A method for estimating near-surface air temperature with high spatial and temporal resolution in the cloud sky

Abstract

The invention relates to a method for estimating near-ground temperature of clouds and sky with high spatiotemporal resolution, and belongs to the technical field of atmospheric remote sensing. The method uses cloud top height and cloud top temperature of spatial resolution of geostationary meteorological satellite, temperature, elevation, vegetation index, latitude, longitude and time information predicted by numerical model, and constructs a temperature model of geostationary meteorological satellite with original spatial resolution based on a neural network model. It is assumed that the near-ground temperature difference between a high spatial resolution sub-pixel and the geostationary meteorological satellite original spatial resolution pixel where the sub-pixel is located is caused by the difference in elevation and normalized vegetation index of the two pixels, and a temperature difference model is constructed. On this basis, a near-ground temperature estimation model of clouds and sky with high spatiotemporal resolution is derived. A near-ground temperature estimation model of clouds and sky with high spatiotemporal resolution is constructed based on a generative adversarial neural network. The method can obtain the near-ground temperature spatial distribution under high-precision clear sky conditions that is more refined than satellite observation.

Description

Technical Field

[0001] The invention belongs to the technical field of atmospheric remote sensing, and in particular relates to a method for estimating near-ground air temperature with high temporal and spatial resolution in the cloud sky. Background Art

[0002] Surface air temperature (T air ) usually refers to the atmospheric temperature measured by a louvered box 1.5-2.0 meters above the ground, and is an important parameter for describing the near-ground atmospheric environment. It is one of the basic observation items of meteorological stations and is widely used in hydrology, ecology, climatology, epidemiology, environmental science, and residents' energy consumption. Changes in near-ground air temperature are affected by many factors such as solar radiation, longitude and latitude, surface type, soil moisture, cloud cover, and altitude. The temperature varies greatly with time and space, and shows spatial heterogeneity. High-precision, high-resolution near-ground air temperature is one of the key input parameters for many land surface processes and climate models, and is also of great significance for better understanding and simulation of complex surface processes.

[0003] At present, the near-surface temperature is mainly obtained through routine observations at meteorological stations. Although station observations can provide high temporal resolution and high-precision temperature, they can only obtain temperature data at discrete points, so they cannot well reflect the spatial heterogeneity of temperature. Although the spatial distribution of temperature can be obtained by spatial interpolation of station data, its accuracy is usually affected by the density of meteorological stations, terrain, and landscape conditions. Generally, it is difficult to obtain a high-precision spatial distribution of temperature, especially in complex terrain areas with fewer stations.

[0004] Different from the temperature observed by meteorological stations and estimated by satellites, numerical forecasting and assimilation models can provide spatially fully distributed temperature data, providing important temperature information for areas with clouds and areas without meteorological station observations. These data usually have coarse temporal and spatial resolutions (spatial resolution is usually 0.25°–0.5°, and temporal resolution is usually 3 hours), which will bring greater uncertainty to the application, especially in some areas with complex terrain.

[0005] Unlike meteorological station observations, numerical forecasting and assimilation models, such as the Global Forecast System (GFS), the fifth generation of reanalysis data of the European Center for Medium-Range Weather Forecasts (ERA5), the Global Land Data Assimilation System (GLDAS) and the Chinese Land Data Assimilation System (CLDAS), can provide seamless / full coverage of Tair data on a regional or global scale. However, these data usually have coarse temporal and spatial resolutions (e.g., spatial resolution of 0.25°–0.5° and temporal resolution of 3 hours) (Rao, Liang et al. 2019). This will bring greater uncertainty to applications, especially in areas with complex terrain (Shuai, Zhu et al. 2018).

[0006] Meteorological satellites can achieve global and large-scale observations. For example, a single geostationary meteorological satellite can continuously observe nearly 1 / 3 of the world's surface and atmosphere. In the past decade, meteorological satellite observation data has been widely used to estimate near-surface temperature (Pepin, Maeda et al. 2016, Good, Ghent et al. 2017). The surface temperature inverted by satellite thermal infrared sensors has a strong correlation with the near-surface atmosphere. Therefore, the existing estimation of clear sky near-surface temperature is usually based on the relationship between satellite surface temperature and air temperature.

[0007] When clouds are present, although satellite infrared observations cannot obtain the surface temperature, cloud top temperature and cloud top height can be obtained. Li Huabin et al. (Li et al. 2021) estimated the temperature in the cloud sky based on the vertical temperature lapse rate using the CTH and CTT products of VIIRS / NPP and other auxiliary data. The results show that the estimated temperature RMSE is less than 2°C, and it is feasible to use satellite cloud products to estimate the temperature under cloud conditions. This provides another idea for estimating the temperature in the cloud sky.

[0008] The new generation of geostationary meteorological satellites, such as GOES-R, Himawari-8 / 9 and FY-4A / B, can provide high temporal resolution observations of the Earth's land, ocean and atmosphere, but their spatial resolution is poor, and they can only provide 2-4km observations of surface temperature, cloud top height and cloud top temperature. With the needs of relevant scientific research and business applications such as urban heat island research and climate model refinement, higher requirements are placed on the spatial resolution of temperature. How to obtain all-weather near-surface temperature data with high temporal and spatial resolution has become a focus of attention. At present, there is no research on using geostationary meteorological satellite data to carry out high temporal and spatial resolution near-surface temperature estimation under cloudy conditions. Summary of the invention

[0009] The purpose of the present invention is to solve the defects of the above-mentioned prior art and provide a method for estimating the near-ground temperature of the sky with high spatiotemporal resolution. Specifically, the cloud top height and cloud top temperature of the spatial resolution of the geostationary meteorological satellite, the temperature predicted by the numerical model, the elevation, the vegetation index, the latitude, the longitude and the time information are used to construct a cloud and sky temperature model with the original spatial resolution of the geostationary meteorological satellite based on the neural network model. Assuming that the difference in near-ground temperature between the high spatial resolution sub-pixel and the original spatial resolution pixel of the geostationary meteorological satellite where it is located is caused by the difference in elevation and normalized vegetation index of the two pixels, a temperature difference model between the high spatial resolution sub-pixel and the original spatial resolution of the geostationary meteorological satellite is constructed. Based on the two constructed models, a near-ground temperature estimation model with high spatiotemporal resolution of the sky is derived, and a near-ground temperature estimation model with high spatiotemporal resolution under cloud and sky conditions is constructed using a generative adversarial neural network. The present invention can obtain the spatial distribution of near-ground temperature under high-precision clear sky conditions that is more refined than satellite observations.

[0010] The present invention mainly solves the problem of estimating the near-ground air temperature with high temporal and spatial resolution using the cloud top height and cloud top temperature data of a geostationary meteorological satellite under cloudy conditions.

[0011] A method for estimating near-surface air temperature with high spatial and temporal resolution in the cloud sky comprises the following steps:

[0012] Step 1. Collect the temperature at meteorological observation stations, cloud top height and cloud top temperature from geostationary meteorological satellites, GFS forecast temperature, high spatial resolution elevation, normalized difference vegetation index and other auxiliary data.

[0013] The other auxiliary data include: latitude, longitude, Julian day and observation time information of the pixel of the geostationary meteorological satellite.

[0014] Step 2. Perform temporal and spatial matching of the collected data according to the latitude, longitude and time information of the meteorological observation station to obtain the meteorological observation station temperature, cloud top height and cloud top temperature, GFS forecast temperature, two high spatial resolution elevations (original spatial resolution and high spatial resolution of geostationary meteorological satellites), normalized difference vegetation index and other auxiliary data that match time and space.

[0015] The temporal and spatial matching includes: selecting the satellite observation data closest to the meteorological observation station according to the measurement time information of the meteorological observation station. Further, using the location information of the meteorological observation station to extract the cloud top height, cloud top temperature and longitude and latitude information of the meteorological satellite pixel closest to it. The high spatial resolution elevation and normalized vegetation index are spatially averaged to obtain the lower spatial resolution elevation and normalized vegetation index of the nearest stationary meteorological satellite pixel.

[0016] Step 3. Using the cloud top height and cloud top temperature of the geostationary meteorological satellite spatial resolution, the numerical model predicted temperature, elevation, normalized difference vegetation index, latitude, longitude and time information, a geostationary meteorological satellite original spatial resolution temperature model is constructed based on the neural network model.

[0017] Step 4. Assuming that the near-surface temperature difference between the high spatial resolution sub-pixel and the original spatial resolution pixel of the geostationary meteorological satellite where it is located is caused by the difference in elevation and normalized vegetation index between the two pixels, a temperature difference model between the high spatial resolution sub-pixel and the original spatial resolution of the geostationary meteorological satellite is constructed.

[0018] Step 5. Based on the original spatial resolution temperature model of the geostationary meteorological satellite and the high spatial resolution sub-pixel and geostationary meteorological satellite original spatial resolution temperature difference model, a cloud sky high temporal and spatial resolution near-surface temperature estimation model is derived.

[0019] Step 6. Based on the cloud-to-sky high temporal and spatial resolution near-surface temperature estimation model and the spatiotemporally matched meteorological observation stations, satellite cloud top temperature and cloud top height, GFS forecast temperature, elevation and normalized vegetation index with two spatial resolutions and other auxiliary data, a cloud-to-sky high temporal and spatial resolution near-surface temperature estimation model is constructed based on a generative adversarial neural network.

[0020] Furthermore, in the cloud-sky high temporal and spatial resolution near-surface air temperature estimation method described above, the high spatial resolution elevation and normalized difference vegetation index of step 1 are derived from the 30-meter elevation of SRTM and the 250-meter vegetation index of MODIS, respectively.

[0021] Furthermore, in the cloud-sky high temporal and spatial resolution near-surface temperature estimation method described above, the two spatial resolutions of elevation and normalized vegetation index in step 2 include: the original spatial resolution elevation and normalized vegetation index corresponding to the pixels observed by the stationary meteorological satellite; and the 250m high spatial resolution elevation and normalized vegetation index matching the meteorological station space.

[0022] Further, in the cloud-sky high temporal and spatial resolution near-surface temperature estimation method as described above, step 3 of the geostationary meteorological satellite original spatial resolution near-surface temperature model includes the following contents:

[0023] T air,原始 =f1(CTH,CTT,DEM,NDVI,T GFS,a ,LAT,LON,JD,hour) (1)

[0024] Where T air,低 is the surface temperature at the original spatial resolution of the geostationary meteorological satellite, CTT, CTH, DEM, VDVI, T GFS,a, LAT, LON, JD, and hour are the cloud top temperature, cloud top height, elevation, normalized difference vegetation index, GFS forecast temperature, latitude, longitude, Julian day, and hour of the original spatial resolution of the geostationary meteorological satellite.

[0025] Furthermore, in the cloud sky high temporal and spatial resolution near-ground temperature estimation method as described above, the temperature difference model between the high spatial resolution sub-pixel and the original spatial resolution of the geostationary meteorological satellite in step 4 includes the following contents:

[0026] ΔT air =T air,原始 -T air,高 =f2(DEM 原始 ,DEM 高 ,NDVI 原始 ,NDVI 高 ) (2)

[0027] Where, ΔT air is the temperature difference between the high spatial resolution pixel and the original spatial resolution pixel of the geostationary meteorological satellite, T air,原始 and T air,高 They are the original spatial resolution of the geostationary meteorological satellite, the high-resolution temperature, and the DEM 原始 and DEM 高 They are the original spatial resolution of the geostationary meteorological satellite, the high-resolution elevation, and the NDVI 原始 and NDVI 高 They are respectively the original spatial resolution of geostationary meteorological satellites and the high-resolution vegetation index.

[0028] Further, in the cloud-sky high temporal and spatial resolution near-surface temperature estimation method described above, step 5 deriving the cloud-sky high temporal and spatial resolution near-surface temperature estimation model includes the following contents:

[0029] The cloud-sky high temporal and spatial resolution temperature estimation model includes: According to formula (2),

[0030] T air,高 =T air,原始 -f2(DEM 低 ,DEM 高 ,NDVI 低 ,NDVI 高 ) (3)

[0031] Substituting formula (1) into formula (3), the high temporal and spatial resolution temperature of the sky can be expressed as:

[0032] T air,高 =f1(CTH,CTT,DEM,NDVI,TGFS ,a ,LAT,LON,JD,hour)-f2(DEM原始 ,DEM 高 ,NDVI 原始 ,NDVI 高 ) (4)

[0033] Considering that the above models are all nonlinear estimation functions, the cloud-sky high temporal and spatial resolution temperature estimation model can be directly expressed as:

[0034] T air,高 =f(CTH,CTT,DEM 原始 ,DEM 高 ,NDVI 原始 ,NDVI 高 ,T GFS,a ,LAT,LON,JD,hour) (5)

[0035] This cloud sky high temporal and spatial resolution temperature estimation model can be implemented using neural networks.

[0036] Beneficial effects of the present invention:

[0037] The cloud top height and cloud top temperature data of existing geostationary meteorological satellites have a high spatial resolution of 2-4km, which limits the spatial resolution of the near-ground temperature in the cloud sky based on geostationary meteorological satellites to 2-4km, and cannot meet the demand for high temporal and spatial resolution of near-ground temperature.

[0038] The high spatial resolution estimation model based on high spatial resolution elevation and normalized difference vegetation index data proposed in the present invention can improve the spatial resolution of the estimated near-surface temperature from 2-4km to 250m by introducing the original spatial resolution of the geostationary meteorological satellite, high-resolution elevation and normalized difference vegetation index into the model.

[0039] The present invention directly realizes the estimation of near-surface temperature with high spatial resolution through a model, which effectively reduces the transmission of model errors. In addition, the present invention is also applicable to the near-surface temperature with high spatial resolution under cloudy sky conditions of polar-orbiting meteorological satellites. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 A flow chart of a method for estimating near-surface air temperature with high spatial and temporal resolution in the sky provided by an embodiment of the present invention;

[0041] Figure 2 A two-dimensional histogram of the temperature estimated by the embodiment of the present invention and the station temperature;

[0042] Figure 3 Graph 1 is a spatial distribution diagram of the root mean square error of the station according to an embodiment of the present invention. DETAILED DESCRIPTION

[0043] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention is described clearly and completely below. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0044] like Figure 1 As shown, the present invention mainly solves the problem of estimating near-surface temperature with high temporal and spatial resolution based on geostationary meteorological satellite data. The embodiment of the present invention is based on the cloud top height and cloud top temperature data of the FY-4A satellite imager (AGRI) to carry out near-surface temperature estimation with high temporal and spatial resolution in the sky in Anhui Province. The near-surface temperature estimation with high temporal and spatial resolution provided by the embodiment of the present invention includes the following steps:

[0045] Step 1. Collect the temperature of meteorological observation stations in Anhui Province, the cloud top height and cloud top temperature of the original spatial resolution of the FY-4A satellite imager (AGRI), the GFS forecast temperature, the elevation and normalized difference vegetation index with high spatial resolution, and other auxiliary data.

[0046] The other auxiliary data include: latitude, longitude, Julian day and observation time information of the FY-4A satellite observation point.

[0047] Step 2. Perform temporal and spatial matching of the collected data according to the latitude, longitude and time information of the meteorological observation station to obtain the meteorological observation station temperature, FY-4A satellite imager cloud top height and cloud top temperature, GFS forecast temperature, elevation at two spatial resolutions, normalized difference vegetation index and other auxiliary data that match time and space.

[0048] The temporal and spatial matching includes: selecting the FY-4A satellite observation data closest to the observation according to the measurement time information of the meteorological observation station. Further, using the location information of the meteorological observation station to extract the cloud top height, cloud top temperature and longitude and latitude information of the FY-4A satellite pixel closest to it. The high spatial resolution elevation and normalized vegetation index are spatially averaged to obtain the lower spatial resolution elevation and normalized vegetation index of the nearest FY-4A satellite pixel.

[0049] Step 3. Using the cloud top height and cloud top temperature of the FY-4A satellite spatial resolution, the GFS numerical model predicted temperature, elevation, normalized difference vegetation index, latitude, longitude and time information, a temperature model with the original spatial resolution of the FY-4A satellite was constructed based on the neural network model.

[0050] Step 4. Assuming that the near-surface temperature difference between the high spatial resolution (250 m) sub-pixel and the original spatial resolution pixel of the FY-4A satellite where it is located is caused by the difference in elevation and normalized vegetation index between the two pixels, a temperature difference model between the high spatial resolution sub-pixel and the original spatial resolution of the FY-4A satellite is constructed.

[0051] Step 5. Based on the FY-4A satellite original spatial resolution temperature model, the high spatial resolution (250m) sub-pixel and FY-4A satellite original spatial resolution temperature difference model, a cloud sky high temporal and spatial resolution near-surface temperature estimation model is derived.

[0052] Step 6. Based on the cloud-to-sky high temporal and spatial resolution near-surface temperature estimation model and the spatiotemporally matched meteorological observation stations, FY-4A satellite cloud top temperature and cloud top height, GFS forecast temperature, elevation and normalized vegetation index with two spatial resolutions and other auxiliary data, a cloud-to-sky high temporal and spatial resolution near-surface temperature estimation model is constructed based on a generative adversarial neural network.

[0053] Furthermore, in the cloud-sky high temporal and spatial resolution near-surface air temperature estimation method described above, the high spatial resolution elevation and normalized difference vegetation index of step 1 are derived from the 30-meter elevation of SRTM and the 250-meter vegetation index of MODIS, respectively.

[0054] Furthermore, in the cloud-sky high temporal and spatial resolution near-surface temperature estimation method described above, the two spatial resolutions of elevation and normalized vegetation index in step 2 include: the original spatial resolution elevation and normalized vegetation index corresponding to the FY-4A satellite observation pixels; and the 250m high spatial resolution elevation and normalized vegetation index matching the meteorological station space.

[0055] Further, in the cloud-sky high temporal and spatial resolution near-surface temperature estimation method described above, the FY-4A satellite original spatial resolution near-surface temperature model in step 3 includes the following contents:

[0056] T air,原始 =f1(CTH,CTT,DEM,NDVI,T GFS,a ,LAT,LON,JD,hour) (1)

[0057] Where T air,低 Near-surface temperature at the original spatial resolution of geostationary meteorological satellites, CTT, CTH, DEM, NDVI, T GFS,a , LAT, LON, JD, and hour are the cloud top temperature, cloud top height, elevation, normalized difference vegetation index, GFS forecast temperature, latitude, longitude, Julian day, and hour of the original spatial resolution of the FY-4A satellite.

[0058] Further, in the cloud sky high temporal and spatial resolution near-surface temperature estimation method described above, the temperature difference model between the high spatial resolution sub-pixel and the original spatial resolution of the FY-4A satellite in step 4 includes the following contents:

[0059] ΔT air =T air,原始 -T air,高 =f2(DEM 原始 ,DEM 高 ,NDVI 原始 ,NDVI 高 ) (2)

[0060] Where, ΔT air is the temperature difference between the high spatial resolution pixel and the original spatial resolution pixel of the FY-4A satellite, T air,原始 and T air,高 They are the original spatial resolution of the FY-4A satellite, the high-resolution temperature, and the DEM 原始 and DEM 高 They are the original spatial resolution of the FY-4A satellite, high-resolution elevation, and NDVI. 原始 and NDVI 高 They are respectively the original spatial resolution and high-resolution vegetation index of the FY-4A satellite.

[0061] Further, in the cloud-sky high temporal and spatial resolution near-surface temperature estimation method described above, step 5 deriving the cloud-sky high temporal and spatial resolution near-surface temperature estimation model includes the following contents:

[0062] The cloud-sky high temporal and spatial resolution temperature estimation model includes: According to formula (2),

[0063] T air,高 =T air,原始 -f2(DEM 低 ,DEM 高 ,NDVI 低 ,NDVI 高 ) (3)

[0064] Substituting formula (1) into formula (3), the high temporal and spatial resolution temperature of the sky can be expressed as:

[0065] T air,高 =f1(CTH,CTT,DEM,NDVI,T GFS,a ,LAT,LON,JD,hour)-f2(DEM 原始 ,DEM 高 ,NDVI 原始 ,NDVI 高 ) (4)

[0066] Considering that the above models are all nonlinear estimation functions, the cloud-sky high temporal and spatial resolution temperature estimation model can be directly expressed as:

[0067] T air,高 =f(CTH,CTT,DEM 原始 ,DEM 高 ,NDVI 原始 ,NDVI 高 ,T GFS,a ,LAT,LON,JD,hour) (5)

[0068] This cloud-sky high temporal and spatial resolution temperature estimation model is implemented by a generative adversarial neural network.

[0069] like Figure 2 As shown in Figure 2, the root mean square error between the temperature estimated by AGRI and the temperature observed by the meteorological stations is less than 1.45°C, which is better than the accuracy reported in the existing literature. Figure 3 As shown, it can be seen that the estimated results show good details of temperature changes, which illustrates the good applicability of the algorithm.

[0070] 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 make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for estimating near-surface air temperature with high spatial and temporal resolution in the cloud sky, characterized in that: The method comprises: Step 1. Collect the temperature at the meteorological observation station, the cloud top height and temperature of the geostationary meteorological satellite, the GFS forecast temperature, the high spatial resolution elevation, the normalized difference vegetation index and other auxiliary data; The other auxiliary data include: latitude, longitude, Julian day and observation time information of the pixel of the geostationary meteorological satellite; Step 2. Perform temporal and spatial matching of the collected data according to the latitude and longitude and time information of the meteorological observation station, and obtain the meteorological observation station temperature, cloud top height and cloud top temperature, GFS forecast temperature, elevation at two spatial resolutions, normalized difference vegetation index and other auxiliary data that match in time and space; The time and space matching includes: selecting the satellite observation data closest to the observation according to the measurement time information of the meteorological observation station, extracting the cloud top height, cloud top temperature and longitude and latitude information of the meteorological satellite pixel closest to the meteorological observation station by using the location information of the meteorological observation station, spatially averaging the high spatial resolution elevation and normalized vegetation index, and obtaining the lower spatial resolution elevation and normalized vegetation index of the nearest stationary meteorological satellite pixel; Step 3. Using the cloud top height and cloud top temperature of the geostationary meteorological satellite spatial resolution, the numerical model predicted temperature, elevation, normalized difference vegetation index, latitude, longitude and time information, a near-surface temperature model with the original spatial resolution of the geostationary meteorological satellite is constructed based on the neural network model; Step 4. Assuming that the near-surface temperature difference between the high spatial resolution sub-pixel and the original spatial resolution pixel of the geostationary meteorological satellite where it is located is caused by the difference in elevation and normalized vegetation index between the two pixels, a temperature difference model between the high spatial resolution sub-pixel and the original spatial resolution pixel of the geostationary meteorological satellite is constructed; Step 5. Based on the original spatial resolution temperature model of the geostationary meteorological satellite and the temperature difference model between the high spatial resolution sub-pixel and the original spatial resolution temperature of the geostationary meteorological satellite, a cloud sky high temporal and spatial resolution near-surface temperature estimation model is derived; Step 6. Based on the cloud-to-sky high temporal and spatial resolution near-surface temperature estimation model and the spatiotemporally matched meteorological observation stations, satellite cloud top temperature and cloud top height, GFS forecast temperature, elevation and normalized vegetation index with two spatial resolutions and other auxiliary data, a cloud-to-sky high temporal and spatial resolution near-surface temperature estimation model is constructed based on a generative adversarial neural network.

2. The method for estimating near-surface air temperature with high spatial and temporal resolution in the sky according to claim 1, characterized in that: The high spatial resolution elevation and normalized difference vegetation index of step 1 are respectively derived from the 30-meter elevation of SRTM and the 250-meter vegetation index of MODIS.

3. The method for estimating near-surface air temperature with high spatial and temporal resolution in the sky according to claim 1, characterized in that: The two spatial resolutions of elevation and normalized vegetation index in step 2 are: the original spatial resolution elevation and normalized vegetation index corresponding to the pixels observed by the stationary meteorological satellite; and the 250m high spatial resolution elevation and normalized vegetation index matching the meteorological station space.

4. The method for estimating near-surface air temperature with high spatial and temporal resolution in the sky according to claim 1, characterized in that: The near-surface temperature model of the original spatial resolution of the geostationary meteorological satellite in step 3 is: T air,原始 =f1(CTH,CTT,DEM,NDVI,T GFS,a ,LAT,LON,JD,hour) (1) Among them, T air,原始 is the surface temperature at the original spatial resolution of the geostationary meteorological satellite, CTT, CTH, DEM, NDVI, T GFS,a , LAT, LON, JD, and hour are the cloud top temperature, cloud top height, elevation, normalized difference vegetation index, GFS forecast temperature, latitude, longitude, Julian day, and hour of the original spatial resolution of the geostationary meteorological satellite.

5. The method for estimating near-surface air temperature with high spatial and temporal resolution in the sky according to claim 4, characterized in that: The temperature difference model between the high spatial resolution sub-pixel and the original spatial resolution of the geostationary meteorological satellite in step 4 is: ΔT air =T air,原始 -T air,高 =f2(DEM 原始 ,DEM 高 ,NDVI 原始 ,NDVI 高 ) (2) Where, ΔT air is the temperature difference between the high spatial resolution pixel and the original spatial resolution pixel of the geostationary meteorological satellite, T air,原始 and T air,高 They are the original spatial resolution of the geostationary meteorological satellite, the high-resolution temperature, and the DEM 原始 and DEM 高 They are the original spatial resolution of the geostationary meteorological satellite, the high-resolution elevation, and the NDVI 原始 and NDVI 高 They are respectively the original spatial resolution of geostationary meteorological satellites and the high-resolution vegetation index.

6. A method for estimating near-surface air temperature with high spatial and temporal resolution in the sky according to claim 5, characterized in that: The cloud sky high temporal and spatial resolution near-surface temperature estimation model derived in step 5 is: The cloud-sky high temporal and spatial resolution temperature estimation model includes: According to formula (2), T air,高 =T air,原始 -f2(DEM 低 ,DEM 高 ,NDVI 低 ,NDVI 高 ) (3) Substituting formula (1) into formula (3), the cloud-sky high temporal and spatial resolution temperature is expressed as: T air,高 =f1(CTH,CTT,DEM,NDVI,T GFS,a ,LAT,LON,JD,hour)-f2(DEM 原始 ,DEM 高 ,NDVI 原始 ,NDVI 高 ) (4) Considering that the above models are all nonlinear estimation functions, the cloud-sky high temporal and spatial resolution temperature estimation model is directly expressed as: T air,高 =f(CTH,CTT,DEM 原始 ,DEM 高 ,NDVI 原始 ,NDVI 高 ,T GFS,a ,LAT,LON,JD,hour) (5) This estimation model is implemented using a neural network.

Images