Stratified precipitable water amount correction method, system and equipment

Abstract

The invention discloses a method, system and equipment for correcting layered precipitable water, and relates to the field of atmospheric remote sensing, and the method comprises the steps: obtaining the atmospheric precipitable water of a GNSS observation station according to the data of the GNSS observation station corresponding to the time and position of the layered precipitable water; inputting the atmospheric precipitable water amount of the GNSS observation station into the water vapor bedding model to obtain the layered precipitable water amount of the GNSS observation station; determining satellite remote sensing layered precipitable water and GNSS observation station layered precipitable water at the same time, respectively determining corresponding positions of the satellite remote sensing layered precipitable water and the GNSS observation station layered precipitable water, and obtaining the GNSS observation station layered precipitable water in which the distance between the corresponding position of the GNSS observation station layered precipitable water and the corresponding position of the satellite remote sensing layered precipitable water is smaller than a preset distance; and according to the layered precipitable water amount of the selected GNSS observation station, correcting the layered precipitable water amount obtained by the satellite remote sensing technology by using a linear difference correction model. According to the method, high-precision correction can be carried out on the layered precipitable water quantity product detected by the remote sensing satellite sensor through the layered precipitable water quantity with high temporal-spatial resolution.

Description

Technical Field

[0001] The invention relates to the field of atmospheric remote sensing, and in particular to a correction method, system and equipment for layered precipitable water. Background Art

[0002] Atmospheric water vapor is an important greenhouse gas in the Earth's ecosystem and plays a vital role in climate change, global water cycle and energy balance. Atmospheric precipitable water is a commonly used indicator of atmospheric water vapor content, which represents the total amount of water vapor contained in the atmospheric column from the Earth's surface to the top of the troposphere per unit area. Further research has found that layered precipitable water represents the amount of water vapor in the atmospheric column within a specific vertical range, which can directly describe the distribution and changes of atmospheric water vapor at different altitudes, and provides effective data support for monitoring the evolution of layer-by-layer water vapor advection and instability under extreme weather. Sensors carried by geostationary satellites can monitor layered precipitable water in three specific vertical intervals, namely, precipitable water information in the low-level interval, the middle-level interval and the high-level interval. The layered precipitable water products detected by satellite sensors are easily affected by clouds, rainfall and surface reflection, resulting in limited observation accuracy, which restricts its research scope in the field of atmospheric water vapor. Therefore, accurate correction of layered precipitable water products on a global scale has become one of the important issues to be solved in the field of satellite remote sensing water vapor detection.

[0003] In the prior art, one method is to use the atmospheric precipitable water measured by the high-precision global navigation satellite system GNSS (GlobalNavigation Satellite System) as a reference value, construct a linear correction model or a correction model based on machine learning technology, correct the atmospheric precipitable water product, and then calculate the layered precipitable water. Another method is to use a radiosonde, which can obtain layered precipitable water information through vertical integration processing.

[0004] However, GNSS can provide continuous observations with high temporal resolution but low spatial resolution, while radiosondes provide observations with high spatial resolution but low temporal resolution, resulting in the fact that the accuracy of stratified precipitable water detected by remote sensing satellite sensors is still not ideal. Summary of the invention

[0005] The embodiments of the present invention provide a method, system and device for correcting stratified precipitable water, which can solve the problem in the prior art that the accuracy of stratified precipitable water detected by remote sensing satellite sensors is not ideal.

[0006] The embodiment of the present invention provides a correction method for layered precipitable water, comprising the following steps: obtaining atmospheric precipitable water and layered precipitable water containing time and position information at high spatiotemporal resolution; and training an artificial neural network for obtaining layered precipitable water according to the atmospheric precipitable water as a water vapor stratification model; obtaining atmospheric precipitable water at the GNSS observation station according to GNSS observation station data corresponding to the time and position of the layered precipitable water; inputting the atmospheric precipitable water at the GNSS observation station into the water vapor stratification model to obtain layered precipitable water at the GNSS observation station; determining the satellite remote sensing layered precipitable water and the GNSS observation station layered precipitable water at the same time, and respectively determining the corresponding positions of the GNSS observation station layered precipitable water and the satellite remote sensing layered precipitable water, and obtaining the GNSS observation station layered precipitable water whose distance between the corresponding position of the GNSS observation station layered precipitable water and the corresponding position of the satellite remote sensing layered precipitable water is less than a preset distance; and correcting the layered precipitable water obtained by satellite remote sensing technology using a linear difference correction model according to the obtained GNSS observation station layered precipitable water.

[0007] Furthermore, the training of an artificial neural network for obtaining layered precipitable water according to atmospheric precipitable water as a water vapor stratification model specifically comprises the following steps: obtaining an artificial neural network ANN including an input layer, a hidden layer and an output layer structure; inputting an atmospheric precipitable water data set from the input layer of the artificial neural network ANN, outputting low-layer precipitable water, middle-layer precipitable water and high-layer precipitable water through the output layer of the artificial neural network ANN, and training the artificial neural network ANN; and using the trained artificial neural network ANN as a water vapor stratification model.

[0008] Furthermore, the linear differential correction model is formulated as follows:

[0009]

[0010] Among them, LPW low LPW mid and LPW high They represent the stratified precipitable water content at the lower, middle and upper levels obtained by satellite remote sensing technology before correction, respectively. low and b low 、a mid and b mid 、a high and b high They represent the correction model coefficients corresponding to the three stratified precipitable water, ΔLPW low , ΔLPW mid , ΔLPW high They respectively represent the differential correction values ​​of three types of satellite remote sensing stratified precipitable water and the corresponding GNSS observation station stratified precipitable water.

[0011] Furthermore, the preset distance is 5 kilometers.

[0012] Furthermore, the obtaining of atmospheric precipitable water and layered precipitable water containing time and location information specifically includes: obtaining atmospheric precipitable water and layered precipitable water containing a specific year, annual accumulated day and daily accumulated time, as well as a specific longitude, latitude and geodetic altitude.

[0013] An embodiment of the present invention provides a correction system for stratified precipitable water, including:

[0014] The data acquisition module is used to obtain atmospheric precipitable water and layered precipitable water containing time and position information at high temporal and spatial resolution; and train an artificial neural network for obtaining layered precipitable water according to the atmospheric precipitable water as a water vapor stratification model; the layered precipitation matching module is used to obtain the atmospheric precipitable water of the GNSS observation station according to the GNSS observation station data corresponding to the time and position of the layered precipitable water; the atmospheric precipitable water of the GNSS observation station is input into the water vapor stratification model to obtain the layered precipitable water of the GNSS observation station; the satellite remote sensing layered precipitable water and the GNSS observation station layered precipitable water at the same time are determined, and the corresponding positions of the GNSS observation station layered precipitable water and the satellite remote sensing layered precipitable water are determined respectively, and the GNSS observation station layered precipitable water is obtained when the distance between the corresponding position of the GNSS observation station layered precipitable water and the corresponding position of the satellite remote sensing layered precipitable water is less than a preset distance; the layered precipitation correction module is used to correct the layered precipitable water obtained by satellite remote sensing technology using a linear difference correction model according to the obtained GNSS observation station layered precipitable water.

[0015] An embodiment of the present invention provides a computer device, including: a memory and a processor; the memory stores a computer program, and the processor implements the above-mentioned method for correcting stratified precipitation when executing the computer program.

[0016] The embodiment of the present invention provides a correction method, system and device for stratified precipitation, which have the following beneficial effects compared with the prior art:

[0017] The present invention uses a linear differential correction model to correct the layered precipitable water obtained by satellite remote sensing technology based on the obtained GNSS observation station layered precipitable water. During correction, the accuracy of the GNSS observation station layered precipitable water will affect the accuracy of the satellite remote sensing layered precipitable water. The GNSS observation station layered precipitable water with high temporal and spatial resolution has the characteristics of frequent data collection in time and intensive data collection in space, so the data accuracy is high. Therefore, it is necessary to obtain the GNSS observation station layered precipitable water with high temporal and spatial resolution. Through the atmospheric precipitable water containing time and position information under high temporal and spatial resolution and the water vapor stratification model trained with the layered precipitable water, the layered precipitable water obtained has high temporal and spatial resolution. After the atmospheric precipitable water of the GNSS observation station is input into the water vapor stratification model, the layered precipitable water of the GNSS observation station obtained has high temporal and spatial resolution.

[0018] Finally, the stratified precipitable water obtained by the corrected satellite remote sensing technology has high accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 A flow chart of a method for correcting stratified precipitation provided by an embodiment of the present invention;

[0020] Figure 2 A flow chart of a method for correcting stratified precipitation provided by an embodiment of the present invention;

[0021] Figure 3 The information diagram required for spatiotemporal matching provided by the embodiment of the present invention;

[0022] Figure 4 A schematic diagram of stratified precipitable water provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0023] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific embodiments of the present invention are described in detail below in conjunction with the accompanying drawings. In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention can be implemented in many other ways different from those described herein, and those skilled in the art can make similar improvements without violating the connotation of the present invention, so the present invention is not limited by the specific embodiments disclosed below.

[0024] See also Figure 1 to Figure 4 The embodiment of the present invention provides a method for correcting stratified precipitable water, comprising the following steps:

[0025] Step 1: Collect reanalysis data ERA5 (ECMWF Reanalysis of the Global Climatesystem, 5th generation) and sounding data to provide location information, time information, and meteorological data such as temperature, air pressure and specific humidity. ERA5 and sounding data have high temporal and spatial resolution. High temporal and spatial resolution has the characteristics of frequent data collection in time and intensive data collection in space. Its data is highly accurate and has high research value. The atmospheric precipitable water PWV (Precipitable Water Vapor) and layered precipitable water LPW (Layer Precipitable Water) are calculated using air pressure and specific humidity data to obtain atmospheric precipitable water PWV and layered precipitable water LPW containing time and location information at high temporal and spatial resolution.

[0026] The atmospheric precipitable water volume PWV can be calculated using formula (1):

[0027]

[0028] Among them, p sur and p top are the surface pressure and the tropopause pressure, respectively, p i and q i are the air pressure and specific humidity of the i-th layer in the ERA5 and sounding profile data, respectively; g is the local gravity acceleration; and n is the total number of layers in the sounding profile.

[0029] The stratified precipitation water LPW can be calculated using formula (2):

[0030]

[0031] Among them, LPW low LPW mid and LPW high represent the low-level precipitable water, the middle-level precipitable water and the high-level precipitable water respectively. Correspondingly, p a 、p b and p c represents the air pressure of the ath layer, bth layer and cth layer divided by the sounding profile in the low-level interval, middle-level interval and high-level interval respectively, q a ,q b and q c represents the relative humidity of the ath layer, bth layer and cth layer in the low-level interval, middle-level interval and high-level interval respectively, p low 、p mid and p high Represents the air pressure at the boundaries of the low, middle and high levels, n a Indicates the total number of layers of the sounding profile in the low-level interval, nb Indicates the total number of layers of the sounding profile in the middle layer interval, n c It represents the total number of layers of the sounding profile in the high-level interval, which can be calculated using formula (3):

[0032]

[0033] Step 2: Establish a global water vapor stratification model based on artificial neural network. First, construct the input layer, hidden layer and output layer structure of the artificial neural network ANN (Artificial Neural Network). The input layer includes the location information, time information and atmospheric precipitable water PWV data of each ERA5 or sounding observation point. The hidden layer has several layers, each layer contains several neurons, and the activation function is set for each hidden layer. The output layer is the three stratified precipitable water data LPW of ERA5 or sounding observation points. low LPW mid and LPW high .

[0034] Step 3: Use the ten-fold cross-validation method to train and test the artificial neural network model, use the mean absolute error (MAE) loss function to supervise the training process, and use the minimum mean absolute error of the test results as the criterion to determine the optimal parameter information of the artificial neural network model.

[0035] Step 4: Use high-precision processing software to process the data of the Global Navigation Satellite System (GNSS) (hereinafter referred to as GNSS) that is synchronized with the layered precipitable water and in the same area to obtain the high-temporal and spatial GNSS tropospheric zenith total delay ZTD (Zenith Total Delay). Use formula (4) to calculate the atmospheric precipitable water at the GNSS station:

[0036]

[0037] Among them, ρ w represents the density of liquid water, R v represents the gas constant of atmospheric water vapor, k'2 and k3 both represent the refractive index constant, T m represents the weighted average temperature of the atmosphere, P S , and H S They represent the air pressure, latitude and geodetic height at the GNSS station respectively.

[0038] Step 5: Substitute the atmospheric precipitable water volume (GNSS PWV) and other input information into the trained global water vapor stratification model based on artificial neural network to obtain three types of GNSS stratified precipitable water data (LPW) corresponding to GNSS PWV. low LPW mid and LPW high .

[0039] Step 6: Temporally and spatially match the three stratified precipitable water products and the three GNSS LPW stratified precipitable water products. low , GNSS LPW mid and GNSS LPW high As a reference value, construct FY-4LPW low 、FY-4LPW mid and FY-4LPW high The corresponding linear differential correction model, FY-4 (FengYun-4) is the fourth generation of polar orbiting meteorological satellite in the Fengyun meteorological satellite series, which realizes high-precision and high-temporal and spatial correction of stratified precipitable water.

[0040] The location information is longitude, latitude and geodetic height, and the time information is year, annual cumulative day and daily cumulative hour. The number of hidden layers of the water vapor stratification model is 5 layers, and the number of neurons in each layer is 128, 64, 32 and 8 respectively. The activation function of the hidden layer is the ReLU (Rectified Linear Unit) function.

[0041] The specific steps of the ten-fold cross-validation method are: divide the valid data set into ten parts, and then use each part as a validation set in turn, and the other data as a training set for training and verification. Finally, after obtaining a satisfactory model hyperparameter, use all valid data sets as training data sets, and use the model hyperparameters to train to obtain a model.

[0042] The mean absolute error MAE loss function is shown in formula (5):

[0043]

[0044] Among them, n represents the total amount of sample data, y j and y' j Represent the j-th predicted value and target value respectively.

[0045] Other input information includes the longitude, latitude and geodetic height of the GNSS station, the year, annual cumulative day and daily cumulative time corresponding to the GNSS PWV data, and the surface pressure and temperature at the GNSS station.

[0046] The steps of spatiotemporal matching are as follows: first, according to the time information of the three types of stratified precipitable water detected by satellite remote sensing, the data at the same time are searched in all GNSS LPW products, which are the time-matched GNSS LPW data; then, according to the site location information in the GNSS LPW data and the pixel point location information of the stratified precipitable water data detected by satellite remote sensing, an appropriate distance threshold is set, and the data with a spatial distance between the GNSS site and the pixel point less than the threshold is used as the spatial matching data, thus forming the spatiotemporal matching GNSS LPW data and satellite remote sensing LPW data.

[0047] The linear differential correction model is shown in formula (6):

[0048]

[0049] Among them, LPW low LPW mid and LPW high They represent the three types of satellite remote sensing stratified precipitable water before correction, a low and b low 、a mid and b mid 、a high and b high They represent the correction model coefficients corresponding to the three stratified precipitable water, ΔLPW low , ΔLPW mid and ΔLPW high They represent the differential corrections of the three types of stratified precipitable water.

[0050] The site location information is the longitude and latitude information of the GNSS station, and the pixel location information is the longitude and latitude information of the pixel in the satellite remote sensing image. The appropriate distance threshold is 5 km.

[0051] The present invention utilizes high-precision global ERA5 reanalysis data and sounding data to construct a global water vapor stratification model based on artificial neural network, combines it with global high-precision GNSSPWV data, and then uses the water vapor stratification model to invert GNSS layered precipitable water data. A linear differential correction model is constructed with the global GNSS layered precipitable water as a reference value, so as to achieve high-temporal, high-precision correction of satellite remote sensing layered precipitable water products, and effectively improve the quality of satellite remote sensing layered precipitable water products.

[0052] The specific advantages include: using high-precision ERA5 reanalysis data and sounding data to construct a global water vapor stratification model based on artificial neural networks, which can accurately estimate high-precision global GNSSLPW data in real time, solving the key problem that traditional GNSSPWV data cannot be used for stratified precipitable water product correction; further, using the globally densely distributed high-precision GNSSLPW data as a reference value to construct a linear differential correction model, the high temporal and spatial resolution advantages of GNSS water vapor detection methods can be fully utilized, greatly improving the serious deficiencies in the temporal resolution and spatial distribution of the stratified precipitable water products provided by radiosondes, and realizing high-temporal and high-precision correction of satellite remote sensing stratified precipitable water products.

[0053] An embodiment of the present invention provides a correction system for stratified precipitable water, including:

[0054] The data acquisition module is used to obtain the atmospheric precipitable water and layered precipitable water containing time and position information at high spatiotemporal resolution; and train the artificial neural network for obtaining the layered precipitable water according to the atmospheric precipitable water as a water vapor stratification model. The layered precipitation matching module is used to obtain the atmospheric precipitable water of the GNSS observation station according to the GNSS observation station data corresponding to the time and position of the layered precipitable water; determine the satellite remote sensing layered precipitable water and the GNSS observation station layered precipitable water at the same time, and respectively determine the corresponding positions of the GNSS observation station layered precipitable water and the satellite remote sensing layered precipitable water, and obtain the GNSS observation station layered precipitable water whose distance between the corresponding position of the GNSS observation station layered precipitable water and the corresponding position of the satellite remote sensing layered precipitable water is less than the preset distance. The layered precipitation correction module is used to correct the layered precipitable water obtained by satellite remote sensing technology using a linear difference correction model according to the obtained GNSS observation station layered precipitable water.

[0055] An embodiment of the present invention provides a computer device, including: a memory and a processor; the memory stores a computer program, and the processor implements the steps of a correction method for stratified precipitation when executing the computer program.

[0056] A specific embodiment is as follows:

[0057] 1. Collect the location information, time information, and meteorological data such as temperature, air pressure, and specific humidity provided by ERA5 reanalysis data and sounding data, and use the air pressure and specific humidity data to calculate the atmospheric precipitable water (PWV) and stratified precipitable water (LPW).

[0058] 2. Establish a global water vapor stratification model based on artificial neural network. First, construct the input layer, hidden layer and output layer structure of the artificial neural network. The input layer includes the location information, time information and atmospheric precipitable water PWV data of each ERA5 / sonde observation point. The hidden layer has several layers, each layer contains several neurons, and the activation function is set for each hidden layer. The output layer is the three stratified precipitable water data LPW of the ERA5 / sonde observation point. low LPW mid and LPW high .

[0059] 3. The ten-fold cross-validation method is used to train and test the artificial neural network model, and the mean absolute error (MAE) loss function is used to supervise the training process. The minimum mean absolute error of the test results is used as the criterion to determine the optimal parameter information of the artificial neural network model.

[0060] 4. Use high-precision processing software to process the GNSS data of the same region and synchronized with the layered precipitable water, obtain the high-temporal and spatial GNSS tropospheric zenith total delay ZTD, and calculate the atmospheric precipitable water at the GNSS measuring station.

[0061] 5. Substitute GNSS PWV and other input information into the trained global water vapor stratification model based on artificial neural network to obtain three types of GNSS stratified precipitable water data LPW corresponding to GNSS PWV. low LPW mid and LPW high .

[0062] 6. Temporal and spatial matching of the three stratified precipitable water products and the three GNSS LPW stratified precipitable water products is performed, and the matched GNSS LPW low , GNSS LPW mid and GNSS LPW high As a reference value, the FY-4LPW of the second-generation geosynchronous orbit meteorological satellite FY-4 was constructed. low 、FY-4LPW mid and FY-4LPW high The corresponding linear difference correction model achieves high-precision and high-temporal and spatial correction of stratified precipitable water.

[0063] The above-mentioned embodiments only express several implementation methods of the present invention, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the invention patent. It should be pointed out that, for ordinary technicians in this field, several variations and improvements can be made without departing from the concept of the present invention, and these all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention shall be subject to the attached claims.

Claims

1. A method for correcting stratified precipitation, characterized in that: The following steps are involved: Obtain atmospheric precipitable water and layered precipitable water with time and location information at high temporal and spatial resolution; And train an artificial neural network to obtain stratified precipitable water according to atmospheric precipitable water as a water vapor stratification model; Obtain atmospheric precipitable water at a GNSS observation station based on GNSS observation station data corresponding to the time and location of the layered precipitable water; The atmospheric precipitable water at the GNSS observation station is input into the water vapor stratification model to obtain the stratified precipitable water at the GNSS observation station; Determine the satellite remote sensing layered precipitable water and the GNSS observation station layered precipitable water at the same time, and respectively determine the corresponding positions of the GNSS observation station layered precipitable water and the satellite remote sensing layered precipitable water, and obtain the GNSS observation station layered precipitable water whose distance between the corresponding position of the GNSS observation station layered precipitable water and the corresponding position of the satellite remote sensing layered precipitable water is less than a preset distance; According to the stratified precipitable water obtained from GNSS observation stations, the linear difference correction model is used to correct the stratified precipitable water obtained by satellite remote sensing technology.

2. A method for correcting stratified precipitation according to claim 1, characterized in that: The training of an artificial neural network to obtain stratified precipitable water according to atmospheric precipitable water as a water vapor stratification model specifically comprises the following steps: Obtain an artificial neural network ANN including an input layer, a hidden layer and an output layer structure; The atmospheric precipitable water data set is input from the input layer of the artificial neural network ANN, and the low-level precipitable water, the middle-level precipitable water and the high-level precipitable water are output through the output layer of the artificial neural network ANN to train the artificial neural network ANN; The trained artificial neural network ANN is used as the water vapor stratification model.

3. A method for correcting stratified precipitation according to claim 1, characterized in that: The linear differential correction model is formulated as follows: Among them, LPW low LPW mid and LPW high They represent the stratified precipitable water content at the lower, middle and upper levels obtained by satellite remote sensing technology before correction, respectively. low and b low 、a mid and b mid 、a high and b high They represent the correction model coefficients corresponding to the three stratified precipitable water, ΔLPW low , ΔLPW mid , ΔLPW high They respectively represent the differential correction values ​​of three types of satellite remote sensing stratified precipitable water and the corresponding GNSS observation station stratified precipitable water.

4. A method for correcting stratified precipitation according to claim 1, characterized in that: The preset distance is 5 kilometers.

5. A method for correcting stratified precipitation according to claim 1, characterized in that: The obtaining of atmospheric precipitable water and layered precipitable water including time and location information specifically includes: Obtain atmospheric precipitable water and layered precipitable water with specific year, annual cumulative day and daily cumulative time, as well as specific longitude, latitude and geodetic altitude.

6. A correction system for stratified precipitation, characterized in that: include: A data acquisition module, used to obtain atmospheric precipitable water and layered precipitable water with time and location information at high temporal and spatial resolution; And train an artificial neural network to obtain stratified precipitable water according to atmospheric precipitable water as a water vapor stratification model; A layered precipitation matching module is used to obtain atmospheric precipitable water at a GNSS observation station based on GNSS observation station data corresponding to the time and location of the layered precipitable water; The atmospheric precipitable water at the GNSS observation station is input into the water vapor stratification model to obtain the stratified precipitable water at the GNSS observation station; Determine the satellite remote sensing layered precipitable water and the GNSS observation station layered precipitable water at the same time, and respectively determine the corresponding positions of the GNSS observation station layered precipitable water and the satellite remote sensing layered precipitable water, and obtain the GNSS observation station layered precipitable water whose distance between the corresponding position of the GNSS observation station layered precipitable water and the corresponding position of the satellite remote sensing layered precipitable water is less than a preset distance; The layered precipitation correction module is used to correct the layered precipitable water obtained by satellite remote sensing technology using a linear difference correction model based on the layered precipitable water obtained from the GNSS observation station.

7. A computer device comprising: Memory and processor; The memory stores a computer program, wherein the processor implements a method for correcting stratified precipitation according to any one of claims 1 to 5 when executing the computer program.

Images