the trained multi-layer neural network model to output the rainfall level in the future preset time period; wherein the training of the multi-layer neural network model includes:

[0007] Calculate the PWV, PWV increment, PWV increase rate and download precipitation data of the long time series of GNSS co-located meteorological stations, and divide the precipitation into several rainfall levels;

[0008] Acquire remote sensing data from meteorological satellites in real time, and extract long-time series channel data from GNSS and meteorological stations;

[0009] Matching the PWV, PWV increment, PWV increase rate and precipitation with the channel data in time to form a data set;

[0010] Constructing a multi-layer neural network model, wherein the input layer of the multi-layer neural network model is PWV, PWV increment, PWV increase rate and channel data 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;

[0011] The multi-layer neural network model is trained using the constructed data set to obtain a trained multi-layer neural network model.

[0012] As a further technical solution, after completing the data matching, it also includes:

[0013] The matched data are divided into rainfall data and non-rainfall data according to the standard of whether there is rainfall in the preset time period;

[0014] Balancing the rainfall data and the non-rainfall data to form a data set;

[0015] Divide the dataset into training and validation sets according to the required ratio.

[0016] As a further technical solution, the channel data includes a water vapor channel and a long-wave infrared channel.

[0017] As a further technical solution, the multi-layer neural network model is constructed using a back-propagation neural network.

[0018] As a further technical solution, the rainfall levels are divided into no rain, light rain, moderate rain, heavy rain and extremely heavy rain.

[0019] According to one aspect of the present invention, there is provided a rainfall forecasting device combining GNSS-PWV and meteorological satellite remote sensing, comprising:

[0020] A data acquisition module is used to determine the PWV, PWV increment and PWV increase rate in a certain period of time before the rainfall site to be predicted, and to extract channel data in a certain period of time before the rainfall site to be predicted from meteorological satellite remote sensing data;

[0021] A rainfall forecast module is used to match the PWV, PWV increment and PWV increase rate with the channel data according to time, input the trained multi-layer neural network model, and output the rainfall level within a preset time period in the future; wherein the training of the multi-layer neural network model includes:

[0022] Calculate the PWV, PWV increment, PWV increase rate and download precipitation data of the long time series of GNSS co-located meteorological stations, and divide the precipitation into several rainfall levels;

[0023] Acquire remote sensing data from meteorological satellites in real time, and extract long-time series channel data from GNSS and meteorological stations;

[0024] Matching the PWV, PWV increment, PWV increase rate and precipitation with the channel data in time to form a data set;

[0025] Constructing a multi-layer neural network model, wherein the input layer of the multi-layer neural network model is PWV, PWV increment, PWV increase rate and channel data 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;

[0026] The multi-layer neural network model is trained using the constructed data set to obtain a trained multi-layer neural network model.

[0027] According to one aspect of the present invention, there is provided a ground station equipped with the rainfall forecasting device combining GNSS-PWV and meteorological satellite remote sensing.

[0028] According to one aspect of the present invention, there is provided a computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the rainfall forecasting method combining GNSS-PWV and meteorological satellite remote sensing when executing the computer program.

[0029] 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 GNSS-PWV and meteorological satellite remote sensing.

[0030] Compared with the prior art, the present invention has the following beneficial effects:

[0031] (1) The present invention integrates GNSS-PWV data and meteorological satellite remote sensing data, and uses a neural network (NN) to build a model for rainfall prediction, thereby reducing the false alarm rate of rainfall forecast based on PWV and improving the accuracy of rainfall forecast.

[0032] (2) The present invention makes up for the lack of prediction ability of numerical weather forecast models for single-point short-term rainfall events, especially in areas with complex terrain or scarce traditional observation data. By utilizing satellite remote sensing data, it provides an accurate single-point short-term rainfall prediction method that can cope with possible extreme rainfall events. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] 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.

[0034] Figure 1 A schematic flow chart of a rainfall forecasting method combining GNSS-PWV and meteorological satellite remote sensing provided in an embodiment of the present invention.

[0035] Figure 2 A schematic diagram of the training process of a multi-layer neural network model provided in an embodiment of the present invention.

[0036] Figure 3 A schematic diagram of the structure of a rainfall forecasting device combining GNSS-PWV and meteorological satellite remote sensing provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0037] The present invention aims at the current single-station forecasting method, in which GNSS is insensitive to cloud observation and fails to make full use of cloud observation. By combining the cloud information captured by meteorological remote sensing satellites, the present invention uses these data and PWV data as input for rainfall forecasting, thereby reducing the false alarm rate of PWV-based forecasts and improving the accuracy of rainfall forecasts.

[0038] It should be noted that the technologies involved in the embodiments of the present invention are all well-known technologies unless otherwise specified.

[0039] 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.

[0040] 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. 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.

[0041] The embodiment of the present invention provides a rainfall forecasting method combining GNSS-PWV and meteorological satellite remote sensing, such as Figure 1 As shown, including:

[0042] Step 1, determining the PWV, PWV increment and PWV increase rate in a certain time period before the rainfall station to be predicted, and extracting the channel data in a certain time period before the rainfall station to be predicted from the meteorological satellite remote sensing data;

[0043] Step 2, after matching the PWV, PWV increment and PWV increase rate with the channel data according to time, input them into the trained multi-layer neural network model to output the rainfall level in the future preset time period.

[0044] The embodiment of the present invention integrates GNSS-PWV data and meteorological satellite remote sensing data, and uses a neural network (NN) to build a model for rainfall prediction, thereby reducing the false alarm rate of rainfall forecast based on PWV, thereby improving the accuracy of rainfall forecast.

[0045] Wherein, the training of the multi-layer neural network model is as follows: Figure 2 As shown, including:

[0046] Step 2.1, calculate the PWV, PWV increment, PWV increase rate and download precipitation data of the long time series of the GNSS co-located meteorological station, and divide the precipitation into several rainfall levels.

[0047] Specifically, this step calculates the PWV, PWV increment, PWV increase rate and downloaded precipitation data of the long time series at the GNSS co-located meteorological station, and divides the precipitation into the following rainfall levels according to the World Meteorological Organization's classification standards for different precipitation 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).

[0048] Step 2.2, obtain the remote sensing data of the meteorological satellite in real time, and extract the long-term channel data of the GNSS and meteorological stations from the remote sensing data.

[0049] Alternatively, remote sensing data from meteorological satellites can be acquired in real time, typically with a temporal resolution of 10 minutes and a spatial resolution of 1 km–4 km.

[0050] Furthermore, after acquiring the remote sensing data, remote sensing image processing software is used to extract long-time series channel data at GNSS and meteorological stations from the satellite remote sensing data, including water vapor channel data and long-wave infrared channel data.

[0051] Step 2.3, matching the PWV, PWV increment, PWV increase rate and precipitation with the channel data in time to form a data set.

[0052] Specifically, after PWV, PWV increment, PWV increase rate, precipitation and the above channel data are matched by time, the matched data are divided into rainfall data and non-rainfall data according to the standard of whether there is rainfall within a preset time period (such as 1 hour).

[0053] Furthermore, the balance between rainfall data and non-rainfall data is achieved through the random downsampling method, that is, the data sample size of the same size as the rainfall data is randomly selected from the non-rainfall data to form a data set together, and the data set is divided into a training set and a validation set.

[0054] Step 2.4, construct a multi-layer neural network model, the input layer of the multi-layer neural network model is the PWV, PWV increment, PWV increase rate and channel data in several time periods before different moments (such as 6 hours), and the output layer is the rainfall level in a preset time period (such as 1 hour) in the future of the corresponding moment.

[0055] Step 2.5, use the constructed data set to train the multi-layer neural network model to obtain a trained multi-layer neural network model.

[0056] Specifically, the training set and validation set divided previously are input into the multi-layer neural network model for training, and the model parameters are saved.

[0057] In an embodiment of the present invention, when using a trained multi-layer neural network model to predict rainfall, the PWV, PWV increment and PWV increase rate of the point where the rainfall is to be predicted are first determined, and then the above-mentioned channel data of the site in the previous several hours (such as 6 hours) are extracted from the meteorological satellite remote sensing data, and the data are matched by time and substituted into the trained model to obtain the rainfall level in the future period (such as 1 hour).

[0058] As a preferred embodiment, the present invention introduces an embodiment of building a model using a back propagation neural network based on a GNSS co-located meteorological station and a meteorological remote sensing satellite Himawari-8 (the full disk observation area is 60°S-60°N, 80°E-160°W), and the specific steps include:

[0059] (1) PWV, PWV increment, and PWV change rate of multiple GNSS co-located meteorological stations from 2010 to 2011 were calculated, and the precipitation from 2010 to 2011 of the corresponding stations was downloaded. The precipitation was classified into the following rainfall levels according to the World Meteorological Organization's classification standards for different precipitation 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). The temporal resolution of PWV data and precipitation data is 10 minutes.

[0060] (2) Download Himawari-8 real-time multi-band data with a temporal resolution of 10 min and a spatial resolution of 1 km.

[0061] (3) Extract long-term channel data of GNSS co-located meteorological stations from satellite remote sensing data, including water vapor channel (6.2μm, 6.9μm, 7.3μm) and long-wave infrared channel (8.6μm, 9.6μm, 10.4μm, 11.2μm, 12.4μm, 13.3μm).

[0062] (4) Match PWV, PWV increment, PWV increase rate, precipitation and the above 9 channel data by time. The matched data are divided into rainfall data and non-rainfall data according to the standard of whether there is rainfall within 1 hour.

[0063] (5) The rainfall data and non-rainfall data are balanced by random downsampling, that is, data of the same size as the rainfall data are randomly selected from the non-rainfall data to form a data set. The training set and validation set are randomly selected (here the ratio of the data set is set to 8:2).

[0064] (6) A backward neural network (BPNN) model with three hidden layers was constructed. The input layer is the channel data, PWV, PWV increment, and PWV increase rate of the six hours before the above different times. The output layer is the rainfall level within one hour after the corresponding time, and the output is no rain (0), light rain (1), moderate rain (2), heavy rain (3), and extremely heavy rain (4).

[0065] (7) Input the training set and validation set into BPNN, train the model until convergence, and save the model parameters.

[0066] When using the model constructed by the above embodiment to predict rainfall, first determine the PWV, PWV increment, and PWV increase rate of the point where rainfall is to be predicted in the previous 6 hours. Then, extract the above channel data of the site in the previous 6 hours from the meteorological satellite remote sensing data, match the data by time, and substitute it into the trained model to obtain the rainfall level in the next 1 hour.

[0067] 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 GNSS-PWV and meteorological satellite remote sensing, which is used to execute the rainfall forecasting method combining GNSS-PWV and meteorological satellite remote sensing in the above method embodiment.

[0068] See also Figure 3 The device comprises: a data acquisition module, which is used to determine the PWV, PWV increment and PWV increase rate in a certain time period before the rainfall station to be predicted, and extract the channel data in a certain time period before the rainfall station to be predicted from the meteorological satellite remote sensing data; a rainfall forecast module, which is used to match the PWV, PWV increment and PWV increase rate with the channel data according to time, input them into the trained multi-layer neural network model, and output the rainfall level in the future preset time period; wherein the training of the multi-layer neural network model comprises: obtaining the PWV, PWV increment, PWV increase rate and precipitation of the long time series of the GNSS co-located meteorological station, and inputting the PWV, PWV increment and PWV increase rate into the trained multi-layer neural network model; and outputting the rainfall level in the future preset time period. The precipitation is divided into several rainfall levels; remote sensing data of meteorological satellites are acquired in real time, and channel data of long time series at GNSS and parallel-site meteorological stations are extracted from the remote sensing data; the PWV, PWV increment, PWV increase rate and precipitation are matched with the channel data in time to form a data set; a multi-layer neural network model is constructed, wherein the input layer of the multi-layer neural network model is the PWV, PWV increment, PWV increase rate and channel data in several time periods before different moments, and the output layer is the rainfall level in a preset time period in the future of the corresponding moment; the multi-layer neural network model is trained using the constructed data set to obtain a trained multi-layer neural network model.

[0069] The rainfall forecasting device combining GNSS-PWV and meteorological satellite remote sensing provided by the embodiment of the present invention aims at solving the problems existing in the existing rainfall forecasting. Figure 3 Several modules in the system integrate GNSS-PWV data and meteorological satellite remote sensing data, use neural network to build a model for rainfall prediction, reduce the false alarm rate of PWV-based rainfall forecast, and thus improve the accuracy of rainfall forecast.

[0070] 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:

[0071] Based on the content of the above device embodiment, as a preferred embodiment, the rainfall forecasting device combining GNSS-PWV and meteorological satellite remote sensing provided in the embodiment of the present invention, in the training of the multi-layer neural network model, after completing the data matching, further includes:

[0072] The matched data are divided into rainfall data and non-rainfall data according to the standard of whether there is rainfall in the preset time period;

[0073] Balancing the rainfall data and the non-rainfall data to form a data set;

[0074] Divide the dataset into training and validation sets according to the required ratio.

[0075] Based on the contents of the above device embodiments, as a preferred embodiment, in the rainfall forecasting device combining GNSS-PWV and meteorological satellite remote sensing provided in the embodiments of the present invention, the channel data includes a water vapor channel and a long-wave infrared channel.

[0076] Based on the contents of the above-mentioned device embodiments, as a preferred embodiment, in the rainfall forecasting device combining GNSS-PWV and meteorological satellite remote sensing provided in the embodiments of the present invention, the multi-layer neural network model is constructed using a back-propagation neural network.

[0077] Based on the contents of the above-mentioned device embodiments, as a preferred embodiment, in the rainfall forecasting device combining GNSS-PWV and meteorological satellite remote sensing provided in the embodiments of the present invention, the rainfall levels are divided into no rain, light rain, moderate rain, heavy rain and extremely heavy rain in turn.

[0078] The embodiment of the present invention further provides a ground station, which is equipped with the above-mentioned rainfall forecasting device combining GNSS-PWV and meteorological satellite remote sensing, and is used to perform the following steps when performing rainfall forecasting:

[0079] Determine the PWV, PWV increment and PWV increase rate in a certain time period before the rainfall site to be predicted, and extract channel data in a certain time period before the rainfall site to be predicted from the meteorological satellite remote sensing data;

[0080] After matching the PWV, PWV increment and PWV increase rate with the channel data in time, they are input into the trained multi-layer neural network model to output the rainfall level in the future preset time period. The training of the multi-layer neural network model includes: calculating the PWV, PWV increment, PWV increase rate and downloading precipitation data of the long time series of the GNSS co-located meteorological station, and dividing the precipitation into several rainfall levels; acquiring the remote sensing data of the meteorological satellite in real time, and extracting the channel data of the long time series at the GNSS co-located meteorological station from the remote sensing data; matching the PWV, PWV increment, PWV increase rate and precipitation with the channel data in time to form a data set; constructing a multi-layer neural network model, the input layer of the multi-layer neural network model is the PWV, PWV increment, PWV increase rate and channel data in several time periods before different moments, and the output layer is the rainfall level in the preset time period in the future of the corresponding moment; using the constructed data set to train the multi-layer neural network model to obtain a trained multi-layer neural network model.

[0081] The embodiment of the present invention further provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the steps of the rainfall forecasting method combining GNSS-PWV and meteorological satellite remote sensing are implemented as follows:

[0082] Determine the PWV, PWV increment and PWV increase rate in a certain time period before the rainfall site to be predicted, and extract channel data in a certain time period before the rainfall site to be predicted from the meteorological satellite remote sensing data;

[0083] After matching the PWV, PWV increment and PWV increase rate with the channel data in time, they are input into the trained multi-layer neural network model to output the rainfall level in the future preset time period. The training of the multi-layer neural network model includes: calculating the PWV, PWV increment, PWV increase rate and downloading precipitation data of the long time series of the GNSS co-located meteorological station, and dividing the precipitation into several rainfall levels; acquiring the remote sensing data of the meteorological satellite in real time, and extracting the channel data of the long time series at the GNSS co-located meteorological station from the remote sensing data; matching the PWV, PWV increment, PWV increase rate and precipitation with the channel data in time to form a data set; constructing a multi-layer neural network model, the input layer of the multi-layer neural network model is the PWV, PWV increment, PWV increase rate and channel data in several time periods before different moments, and the output layer is the rainfall level in the preset time period in the future of the corresponding moment; using the constructed data set to train the multi-layer neural network model to obtain a trained multi-layer neural network model.

[0084] 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 GNSS-PWV and meteorological satellite remote sensing, as shown below:

[0085] Determine the PWV, PWV increment and PWV increase rate in a certain time period before the rainfall site to be predicted, and extract channel data in a certain time period before the rainfall site to be predicted from the meteorological satellite remote sensing data;

[0086] After matching the PWV, PWV increment and PWV increase rate with the channel data in time, they are input into the trained multi-layer neural network model to output the rainfall level in the future preset time period. The training of the multi-layer neural network model includes: calculating the PWV, PWV increment, PWV increase rate and downloading precipitation data of the long time series of the GNSS co-located meteorological station, and dividing the precipitation into several rainfall levels; acquiring the remote sensing data of the meteorological satellite in real time, and extracting the channel data of the long time series at the GNSS co-located meteorological station from the remote sensing data; matching the PWV, PWV increment, PWV increase rate and precipitation with the channel data in time to form a data set; constructing a multi-layer neural network model, the input layer of the multi-layer neural network model is the PWV, PWV increment, PWV increase rate and channel data in several time periods before different moments, and the output layer is the rainfall level in the preset time period in the future of the corresponding moment; using the constructed data set to train the multi-layer neural network model to obtain a trained multi-layer neural network model.

[0087] In summary, the present invention has the following advantages:

[0088] (1) The present invention integrates GNSS-PWV data and meteorological satellite remote sensing data, and uses a neural network (NN) to build a model for rainfall prediction, thereby reducing the false alarm rate of rainfall forecast based on PWV and improving the accuracy of rainfall forecast.

[0089] (2) The present invention makes up for the lack of prediction ability of numerical weather forecast models for single-point short-term rainfall events, especially in areas with complex terrain or scarce traditional observation data. By utilizing satellite remote sensing data, it provides an accurate single-point short-term rainfall prediction method that can cope with possible extreme rainfall events.

[0090] 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 forecasting method combining GNSS-PWV and meteorological satellite remote sensing, characterized in that: include: Determine the PWV, PWV increment and PWV increase rate in a certain time period before the rainfall site to be predicted, and extract channel data in a certain time period before the rainfall site to be predicted from the meteorological satellite remote sensing data; After matching the PWV, PWV increment and PWV increase rate with the channel data according to time, the PWV, PWV increment and PWV increase rate are input into the trained multi-layer neural network model to output the rainfall level in the future preset time period; wherein the training of the multi-layer neural network model includes: Calculate the PWV, PWV increment, PWV increase rate and download precipitation data of the long time series of GNSS co-located meteorological stations, and divide the precipitation into several rainfall levels; Acquire remote sensing data from meteorological satellites in real time, and extract long-time series channel data from GNSS and meteorological stations; Matching the PWV, PWV increment, PWV increase rate and precipitation with the channel data in time to form a data set; Constructing a multi-layer neural network model, wherein the input layer of the multi-layer neural network model is PWV, PWV increment, PWV increase rate and channel data 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; The multi-layer neural network model is trained using the constructed data set to obtain a trained multi-layer neural network model.

2. The rainfall forecasting method combining GNSS-PWV and meteorological satellite remote sensing according to claim 1, characterized in that: After completing the data matching, it also includes: The matched data are divided into rainfall data and non-rainfall data according to the standard of whether there is rainfall in the preset time period; Balancing the rainfall data and the non-rainfall data to form a data set; Divide the dataset into training and validation sets according to the required ratio.

3. The rainfall forecasting method combining GNSS-PWV and meteorological satellite remote sensing according to claim 1, characterized in that: The channel data includes a water vapor channel and a long-wave infrared channel.

4. The rainfall forecasting method combining GNSS-PWV and meteorological satellite remote sensing according to claim 1, characterized in that: The multi-layer neural network model is constructed using a back-propagation neural network.

5. The rainfall forecasting method combining GNSS-PWV and meteorological satellite remote sensing according to claim 1, characterized in that: The rainfall levels are divided into no rain, light rain, moderate rain, heavy rain and extremely heavy rain.

6. A rainfall forecasting device combining GNSS-PWV and meteorological satellite remote sensing, characterized in that: include: A data acquisition module is used to determine the PWV, PWV increment and PWV increase rate in a certain period of time before the rainfall site to be predicted, and to extract channel data in a certain period of time before the rainfall site to be predicted from meteorological satellite remote sensing data; A rainfall forecast module is used to match the PWV, PWV increment and PWV increase rate with the channel data according to time, input the trained multi-layer neural network model, and output the rainfall level within a preset time period in the future; wherein the training of the multi-layer neural network model includes: Calculate the PWV, PWV increment, PWV increase rate and download precipitation data of the long time series of GNSS co-located meteorological stations, and divide the precipitation into several rainfall levels; Acquire remote sensing data from meteorological satellites in real time, and extract long-time series channel data from GNSS and meteorological stations; Matching the PWV, PWV increment, PWV increase rate and precipitation with the channel data in time to form a data set; Constructing a multi-layer neural network model, wherein the input layer of the multi-layer neural network model is PWV, PWV increment, PWV increase rate and channel data 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; The multi-layer neural network model is trained using the constructed data set to obtain a trained multi-layer neural network model.

7. A ground measuring station, characterized in that: A rainfall forecasting device combining GNSS-PWV and meteorological satellite remote sensing as claimed in claim 6 is provided.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the rainfall forecasting method combining GNSS-PWV and meteorological satellite remote sensing are implemented as described in any one of claims 1 to 5.

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 GNSS-PWV and meteorological satellite remote sensing as claimed in any one of claims 1 to 5.

Images