A Temperature Inversion Method for a Transitional Weather

Abstract

The present invention relates to the field of air temperature inversion, and specifically relates to a method for air temperature inversion of transitional weather. Step 1: Data preparation; Step 2: Based on the prepared data, conduct air temperature inversion with high temporal resolution and high spatial resolution under transitional weather, including the following steps: Step 21: Perform temporal difference and spatial difference on the air temperature data of GFS; Step 22: Match the comprehensive data of FY-4A, the air temperature data released by GFS, the rainfall data and measured air temperature data of meteorological stations, NDVI, DEM slope, aspect, and underlying surface type in terms of time and space according to the time of rainfall occurrence and the longitude and latitude of the stations to form a dataset required for the air temperature inversion model, and develop an air temperature inversion method for transitional weather based on FY-4A satellite by combining the advantages of feedforward neural network; Step 23: Use FNN to establish air temperature inversion models under two kinds of transitional weather processes respectively. With this method, air temperature data with high temporal resolution and high spatial resolution can be obtained under transitional weather conditions.

Description

Technical Field

[0001] The present invention relates to the technical field of air temperature inversion, and particularly to an air temperature inversion method for transitional weather. Background Art

[0002] In recent years, with the rapid development of satellite remote sensing technology, the accuracy of air temperature inversion has been further improved. Transitional weather refers to weather with significant transitional changes, such as rainfall. Research shows that existing air temperature inversion mainly considers the accuracy differences of different air temperature inversion methods, without considering air temperature inversion during rainfall. For transitional weather processes with large air temperature changes, conventional air temperature inversion methods have great difficulty and low accuracy in inversion under transitional weather. Air temperature inversion for transitional weather can improve the accuracy of air temperature inversion in rainy weather, better adapt to the situation of rapid climate change, and have better response ability to the rapid change of air temperature. When the climate turns, it considers the non-linear characteristics of the climate system, and this method can more accurately predict the change trend of air temperature.

[0003] To solve the above problems, the present invention proposes an air temperature inversion method for transitional weather based on FY-4A satellite data. Summary of the Invention

[0004] The purpose of the present invention is to provide an air temperature inversion method for transitional weather, which can obtain air temperature data with high temporal resolution and high spatial resolution under transitional weather conditions by using this method.

[0005] To achieve the above purpose, the technical solution of the present invention is as follows: An air temperature inversion for transitional weather includes the following steps:

[0006] Step 1, data preparation:

[0007] The data preparation is as follows: Collect the comprehensive data of FY-4A; collect the air temperature data of GFS; collect the NDVI provided by Modis; collect the DEM with a spatial resolution of 90m provided by NASA; slope; aspect and land surface type; and perform average resampling to process the spatial resolution of the DEM into 250m; collect the measured rainfall data and air temperature data of meteorological stations throughout the year of 2022;

[0008] And use the station air temperature as the target variable for the air temperature inversion model modeling of the true air temperature;

[0009] Step 2, based on the prepared data, perform air temperature inversion with high temporal resolution and high spatial resolution under transitional weather;

[0010] It includes the following steps:

[0011] Step 21, perform time difference and spatial difference on the air temperature data of GFS;

[0012] Step 22: Perform spatio-temporal matching on the comprehensive data of FY-4A, the temperature data released by GFS, the measured rainfall data and temperature data of meteorological stations, NDVI, DEM slope, aspect, and underlying surface type according to the time of rainfall occurrence and the longitude and latitude of the stations, establish a dataset required for the turning temperature inversion model, and then use a feedforward neural network to train the dataset and read the parameters of the network;

[0013] Step 23: Combine the parameters and data read from the neural network in the previous step. Each type of data has corresponding parameters, construct a non-linear relationship between the estimated temperature and these data, and establish a turning weather temperature inversion model.

[0014] Furthermore, in Step 1, the comprehensive data collected for FY-4A includes cloud top height and cloud top temperature data, and its spatial resolution is 4 km.

[0015] Furthermore, collect the temperature data of GFS, and its time and spatial resolutions are 3 hours and 25 km respectively.

[0016] Furthermore, collect NDVI data with a spatial resolution of 250 m and a time resolution of 16 days provided by Modis and perform projection mosaicking to obtain complete data for a certain area.

[0017] Furthermore, in Step 23, invert the turning weather temperature with high time resolution and high spatial resolution based on the FY-4A satellite through the temperature inversion model to form a 250 m temperature grid, and establish a non-linear relationship between the temperature and the input variables based on FNN modeling. And use FNN to combine the parameters of the training model with the FY-4A satellite data, output the inverted temperature, and compare the measured temperature of the station with the inverted temperature to obtain the correlation, root mean square error, and bias.

[0018] The above scheme has the following beneficial effects: The present invention combines the FY-4A satellite data and the GFS temperature data from the spatial and temporal perspectives by utilizing the advantages of the neural network, and performs temperature inversion with high time resolution and high spatial resolution under turning weather, providing a new idea for the research on turning weather temperature inversion. Through the research on turning weather temperature inversion, more accurate temperature data can be provided, improving the accuracy and timeliness of weather forecasting. It is crucial for predicting weather events;

[0019] The additional aspects and advantages of the present invention will be partially given in the following description, partially will become apparent from the following description, or be understood through the practice of the present invention. Description of the Drawings

[0020] Figure 1 It is a technical route map for the turning weather temperature inversion model;

[0021] Figure 2 Comparison between the model-inverted temperature and the temperature measured at the meteorological station. DETAILED DESCRIPTION

[0022] The technical solution of the present invention will be described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0023] The following is further described in detail through specific implementation methods:

[0024] This application provides a temperature inversion method for transitional weather, which is basically as shown in the attached Figure 1 To Attachment Figure 2 As shown, the following steps are included:

[0025] Step 1, Data Preparation

[0026] Collect the cloud top height (CTH) and cloud top temperature (CTT) data of FY-4A, with a spatial resolution of 4km respectively. Collect the temperature data of GFS, with a temporal and spatial resolution of 3 hours and 25km respectively. Collect the vegetation index (NDVI) data with a spatial resolution of 250m and a temporal resolution of 16 days provided by Modis and perform projection stitching to obtain complete data for a certain area. Collect the (digital elevation model) DEM with a spatial resolution of 90m provided by NASA and perform simple average resampling to process the spatial resolution of DEM to 250m. Slope, aspect and underlying surface type Collect the measured rainfall data and temperature data from the meteorological station, match these data in time and space, form a training set for the transition weather inversion, and use the station temperature as the target variable for the real temperature in the temperature inversion model modeling. Finally, clean the data and remove outliers.

[0027] Step 2: Based on the prepared data, perform high temporal resolution and high spatial resolution temperature inversion under transitional weather conditions:

[0028] The steps include:

[0029] Step 21: Since there are many types of input data, the GFS temperature data itself has large errors and non-uniform resolutions, it is necessary to pre-process the data and perform time and space differences on the data to ensure the uniformity of the data in time and space resolutions;

[0030] Step 22: Spatially and temporally match the cloud top height (CTH) of FY-4A, cloud top temperature data (CTT), air temperature data released by GFS, measured rainfall data and air temperature data at meteorological stations, NDVI, DEM, slope, aspect, and underlying surface type according to the time of rainfall and the longitude and latitude of the stations, establish the dataset required for the turning temperature inversion model, and then use the feedforward neural network to train the dataset and read the parameters of the network.

[0031] Step 23: Use FNN to establish the turning temperature inversion model. Through this model, the turning weather air temperature with high temporal resolution and high spatial resolution based on FY-4A satellite can be inverted to form a 250m air temperature grid. FNN can better model the nonlinear relationship between temperature and input variables and can adapt to the changes of different types of data and input variables. Finally, use FNN to combine the parameters of the training model with satellite data to output the inverted air temperature. Compare the measured air temperature at the station with the inverted air temperature to calculate the correlation, root mean square error, and bias.

[0032] The present application makes the following explanations based on actual operations:

[0033] The precipitation data and measured near-surface air temperature data of 10 meteorological stations in the Qinghai-Tibet Plateau region throughout 2022 were collected, and the cloud top height (CTH) of FY-4A, cloud top temperature data (CTT), air temperature data released by GFS, Modis / NDVI products, DEM, slope, aspect, and underlying surface type were collected synchronously. The collected data were preprocessed, including time difference, spatial interpolation, data cleaning, and data standardization, and then the adjacent matching method was used for data spatial matching to form the dataset required for modeling, and the turning weather air temperature inversion model was established using these data.

[0034] For the constructed air temperature inversion model, select the data from the moment of no precipitation to the moment of precipitation in the Qinghai-Tibet Plateau region in 2022, take data every other moment, and randomly select half of the data at intervals as the training set of the model, and the remaining half of the data as the test set. Then, the dataset is input into the two models respectively and processed using FNN, and finally the inversion results are output. The entire air temperature inversion process is as Figure 1 shown. The verification results show that the turning weather air temperature inversion model constructed in this study can invert the grid air temperature with high spatial resolution and high precision during rainfall, and its correlation coefficient, root mean square error, and bias are as Figure 2 shown; through the above experiments, the main parameter information of the air temperature inversion model is shown in Table 1 below.

[0035] Table 1 Main parameter information of the air temperature inversion model

[0036]

[0037] Obviously, the above embodiments are merely examples given for clear illustration and not limitations on the implementation manners. For those of ordinary skill in the art, other different forms of changes or modifications can be made based on the above description. It is not necessary and impossible to enumerate all the implementation manners here. And the obvious changes or modifications derived therefrom still fall within the protection scope of the present invention.

Claims

1. A temperature inversion method for transitional weather, characterized in that: The steps include: Step 1, data preparation: Data preparation is as follows: collect comprehensive data of FY-4A; collect temperature data of GFS; collect NDVI provided by Modis; slope; slope aspect; underlying surface type data; Collect the DEM with a spatial resolution of 90m provided by NASA and perform average resampling to process the spatial resolution of DEM to 250m; Collect the measured rainfall and temperature data of meteorological stations throughout 2022; Step 2: Based on the prepared data, high temporal resolution and high spatial resolution temperature inversion is performed under transitional weather conditions; The steps include: Step 21, performing time difference and space difference on the temperature data of GFS; Step 22, the comprehensive data of FY-4A, the temperature data released by GFS, the rainfall data of meteorological stations and the measured temperature data, NDVI, DEM, slope, slope aspect and underlying surface type are matched in time and space according to the time when the station recorded rainfall and the longitude and latitude of the station, and the data set required for the transition temperature inversion model is established, and then the data set is trained using a feedforward neural network to read the parameters of the network; Step 23, combining the parameters and data read in the neural network in the previous step, each type of data has corresponding parameters, constructing a nonlinear relationship between the estimated temperature and these data, and establishing a transitional weather temperature inversion model.

2. The temperature inversion method of transitional weather according to claim 1, characterized in that: In step 1, the comprehensive data of FY-4A including cloud top height and cloud top temperature data are collected with a spatial resolution of 4 km.

3. The temperature inversion method of transitional weather according to claim 2, characterized in that: The temperature data of GFS are collected with a temporal and spatial resolution of 3 hours and 25 km respectively.

4. The temperature inversion method of transitional weather according to claim 3 is characterized in that: The NDVI data with a spatial resolution of 250m and a temporal resolution of 16 days provided by Modis were collected and projected and stitched to obtain complete data for a certain area.

5. The temperature inversion method of transitional weather according to claim 4, characterized in that: In step 23, the high temporal resolution and high spatial resolution transitional weather temperature based on the FY-4A satellite is inverted through the temperature inversion model to form a 250m temperature grid, and the nonlinear relationship between the temperature and the input variable is modeled based on the FNN. The parameters of the training model are combined with the FY-4A satellite data using the FNN to output the inverted temperature. The measured temperature at the station is compared with the inverted temperature to obtain the correlation, root mean square error, and bias.

Images