All-weather surface temperature inversion method based on coupling remote sensing observations and surface energy balance
By coupling remote sensing observation and surface energy balance model, cloud mode is identified and combined with remote sensing observation effect, the data loss problem of thermal infrared remote sensing under cloud conditions is solved, and the accurate inversion of all-weather surface temperature is achieved, and the remote sensing inversion is improved. Accuracy and applicability.
Patent Information
- Application Number
- CN202410601490.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-05-15
- Publication Date
- 2024-11-01
- Estimated Expiration
- 2044-05-15
AI Technical Summary
The existing thermal infrared remote sensing technology cannot obtain surface temperature information under cloud conditions, resulting in serious data loss. The existing methods fail to effectively consider the remote sensing observation geometric mechanism and radiation influence of cloud coverage areas, resulting in surface temperature inversion accuracy and application. Insufficient sex.
The all-weather surface temperature inversion method using coupled remote sensing observation and surface energy balance model is used to identify cloud modes, and the cells in remote sensing images are divided into four types. Combined with the surface energy balance theory and remote sensing observation effects, different surface temperature estimation is used. Methods: targeted quantification of the surface thermal radiation of cloud cells to achieve inversion of surface temperature under the cloud.
It improves the accuracy and applicability of surface temperature estimation, provides a continuous and consistent surface temperature product all-weather, overcomes the problems of data loss and insufficient accuracy of existing methods, and achieves more reliable and efficient remote sensing inversion.
Smart Images

Figure CN118467911B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of remote sensing technology application technology, and in particular to an all-weather surface temperature inversion method that couples remote sensing observation geometry and a surface energy balance model. Background Art
[0002] As the skin temperature of the upper layer of the earth's surface, the surface temperature characterizes the interaction between the earth and the atmosphere, energy exchange and water cycle processes. It is an important parameter that cannot be obtained in many disciplines such as surface ecosystem research, surface energy state balance and surface physical parameter solution. Remote sensing observation has the advantages of continuous spatial coverage, short cycle and low cost. It can provide continuous surface temperature data, which is convenient for people to obtain the spatiotemporal distribution characteristics of surface temperature in a timely manner. At present, thermal infrared remote sensing and passive microwave remote sensing are the main methods used for surface temperature inversion. Although passive microwave remote sensing can penetrate clouds and fog to obtain surface information, its spatial resolution is low, the inversion algorithm is not perfect, and there are problems with low resolution and inversion accuracy. Thermal infrared remote sensing surface temperature has high inversion accuracy and is widely used under clear sky conditions. It has become one of the main sources for obtaining global and regional surface temperature products. However, since the thermal infrared spectrum cannot penetrate clouds and fog, surface temperature information cannot be obtained under cloudy conditions.
[0003] Studies have shown that the data loss caused by cloud pollution exceeds 50%, and it is particularly serious during the day. The lack of surface temperature information in cloud pixels has seriously hindered the application and development of surface temperature products. Therefore, developing an all-weather surface temperature estimation method for thermal infrared surface temperature products and achieving a surface temperature product that is consistent in time and space is one of the current research hotspots in the field of thermal infrared remote sensing.
[0004] Clouds play an important role in surface radiation transmission. Their absorption, reflection and scattering effects reduce surface solar radiation, which affects the overall surface radiation balance. The cloud distribution on remote sensing images is closely related to the solar altitude angle and the satellite observation angle. The larger the relative angle between the solar incidence and the sensor observation direction, the more obvious the cloud position offset. In practical applications, remote sensing data detects cloud positions, but the cloud pixels in the image are only the projection of the cloud in the sensor observation direction, which does not fully reflect the impact of clouds on surface radiation.
[0005] The spatiotemporal reconstruction method based on the radiation energy balance model has physical significance and is simple and effective. The algorithm mainly obtains the surface temperature under the cloud by adding the theoretical clear sky surface temperature to the correction term that considers the effect of cloud cover on the surface temperature. The reconstruction of the required theoretical clear sky surface temperature can be achieved through spatiotemporal interpolation based on neighboring pixels, and the correction term estimation method relies on shortwave net radiation or surface net radiation. In this process, there are two main deficiencies. First, due to the influence of the geometric mechanism of remote sensing observation, the surface radiation and temperature of polar-orbiting satellites have remote sensing observation effects, and the existing methods do not consider this aspect. The second is that the radiation data required for the calculation of the parameters required for the correction term are also affected by the remote sensing observation effect, and are mostly obtained by interpolation of neighboring similar pixels. It is difficult to achieve corresponding to a large range of cloud coverage areas, and the applicable area and scope are limited.
[0006] The previous research results of this research group are Chinese patent CN 114564767 A, which discloses a method for estimating the surface temperature under clouds based on the sun-cloud-satellite observation geometry. It is a deep learning algorithm based on remote sensing observation geometry. The algorithm has the following shortcomings: it only considers the difference in the impact of clouds on solar radiation and divides it into two types, but does not consider the differences in the surface energy balance and radiation transfer physical mechanisms of these two types of cloud pixels themselves, and directly uses the same machine learning training model to estimate the surface temperature of all cloud pixels. Summary of the invention
[0007] The technical problem to be solved by the present invention is to provide an all-weather surface temperature inversion method coupling remote sensing observation and surface energy balance in view of the deficiencies in the prior art.
[0008] The technical solution of the present invention is as follows:
[0009] An all-weather surface temperature inversion method coupling remote sensing observation and surface energy balance includes the following steps X1, X2 and X3:
[0010] X1. Identification of cloud patterns
[0011] Based on the geometric relationship of remote sensing observation, pixels in remote sensing images are divided into four types according to whether the solar radiation received by the surface is affected by cloud blocking and whether the pixels are visible in remote sensing observation:
[0012] Area A: Pixels in the non-overlapping area A4 between the cloud projection area A4 and the cloud area A2 are observable but the solar radiation is blocked. This part is the shadow pixel of the cloud projection;
[0013] Area B, where there are pixels in the overlapping area of cloud area A2 and cloud projection area A4, cannot be observed and the solar radiation is blocked by clouds. The change of the surface temperature in this area is based on the clear sky surface temperature, and the impact of clouds on solar radiation also needs to be considered.
[0014] Area C, where there are non-overlapping pixels in the cloud area A2 and the cloud projection area A4, cannot be observed and the solar radiation is not blocked by clouds. The surface temperature in this part is similar to that in the clear sky area;
[0015] Area D: The solar radiation is observable and not blocked by clouds. The initial clear sky area A1 minus the shadow area A pixel is the actual clear sky area.
[0016] X2. All-weather surface temperature algorithm based on coupling remote sensing observations and radiation energy balance model
[0017] X21 Surface Energy Balance Model
[0018] According to the surface energy balance theory, the relationship between energy fluxes is expressed as:
[0019] S n -F n =G+H+LE (1)
[0020] Where S n is the net solar or shortwave radiation flux at the surface, F n is the surface longwave net radiation flux, G is the ground heat flux, H and LE are the sensible heat flux and latent heat flux respectively;
[0021] According to the force recovery method, by considering the vertical temperature gradient of the surface temperature, as well as the soil thermal conductivity and penetration depth, the relationship between the surface heat flux and the surface temperature is derived, which reflects the physical mechanism of the surface energy balance; formula (1) can be further expressed as:
[0022]
[0023] The coefficient a is a function of the properties of the topsoil, b is a function of the local surface and soil properties, and kg is the thermal conductivity of the soil in Wm -1 K -1 ; △Z is the soil depth, set to 0.1m; then, based on the assumption based on adjacent clear sky pixels, it is assumed that the difference in net solar radiation and surface temperature between two pixels or two moments is only caused by clouds; under this assumption, the differential in formula (2) is approximately a finite difference, which means the difference in shortwave net radiation between two adjacent pixels and the difference in surface temperature between two adjacent pixels; further written as:
[0024]
[0025] where λ = k g / ΔZ,1 / K=1-ab / λ,ΔT s is the surface temperature difference between two adjacent pixels, ΔS n is the difference in shortwave net radiation between two adjacent pixels; therefore, the surface temperature of the cloud pixel is expressed as:
[0026]
[0027] T cloud is the surface temperature of cloud pixels, T clear is the surface temperature of the adjacent clear sky pixel in time or space, ΔS n is the shortwave net radiation difference between the two pixels; K is the soil thermal conductivity, which is related to the soil thermal conductivity; According to formulas (3) and (4), the combined K value and ΔS n The surface temperature difference ΔT affected by the cloud can be calculated s ;
[0028] For a given cloud pixel, its daytime surface temperature under the cloud can be composed of two parts: theoretical clear sky surface temperature T clear and cloud pixel surface temperature correction term ΔT s ;
[0029] X22 remote sensing observation effect and surface energy balance coupling algorithm
[0030] When estimating the surface temperature under clouds based on MODIS, the geometric effect of surface temperature remote sensing observation is considered, and the area where solar radiation is blocked by clouds, i.e., area B, and the area where solar radiation is not blocked, i.e., area C, are studied and estimated separately;
[0031] The surface temperature T of all MODIS cloud pixels cloud Represented as the surface temperature of the cloud pixel in area B and the surface temperature of cloud pixels in area C The set of expressions is:
[0032]
[0033] Among them, the surface temperature in area C Corresponding to the theoretical clear sky surface temperature Because the solar radiation is not blocked by clouds, we have:
[0034]
[0035] Surface temperature in zone B Expressed as ideal clear sky surface temperature and the cloud pixel surface temperature correction term ΔT considering the remote sensing observation effect sc The sum of the surface temperature of pixels in area B It is expressed as the following formula:
[0036]
[0037] According to the above surface energy balance model derivation, it can be known that the cloud pixel surface temperature correction term ΔT considering the remote sensing observation effect sc It is expressed as the following formula:
[0038]
[0039] ΔS nc =NSSR actuat -NSSR clear (9)
[0040] Among them, ΔS nc The NSSR is the difference between the actual shortwave net radiation under cloudy conditions and theoretical clear sky conditions considering the effect of remote sensing observation. actual is the actual shortwave net radiation after removing the remote sensing observation effect under cloudy conditions, NSSR clear is the theoretical clear sky shortwave net radiation, and K is the soil thermal conductivity;
[0041] X3. All-weather surface temperature acquisition
[0042] The surface temperature of all cloud pixels can be obtained by formula (5), and then combined with the actual clear sky surface temperature according to formula (10): and the surface temperature of the shadow area The all-weather surface temperature T can be obtained by coupling remote sensing observations and surface energy balance. all-weather ;
[0043]
[0044] Actual clear sky surface temperature and the surface temperature of the shadow area Obtained from original temperature products.
[0045] The method described, the step X1 comprises:
[0046] X11 prepares MODIS cloud detection data MYD35 products, extracts preliminary clear sky area A1 and cloudy area A2, and the surface temperature product corresponding to area A2 is invalid;
[0047] X12 required angle data preparation: prepare the corresponding geometric positioning file MYD03 data to obtain the remote sensing observation zenith angle θ v , Observation azimuth Sun altitude angle θ s , solar azimuth The azimuth range is -180° to 180°, 90° east, -90° west, 180° south, 0° north;
[0048] X13 Calculation of cloud top height above ground: Obtain cloud top height data from the MYD06 cloud height product of the corresponding area, unify it into meters, and subtract the ground height from the MODIS cloud top height, expressed as H;
[0049] X14 converts the file coordinates of the cloud detection data pixel by pixel into the corresponding projection system map coordinates to obtain the corresponding projection system map horizontal and vertical coordinate values;
[0050] The X15 cloud orthophoto position is determined by adding the map displacement increment based on the horizontal and vertical coordinate values of the map coordinates under the existing projection system. and Obtain the horizontal coordinate and the vertical coordinate of the cloud orthophoto projection system, and convert the map coordinate of the cloud orthophoto projection system into the file coordinate, so as to obtain the corresponding cloud orthophoto area A3;
[0051] X16 Determine the cloud projection position: Calculate the shadow projection position of the cloud area A2 as a whole on the ground, and add displacement increments to the X and Y coordinates of the cloud's orthographic projection on the ground. and The X-coordinate and Y-coordinate positions of the corresponding cloud shadow projection are obtained, and the X-coordinate of the cloud projection and the projection system map coordinates are converted into file coordinates, thereby obtaining the corresponding cloud projection range A4;
[0052] X17 Cloud Pattern Recognition: Based on the geometric relationship of remote sensing observations, pixels in remote sensing images are divided into the four types according to whether the solar radiation received by the surface is affected by cloud obstruction and whether the pixels are visible in remote sensing observations.
[0053] The method, in step X2, The estimation method is:
[0054] The harmonic analysis algorithm based on time series marks all pixels in area C of daily surface temperature data as invalid values. The effective surface temperature values throughout the year are collected pixel by pixel using the time series harmonic analysis algorithm, and then the time series interpolation is performed to obtain the theoretical clear sky surface temperature in area C. Theoretical clear sky surface temperature data for region C That is, the cloud pixel in this area corresponds to
[0055] The method, in step X2, The estimation method is:
[0056] Solar radiation is blocked by clouds in area B. According to formula (7-9), the estimation of the surface temperature of cloud pixels in this area requires three parameters: the theoretical clear sky surface temperature in area B The difference between the actual shortwave net radiation and the theoretical clear sky shortwave net radiation after removing the remote sensing observation effect under K value and cloudy conditions, that is, the shortwave net radiation difference considering the influence of remote sensing observation ΔS nc ;
[0057] B21 Estimation of net radiation difference considering the influence of remote sensing observations;
[0058] B211 Estimation of actual shortwave net radiation after eliminating the effect of remote sensing observation;
[0059] B212 theoretical clear sky shortwave net radiation data reconstruction;
[0060] All pixels in the A2 area are marked as invalid values, and the effective shortwave net radiation values throughout the year are collected pixel by pixel. On this basis, the time series harmonic analysis algorithm is used to perform time series interpolation to obtain the clear sky theoretical shortwave net radiation data in the A2 area. Combined with the radiation data in the A1 area, the theoretical clear sky shortwave net radiation data NSSR can be obtained. clear ;
[0061] B213 Estimation of actual shortwave net radiation difference considering the effect of remote sensing observation
[0062] The actual surface shortwave net radiation NSSR is calculated by removing the cloud state from the remote sensing observation effect. actual Subtract the shortwave net radiation NSSR of the theoretical clear sky state clear , we can get the actual shortwave net radiation difference ΔS considering the remote sensing observation effect nc ;
[0063] B22 Estimation of soil thermal conductivity K
[0064] While keeping the soil thermal conductivity unchanged, based on MODIS satellite data and surface temperature products, combined with morning and afternoon observation data and surface shortwave net radiation data, the ratio of shortwave net radiation difference to temperature difference is calculated daily, and the monthly average is further calculated. For the missing pixels, the adjacent pixels are interpolated to fill them, so as to obtain the soil thermal conductivity K value of the area.
[0065] B23 Calculation of surface temperature of cloud pixels blocking solar radiation
[0066] The harmonic analysis algorithm based on time series marks all pixels in area B of the daily surface temperature data as invalid values. The effective surface temperature values throughout the year are collected pixel by pixel using the time series harmonic analysis algorithm, and then the time series interpolation is performed to obtain the theoretical clear sky surface temperature in area B. According to formula (7-8), the K value and the actual shortwave net radiation difference ΔS considering the remote sensing observation effect are nc As input data, the cloud pixel surface temperature correction term ΔT in area B considering the remote sensing observation effect can be obtained.sc , combined with the theoretical clear sky surface temperature Get the final surface temperature value of the cloud pixel in area B
[0067] The method described, step B211 comprises:
[0068] B2111 Calculation of shortwave broadband albedo at the top of the atmosphere: The narrowband reflectivity is converted into the on-board shortwave broadband albedo. The relationship between the narrowband and broadband is established using the MODTRAN model. A parameter lookup table for different observation conditions is established based on the radiation transfer simulation data to obtain the on-board broadband albedo. Then, based on the given observation zenith angle VZA and solar zenith angle SZA, the parameter lookup table is used for linear interpolation to obtain the corresponding on-board broadband albedo.
[0069] B2112 Estimation of shortwave net radiation before remote sensing observation offset correction: The solar zenith angle, atmospheric water vapor, relative azimuth, atmospheric water vapor content, and wide-band albedo of the top of the atmosphere are used as input data to estimate the surface shortwave net radiation. This data is affected by the geometric effect of remote sensing observation and needs to be further corrected;
[0070] B2113 Shortwave radiation correction caused by remote sensing observation offset: Based on the estimated shortwave net radiation before remote sensing observation offset correction, by determining the cloud and its projection range and mapping the position relationship between each cloud pixel and its shadow projection, the shortwave net radiation of the shadow pixel in area A and the cloud pixel in area B is designated as the shortwave net radiation of the corresponding cloud projection area; then, all the pixels in area B are iteratively processed in the image, and the above process is performed one by one;
[0071] B2114 Estimation of underestimated shortwave radiation caused by remote sensing observations: For pixels in area C, they can be illuminated by the sun. Therefore, the shortwave net radiation of these pixels should be consistent with that under clear sky conditions. The clear sky shortwave net radiation data for the whole year of 2020 is used based on the time series harmonic algorithm to further reconstruct this part of the pixel data on the basis of the previous step. Finally, the actual shortwave radiation data NSSR after eliminating the remote sensing observation effect can be obtained. actual .
[0072] The present invention proposes an all-weather surface temperature inversion method that couples remote sensing observations with a surface energy balance model. This method makes full use of the physical mechanism of surface energy balance, considers the vertical temperature gradient of the surface temperature during the calculation process, and uses the relationship between the surface heat flux and the surface temperature to quantify the change in thermal radiation under the clouds. At the same time, it fully considers the dual effects of the remote sensing observation mechanism on the surface temperature and radiation flux, and eliminates the remote sensing observation effect to estimate the actual radiation flux data. In view of the situation where the solar radiation of cloud pixels is blocked by clouds, different surface temperature estimation methods are used to quantify the surface thermal radiation of cloud pixels in a targeted manner, thereby realizing the targeted remote sensing inversion of the surface temperature under the clouds.
[0073] This method not only overcomes the problems existing in existing methods, but also improves the accuracy and applicability of surface temperature estimation, providing a more reliable and efficient solution for surface temperature remote sensing inversion. BRIEF DESCRIPTION OF THE DRAWINGS
[0074] Figure 1 .Technical route for all-weather surface temperature estimation by coupling remote sensing observations and surface energy balance models;
[0075] Figure 2 .Geometric relationship diagram of remote sensing observations under cloudy conditions;
[0076] Figure 3 .Statistical results of cloud pattern recognition of MYD11A1 daytime surface temperature product in 2020;
[0077] Figure 4 .Distribution map of various pixel types for cloud pattern recognition (observation date DOY2020259);
[0078] Figure 5 .Comparison chart of original surface temperature data and time series harmonic analysis simulation data;
[0079] Figure 6 .Comparison chart of original surface shortwave net radiation data and time series harmonic analysis simulation data;
[0080] Figure 7 .Distribution map of clear sky surface temperature on the 15th of each month in 2020;
[0081] Figure 8 .2020 all-weather surface temperature distribution map on the 15th of each month;
[0082] Figure 9. Time series of all-weather surface temperature and surface temperature observed at SURFRAD sites in 2020; Figure 9-1 US-BND station; Figure 9-2 US-FPK station; Figure 9-3 US-PSU Station; Figure 9-4 US-SXF Station; Figure 9-5 US-TBL station;
[0083] Figure 10. Scatter plot of all-weather surface temperature and site verification in 2020; Figure 10-1 US-BND station; Figure 10-2 US-FPK station; Figure 10-3 US-PSU Station; Figure 10-4 US-SXF Station; Figure 10-5 US-TBL station;
[0084] Figure 11. Histogram of all-weather surface temperature and site verification distribution in 2020; Figure 11-1 US-BND station; Figure 11-2 US-FPK station; Figure 11-3 US-PSU Station; Figure 11-4 US-SXF Station; Figure 11-5 US-TBL station; DETAILED DESCRIPTION
[0085] The present invention is described in detail below in conjunction with specific embodiments.
[0086] refer to Figure 1 , an all-weather surface temperature inversion method coupling remote sensing observations and a surface energy balance model, comprising the following steps:
[0087] X1. Identification of cloud patterns
[0088] Depending on whether the surface solar radiation corresponding to the pixel under the cloud is affected by cloud shielding, different methods need to be used to estimate the surface temperature under the cloud. First, the specific positions corresponding to the actual clear sky pixel, shadow pixel, and two types of cloud pixels with different cloud shielding conditions are identified based on the geometric relationship model of solar radiation and remote sensing observations. Based on the geometric relationship of remote sensing observations, the actual position of the cloud can be directly determined by using the cloud height, the zenith angle and azimuth angle of the sun and satellite, and the position of the cloud on the remote sensing image as input data, and the spatial position of the surface shadow can be derived based on the orthophoto position.
[0089] X11 First, prepare the MODIS cloud detection data MYD35 product, extract the preliminary clear sky area A1 and the cloudy area A2, and the surface temperature product corresponding to the A2 area is an invalid value;
[0090] X12 required angle data preparation: prepare the corresponding geometric positioning file MYD03 data to obtain the remote sensing observation zenith angle θ v , Observation azimuth Sun altitude angle θ s , solar azimuth The azimuth range is -180° to 180°, 90° east, -90° west, 180° south, and 0° north.
[0091] X13 Calculation of cloud top height above ground: Obtain cloud top height data from the MYD06 cloud height product of the corresponding area, unify it into meters, and subtract the ground height (DEM product) from the MODIS cloud top height (cloud height product), expressed as H.
[0092] X14 converts the file coordinates of the cloud detection data to the corresponding projection system map coordinates pixel by pixel, and obtains the corresponding projection system map horizontal and vertical coordinate values.
[0093] The orthophoto position of the X15 cloud is determined by adding the map displacement increment based on the horizontal and vertical coordinate values of the map coordinates under the existing projection system. and The horizontal coordinate and the vertical coordinate of the orthophoto projection system of the cloud are obtained, and the map coordinates of the orthophoto projection system of the cloud are converted into file coordinates, so as to obtain the corresponding orthophoto area A3 of the cloud.
[0094] X16 Determine the cloud projection position: Calculate the shadow projection position of the cloud area A2 as a whole on the ground, and add displacement increments to the X and Y coordinates of the cloud's orthographic projection on the ground. and The X-coordinate and Y-coordinate positions of the corresponding cloud shadow projection are obtained, and the X-coordinate of the cloud projection and the projection system map coordinates are converted into file coordinates, thereby obtaining the corresponding cloud projection range A4.
[0095] X17 Cloud Pattern Recognition: Based on the geometric relationship of remote sensing observations, the pixels in remote sensing images are divided into four types according to whether the solar radiation received by the surface is affected by cloud blocking and whether the pixels are visible in remote sensing observations:
[0096] In area A, the pixels in the non-overlapping area A4 between the cloud projection area A4 and the cloud area A2 are observable but the solar radiation is blocked. This part is the shadow pixel of the cloud projection.
[0097] In area B, there are pixels in the overlapping area of cloud area A2 and cloud projection area A4, which cannot be observed and the solar radiation is blocked by clouds. The change of surface temperature in this part is based on the clear sky surface temperature, and the impact of clouds on solar radiation also needs to be considered.
[0098] In area C, there are pixels in the non-overlapping area A2 of cloud area A2 and cloud projection area A4, which are unobservable and the solar radiation is not blocked by clouds. The surface temperature of this part is similar to that of the clear sky area.
[0099] In area D, the solar radiation is observable and not blocked by clouds. The preliminary clear sky area A1 minus the shadow area A pixel is the actual clear sky area.
[0100] Figure 2 The figure is a schematic diagram of the geometric relationship of remote sensing observation under cloudy conditions. It estimates whether the solar radiation of cloud pixels is blocked. Figure 3 It is the statistical result of cloud pattern recognition of MYD11A1 surface temperature product throughout 2020; Figure 4 Distribution map of various types of pixels for cloud pattern identification on observation date DOY2020259.
[0101] X2. Algorithm for surface temperature below clouds by coupling remote sensing observations and radiation energy balance model
[0102] X21 Surface Energy Balance Model Algorithm
[0103] The earth's surface is often covered by clouds, which is one of the main factors of global or regional weather or climate change. Clouds transmit, absorb and reflect solar radiation, and emit long-wave radiation. Clouds control the solar incident radiation and atmospheric long-wave radiation. Different types of clouds have different effects, and medium and low precipitation clouds have a greater impact on solar radiation. Clouds have a decisive influence on the ground radiation flux balance, surface temperature changes and surface energy balance.
[0104] This patent uses the surface energy balance theory to estimate the surface temperature under MODIS clouds. Specifically, according to the surface energy balance theory, the relationship between energy fluxes is expressed as:
[0105] S n -F n =G+H+LE (1)
[0106] Where S n is the net solar or shortwave radiation flux at the surface, F n is the surface longwave net radiation flux, G is the ground heat flux, H and LE are the sensible heat flux and latent heat flux respectively.
[0107] According to the force recovery method, by considering the vertical temperature gradient of the surface temperature, as well as the soil thermal conductivity and penetration depth, the relationship between the surface heat flux and the surface temperature is derived, which reflects the physical mechanism of the surface energy balance. Therefore, formula (1) can be further expressed as:
[0108]
[0109] Then, according to the assumption based on adjacent clear sky pixels, it is assumed that the difference in shortwave net radiation and surface temperature between two pixels or two moments is only caused by clouds. Under this assumption, the differential in Equation 2 can be approximated as a finite difference, which means the difference in shortwave net radiation between two adjacent pixels and the difference in surface temperature between two adjacent pixels. Therefore, it can be further written as:
[0110]
[0111] where λ = k g / ΔZ,1 / K=1-ab / λ,ΔT s is the surface temperature difference between two adjacent pixels (i.e., a pixel covered by clouds and its neighboring pixel under clear sky), ΔS n is the difference in shortwave net radiation between two adjacent pixels; therefore, the surface temperature of the cloud pixel can be expressed as:
[0112]
[0113] T cloudis the surface temperature of cloud pixels, T clear is the surface temperature of the adjacent clear sky pixel in time or space, ΔS n is the shortwave net radiation difference between the two pixels; K is the soil thermal conductivity, which is related to the soil thermal conductivity; According to formulas (3) and (4), the combined K value and ΔS n The surface temperature difference ΔT affected by the cloud can be calculated s ;
[0114] For a given cloud pixel, its daytime surface temperature under the cloud can be composed of two parts: theoretical clear sky surface temperature T clear and cloud pixel surface temperature correction term ΔT s .
[0115] X22 remote sensing observation effect and surface energy balance coupling algorithm
[0116] When estimating the surface temperature under clouds based on MODIS, the geometric effect of surface temperature remote sensing observation is taken into account, and the area where solar radiation is blocked by clouds (area B) and the area where solar radiation is not blocked (area C) are studied and estimated separately.
[0117] The surface temperature T of all MODIS cloud pixels cloud It can be expressed as the surface temperature of the cloud pixel in area B and the surface temperature of cloud pixels in area C The set of expressions is:
[0118]
[0119] Among them, the surface temperature in area C Corresponding to the theoretical clear sky surface temperature Because the solar radiation is not blocked by clouds, we have:
[0120]
[0121] Area B is the projection area of the sun's incident direction, where solar radiation is blocked by clouds. The surface temperature in this area Expressed as ideal clear sky surface temperature and the cloud pixel surface temperature correction term ΔT considering the remote sensing observation effect sc According to formula 4, as long as the shortwave net radiation difference ΔS under cloudy and ideal clear sky conditions considering the geometric effect of remote sensing observation is obtained, nc and K value, the cloud pixel surface temperature correction term ΔT considering the remote sensing observation effect can be obtained according to Formula 3 sc , combined with the theoretical clear sky surface temperature The actual surface temperature of the pixel in area B can be calculated. It is expressed as the following formula:
[0122]
[0123] Based on the surface energy balance model, the cloud pixel surface temperature correction term ΔT considering the remote sensing observation effect is derived. sc It can be expressed as formula (8-9):
[0124]
[0125] ΔS nc =NSSR actual -NSSR clear (9)
[0126] Among them, ΔS nc The NSSR is the difference between the actual shortwave net radiation under cloudy conditions and theoretical clear sky conditions considering the effect of remote sensing observation. actual is the actual shortwave net radiation after removing the remote sensing observation effect under cloudy conditions, NSSR clear is the theoretical clear sky shortwave net radiation, and K is the soil thermal conductivity coefficient.
[0127] B1. Estimates
[0128] The surface temperature of the cloud pixels in area C is equivalent to the theoretical clear sky surface temperature. The surface temperature has periodic variation characteristics on the annual and daily scales. The time series harmonic analysis method is a combination of the least squares method and Fourier transform, which combines the two methods of smoothing and filtering. The iterative fitting of the least squares method is used to remove the points in the time series values that are less affected by cloud pollution. The decomposition and reconstruction of the curve are achieved by means of the forward and inverse transforms of Fourier in the time domain and frequency domain, thereby achieving the purpose of cloud removal and reconstruction of time series remote sensing images. The time series harmonic decomposition method fully considers the characteristics of vegetation growth periodicity and the change cycle of the data itself when reconstructing images. It can reconstruct time series images with frequency curves representing different growth cycles, and truly reflect the periodic variation law of the data.
[0129] Therefore, the harmonic analysis algorithm of time series can effectively capture the periodicity and trend of temperature changes, and the annual periodic variation law of surface temperature can be used to reconstruct these pixels based on the harmonic analysis algorithm of time series. Figure 5 It is a comparison chart of the original surface temperature data and the smoothed data set of the harmonic analysis algorithm of the time series. Table 1 shows the parameter settings of the harmonic analysis algorithm of the time series.
[0130] In the specific implementation, the surface temperature data of the study area in 2020 is used. The daily surface temperature data in area C are marked as invalid values based on the harmonic analysis algorithm of the time series. The effective surface temperature values throughout the year are collected pixel by pixel using the time series harmonic analysis algorithm. Then, the time series interpolation or the integrated regression model with clear sky surface temperature as the target variable is performed to obtain the theoretical clear sky surface temperature in area C. As mentioned above, the theoretical clear sky surface temperature data for region C That is, the cloud pixel in this area corresponds to
[0131] Table 1 Parameter settings for harmonic analysis of ideal clear sky surface temperature estimation time series (unit: K)
[0132]
[0133] B2, Estimates
[0134] Solar radiation is blocked by clouds in area B. According to formula (7-9), the estimation of the surface temperature of cloud pixels in this area requires three parameters: the theoretical clear sky surface temperature in area B The difference between the actual shortwave net radiation and the theoretical clear sky shortwave net radiation after removing the remote sensing observation effect under K value and cloudy conditions, that is, the shortwave net radiation difference considering the influence of remote sensing observation ΔS nc .
[0135] B21 Estimation of net radiation difference considering the influence of remote sensing observations
[0136] B211 Estimation of actual shortwave net radiation after eliminating remote sensing observation effect
[0137] Based on the surface energy balance model, shortwave net radiation is a key input parameter for estimating the surface temperature under clouds. Shortwave net radiation data is affected by remote sensing observation effects and has obvious overestimation or underestimation. The distribution of clouds appears to be offset in remote sensing images due to the remote sensing observation effect.
[0138] 1. Calculation of shortwave broadband albedo at the top of the atmosphere: The narrowband reflectivity is converted into the on-board shortwave broadband albedo. The relationship between the narrowband and the broadband is established using the MODTRAN model. A parameter lookup table for different observation conditions is established based on the radiation transfer simulation data to obtain the on-board broadband albedo. Then, based on the given observation zenith angle (VZA) and solar zenith angle (SZA), the parameter lookup table is used for linear interpolation to obtain the corresponding on-board broadband albedo.
[0139] 2. Estimation of shortwave net radiation before remote sensing observation offset correction: Based on the parameterized relationship of surface shortwave net radiation established by the parameterized model proposed by Li et al. (1993), the solar zenith angle, atmospheric water vapor, relative azimuth, atmospheric water vapor content and wide-band albedo of the top of the atmosphere are used as input data to estimate the surface shortwave net radiation. This data is affected by the geometric effect of remote sensing observations and needs to be further corrected.
[0140] 3. Correction of shortwave radiation caused by offset of remote sensing observations: Based on the estimated shortwave net radiation before offset correction of remote sensing observations, by determining the cloud and its projection range and mapping the positional relationship between each cloud pixel and its shadow projection, the shortwave net radiation of the shadow pixel in area A and the cloud pixel in area B is designated as the shortwave net radiation of the corresponding cloud projection area. Then, all pixels in area B are iteratively processed in the image, and the above process is performed one by one.
[0141] 4. Estimation of underestimated shortwave radiation caused by remote sensing observations: For pixels in area C, they can be illuminated by the sun. Therefore, the shortwave net radiation of these pixels should be consistent with that under clear sky conditions. The clear sky shortwave net radiation data for the whole year of 2020 is used based on the time series harmonic algorithm to further reconstruct this part of the pixel data on the basis of the previous step. Finally, the actual shortwave radiation data NSSR can be obtained after eliminating the remote sensing observation effect. actual .
[0142] Reconstruction of B212 theoretical clear sky shortwave net radiation data
[0143] The surface shortwave net radiation has a significant annual variation cycle. The theoretical clear-sky shortwave net radiation can be estimated by using methods such as the harmonic analysis algorithm of the time series and the integrated regression model with the clear-sky shortwave net radiation as the target variable. Figure 6 The original surface shortwave net radiation data and the schematic diagram of the harmonic analysis simulation data of the time series are shown. Table 2 shows the parameter settings when estimating the theoretical clear sky surface shortwave net radiation using the harmonic analysis algorithm of the time series.
[0144] In the specific implementation, the shortwave net radiation data of the study area in 2020 was used to mark all pixels in the A2 area as invalid values, and the effective shortwave net radiation values for the whole year were collected pixel by pixel. On this basis, the time series harmonic analysis algorithm was used to perform time series interpolation to obtain the clear sky theoretical shortwave net radiation data of the A2 area. Combined with the radiation data of the A1 area, the theoretical clear sky shortwave net radiation data NSSR can be obtained. clear .
[0145] Table 2 Harmonic analysis parameter settings for theoretical clear sky surface shortwave net radiation estimation time series (unit: W / m 2 )
[0146]
[0147] B213 Estimation of actual shortwave net radiation difference considering the effect of remote sensing observation
[0148] The actual surface shortwave net radiation NSSR is calculated by removing the cloud state from the remote sensing observation effect. actual Subtract the shortwave net radiation NSSR of the theoretical clear sky state clear (Assuming the shortwave net radiation in clear sky), the actual shortwave net radiation difference ΔS considering the remote sensing observation effect can be obtained. nc .
[0149] B22 Estimation of soil thermal conductivity K
[0150] The thermal conductivity of soil is mainly determined by its composition and proportion. The size of soil thermal conductivity mainly depends on the soil composition and its proportion. It is assumed that soil thermal conductivity will not change significantly in a certain area in a short period of time. While keeping the soil thermal conductivity unchanged, based on MODIS satellite data and surface temperature products, combined with morning and afternoon observation data and surface shortwave net radiation data, the ratio of shortwave net radiation difference to temperature difference is calculated daily, and the monthly average is further calculated. For the missing pixels, the spatial neighboring pixels are interpolated to fill them, so as to obtain the soil thermal conductivity coefficient K value of the area.
[0151] B23 Calculation of surface temperature of cloud pixels blocking solar radiation
[0152] The harmonic analysis algorithm based on time series marks all pixels in area B of the daily surface temperature data as invalid values. The effective surface temperature values throughout the year are collected pixel by pixel using the time series harmonic analysis algorithm, and then the time series interpolation is performed, or an integrated regression model with clear sky surface temperature as the target variable is used to obtain the clear sky theoretical surface temperature in area B. According to formula (7-8), the K value and the actual shortwave net radiation difference ΔS considering the remote sensing observation effect are nc As input data, the cloud pixel surface temperature correction term ΔT in area B considering the remote sensing observation effect can be obtained. sc , combined with the theoretical clear sky surface temperature Get the final surface temperature value of the cloud pixel in area B
[0153] X3, all-weather surface temperature
[0154] The surface temperature of all cloud pixels can be obtained by formula (5), and then combined with the actual clear sky surface temperature according to formula (10): and the surface temperature of the shadow area The all-weather surface temperature T can be obtained by coupling remote sensing observations and surface energy balance. all-weather ;
[0155]
[0156] Actual clear sky surface temperature and the surface temperature of the shadow area Obtained from original temperature products.
[0157] Figure 7 This is the clear sky surface temperature distribution map on the 15th of each month in 2020. Figure 8 This is the all-weather surface temperature distribution map on the 15th of each month in 2020. Compared with previous studies, this patent distinguishes the conditions in which solar radiation is blocked by cloud pixels, and specifically estimates the surface temperature of cloud pixels. Compared with the previous unified training model calculation results, the boundary between the all-weather surface temperature estimated in the patent and the original clear sky surface temperature is more natural, avoiding some abnormal mutations that existed in previous studies. The same is true inside the cloud pixels. The spatial distribution of surface temperature is more reasonable, and can more accurately reflect the spatial distribution pattern of surface thermal radiation. This proves the effectiveness of the algorithm and reflects its advantages over previous algorithms in terms of spatial distribution.
[0158] In order to verify the accuracy of the surface temperature under the cloud in MODIS data, the measured data of the Surface Radiation Budget Network (SURFRAD) were collected for verification. The data download address is http: / / www.esrl.noaa.gov / gmd / grad / SURFRAD / overview.html.
[0159] The main measurement data of the Surface Radiation Budget (SURFRAD) network are solar and infrared radiation, and auxiliary observations include solar direct and diffuse reflection, photosynthetically active radiation, solar spectrum and meteorological parameters. In addition, uplink / downlink radiation flux observation data can also be obtained, which can be used to derive the site surface temperature.
[0160] Based on the Stefan-Boltzmann law of radiation transfer theory, the surface temperature at each station is calculated using the atmospheric upwelling and downwelling longwave radiation flux data observed by the SURFRAD station. The expression is:
[0161]
[0162] In the formula, F ↑ and F ↓ It is the uplink and downlink thermal radiation flux (3-50 μm) measured synchronously during the satellite transit time, in W / m 2 , σ represents the Stefan-Boltzman constant (5.670367×10 -8 W m -2 ·K-4 ), ε b represents the broadband surface emissivity, which is set to 0.97.
[0163] The present invention collected observations of different surface cover types at different periods throughout 2020 from 5 stations for the verification of surface temperature. Figure 9 is a time series diagram of the all-weather surface temperature in 2020 and the surface temperature observed at the SURFRAD station. As can be seen from the figure, the surface temperature under the clouds estimated by the new algorithm is basically consistent with the surface temperature change trend observed at the station, and the difference between the two is basically within 3K. In addition, the scatter plots and histograms of the surface temperature under the clouds and the corresponding station observation data throughout 2020 were statistically analyzed, and the quantitative evaluation indicators root mean square error RMSE, bias, mean Mean, standard deviation STD and determination coefficient R were calculated. 2 . Figure 10 is a scatter plot of all-weather surface temperature and site verification in 2020, and Figure 11 is a distribution histogram of the difference between all-weather and site surface temperatures in 2020. According to the results of quantitative analysis, after removing outliers, the RMSE of all-weather surface temperature is mainly concentrated between 2.5K and 5K, and the determination coefficient is between 0.71 and 0.95. Except for individual sites with systematic deviations, the deviation (bias) between the estimated surface temperature and the site observation value is around 1K, and Figure 11 shows the average value of the difference between the surface temperature of all cloud pixels and the site observed surface temperature, all of which are within 1K, or even close to 0K. In general, the algorithm proposed in the patent has high accuracy. Moreover, compared with previous results, the patented algorithm adopts a method of direct comparison based on daily site data throughout the year, and a cross-validation method with assimilated data with lower resolution. The verification results are more credible and the accuracy verification results are more reasonable.
[0164] It should be understood that those skilled in the art can make improvements or changes based on the above description, and all these improvements and changes should fall within the scope of protection of the appended claims of the present invention.
Claims
1. An all-weather surface temperature inversion method that couples remote sensing observations and surface energy balance, characterized in that: The following steps are included: X1, X2 and X3: X1. Identification of cloud patterns X11 prepares MODIS cloud detection data MYD35 products, extracts preliminary clear sky area A1 and cloud area A2, and the surface temperature product corresponding to cloud area A2 is an invalid value; X12 required angle data preparation: prepare the corresponding geometric positioning file MYD03 data to obtain the remote sensing observation zenith angle θ v , Observation azimuth Sun altitude angle θ s , solar azimuth The azimuth range is -180° to 180°, 90° east, -90° west, 180° south, 0° north; X13 Calculation of cloud top height above ground: Obtain cloud top height data from the MYD06 cloud height product of the corresponding area, unify it into meters, and subtract the ground height from the MODIS cloud top height, expressed as H; X14 converts the file coordinates of the cloud detection data pixel by pixel into the corresponding projection system map coordinates to obtain the corresponding projection system map horizontal and vertical coordinate values; The X15 cloud orthophoto position is determined by adding the map displacement increment based on the horizontal and vertical coordinate values of the map coordinates under the existing projection system. and Obtain the horizontal coordinate and the vertical coordinate of the cloud orthophoto projection system, convert the map coordinate of the cloud orthophoto projection system into the file coordinate, and thus obtain the corresponding cloud orthophoto area A3; X16 Determine the cloud projection position: Calculate the shadow projection position of the cloud area A2 as a whole on the ground, and add displacement increments to the X and Y coordinates of the cloud's orthographic projection on the ground. and The X-coordinate and Y-coordinate positions of the corresponding cloud shadow projection are obtained, and the X-coordinate of the cloud projection and the projection system map coordinates are converted into file coordinates, thereby obtaining the corresponding cloud projection area A4; X17 Cloud Pattern Recognition: Based on the geometric relationship of remote sensing observations, the pixels in remote sensing images are divided into the following four types according to whether the solar radiation received by the surface is affected by cloud blocking and whether the pixels are visible in remote sensing observations: Area A: Pixels in the non-overlapping area A4 between the cloud projection area A4 and the cloud area A2 are observable but the solar radiation is blocked. This part is the shadow pixel of the cloud projection; Area B, where there are pixels in the overlapping area of cloud area A2 and cloud projection area A4, cannot be observed and the solar radiation is blocked by clouds. The change of the surface temperature in this area is based on the clear sky surface temperature, and the impact of clouds on solar radiation needs to be considered. Area C, where there are non-overlapping pixels in the cloud area A2 and the cloud projection area A4, cannot be observed and the solar radiation is not blocked by clouds. The surface temperature in this part is similar to that in the clear sky area; Area D: The solar radiation is observable and not blocked by clouds. The initial clear sky area A1 minus the shadow area A pixel is the actual clear sky area. X2. All-weather surface temperature algorithm based on coupling remote sensing observations and radiation energy balance model X21 Surface Energy Balance Model According to the surface energy balance theory, the relationship between energy fluxes is expressed as: S n -F n =G+H+LE (1) Where S n is the net solar or shortwave radiation flux at the surface, F n is the surface longwave net radiation flux, G is the ground heat flux, H and LE are the sensible heat flux and latent heat flux respectively; According to the force recovery method, by considering the vertical temperature gradient of the surface temperature, as well as the soil thermal conductivity and penetration depth, the relationship between the surface heat flux and the surface temperature is derived, which reflects the physical mechanism of the surface energy balance; formula (1) can be further expressed as: The coefficient a is a function of the properties of the topsoil, and b is a function of the local surface and soil properties; kg is the thermal conductivity of the soil in Wm -1 K -1 ; ΔZ is the soil depth, set to 0.1m; then, based on the assumption based on adjacent clear sky pixels, it is assumed that the difference in net solar radiation and surface temperature between two pixels or two moments is only caused by clouds; under this assumption, the differential in formula (2) can be approximated as a finite difference, which means the difference in shortwave net radiation between two adjacent pixels and the difference in surface temperature between two adjacent pixels; further written as: where λ = k g / ΔZ,1 / K=1-ab / λ,ΔT s is the surface temperature difference between two adjacent pixels, ΔS n is the difference in shortwave net radiation between two adjacent pixels; Therefore, the surface temperature of the cloud pixel is expressed as: T cloud is the surface temperature of cloud pixels, T clear is the surface temperature of the adjacent clear sky pixel in time or space, ΔS n is the shortwave net radiation difference between the two pixels; K is the soil thermal conductivity, which is related to the soil thermal conductivity; According to formulas (3) and (4), the combined K value and ΔS n The surface temperature difference ΔT affected by the cloud can be calculated s ; For a given cloud pixel, its daytime surface temperature under the cloud consists of two parts: theoretical clear sky surface temperature T clear and cloud pixel surface temperature correction term ΔT s ; X22 remote sensing observation effect and surface energy balance coupling algorithm When estimating the surface temperature under clouds based on MODIS, the geometric effect of surface temperature remote sensing observation is considered, and the area where solar radiation is blocked by clouds, i.e., area B, and the area where solar radiation is not blocked, i.e., area C, are studied and estimated separately; The surface temperature T of all MODIS cloud pixels cloud It can be expressed as the surface temperature of the cloud pixel in area B and the surface temperature of cloud pixels in area C The set of expressions is: Among them, the surface temperature in area C Corresponding to the theoretical clear sky surface temperature Because the solar radiation is not blocked by clouds, we have: Surface temperature in zone B Expressed as ideal clear sky surface temperature and the cloud pixel surface temperature correction term ΔT considering the remote sensing observation effect sc The sum of the surface temperature of pixels in area B It is expressed as the following formula: According to the above surface energy balance model derivation, it can be known that the cloud pixel surface temperature correction term ΔT considering the remote sensing observation effect sc It is expressed as the following formula: ΔS nc =NSSR actual -NSSR clear (9) Among them, ΔS nc The NSSR is the difference between the actual shortwave net radiation under cloudy conditions and theoretical clear sky conditions considering the effect of remote sensing observation. actual is the actual shortwave net radiation after removing the remote sensing observation effect under cloudy conditions, NSSR clear is the theoretical clear sky shortwave net radiation, and K is the soil thermal conductivity; X3. All-weather surface temperature acquisition The surface temperature of all cloud pixels can be obtained by formula (5), and then combined with the actual clear sky surface temperature according to formula (10): and the surface temperature of the shadow area The all-weather surface temperature T can be obtained by coupling remote sensing observations and surface energy balance. all-weather ; Actual clear sky surface temperature and the surface temperature of the shadow area Obtained from original temperature products.
2. The method according to claim 1, characterized in that In the step X2, The estimation method is: The harmonic analysis algorithm based on time series marks all pixels in the C area of daily surface temperature data as invalid values. The effective surface temperature values throughout the year are collected pixel by pixel using the time series harmonic analysis algorithm, and then the time series interpolation is performed to obtain the clear sky theoretical clear sky surface temperature in the C area. Theoretical clear sky surface temperature data for region C That is, the cloud pixel in this area corresponds to 3. The method according to claim 1, characterized in that In the step X2, The estimation method is: Solar radiation is blocked by clouds in area B. According to formula (7-9), the surface temperature estimation of cloud pixels in this area requires three parameters: the theoretical clear sky surface temperature in area B The difference between the actual shortwave net radiation and the theoretical clear sky shortwave net radiation after removing the remote sensing observation effect under K value and cloudy conditions, that is, the shortwave net radiation difference considering the influence of remote sensing observation ΔS nc ; B21 Estimation of net radiation difference considering the influence of remote sensing observations; B211 Estimation of actual shortwave net radiation after eliminating the effect of remote sensing observation; B212 theoretical clear sky shortwave net radiation data reconstruction; All pixels in area A2 are marked as invalid values, and the effective shortwave net radiation values throughout the year are collected pixel by pixel. On this basis, the time series harmonic analysis algorithm is used to perform time series interpolation to obtain the clear sky theoretical shortwave net radiation data in area A2. Combined with the radiation data in area A1, the theoretical clear sky shortwave net radiation data NSSR can be obtained. clear ; B213 Estimation of actual shortwave net radiation difference considering the effect of remote sensing observation The actual surface shortwave net radiation NSSR is calculated by removing the cloud state from the remote sensing observation effect. actual Subtract the shortwave net radiation NSSR of the theoretical clear sky state clear , we can get the actual shortwave net radiation difference ΔS considering the remote sensing observation effect nc ; B22 Estimation of soil thermal conductivity K Under the condition of keeping the soil thermal conductivity unchanged, based on MODIS satellite data and surface temperature products, combined with morning and afternoon observation data and surface shortwave net radiation data, the ratio of shortwave net radiation difference to temperature difference is calculated daily, and the monthly average is further calculated. For the missing pixels, the adjacent pixels are interpolated to fill them, so as to obtain the soil thermal conductivity coefficient K value of the area. B23 Calculation of surface temperature of cloud pixels blocking solar radiation The harmonic analysis algorithm based on time series marks all pixels in area B of the daily surface temperature data as invalid values. The effective surface temperature values throughout the year are collected pixel by pixel using the time series harmonic analysis algorithm, and then the time series interpolation is performed to obtain the theoretical clear sky surface temperature in area B. According to formula (7-8), the K value and the actual shortwave net radiation difference ΔS considering the remote sensing observation effect are nc As input data, the cloud pixel surface temperature correction term ΔT considering the remote sensing observation effect in area B can be obtained. sc , combined with the theoretical clear sky surface temperature Get the final surface temperature value of the cloud pixel in area B 4. The method according to claim 3, characterized in that: The step B211 comprises: B2111 Calculation of shortwave broadband albedo at the top of the atmosphere: The narrowband reflectivity is converted into the on-board shortwave broadband albedo. The relationship between the narrowband and broadband is established using the MODTRAN model. A parameter lookup table for different observation conditions is established based on the radiation transfer simulation data to obtain the on-board broadband albedo. Then, based on the given observation zenith angle VZA and solar zenith angle SZA, the parameter lookup table is used for linear interpolation to obtain the corresponding on-board broadband albedo. B2112 Estimation of shortwave net radiation before remote sensing observation offset correction: The solar zenith angle, atmospheric water vapor, relative azimuth, atmospheric water vapor content, and wide-band albedo of the top of the atmosphere are used as input data to estimate the surface shortwave net radiation. This data is affected by the geometric effect of remote sensing observation and needs to be further corrected; B2113 Shortwave radiation correction caused by remote sensing observation offset: Based on the estimated shortwave net radiation before remote sensing observation offset correction, by determining the cloud and its projection range and mapping the position relationship between each cloud pixel and its shadow projection, the shortwave net radiation of the shadow pixel in area A and the cloud pixel in area B is designated as the shortwave net radiation of the corresponding cloud projection area; then, all the pixels in area B are iteratively processed in the image, and the above process is performed one by one; B2114 Estimation of underestimated shortwave radiation caused by remote sensing observations: For pixels in area C, they can be illuminated by the sun. Therefore, the shortwave net radiation of these pixels should be consistent with that under clear sky conditions. The clear sky shortwave net radiation data for the whole year of 2020 is used based on the time series harmonic algorithm to further reconstruct this part of the pixel data on the basis of the previous step. Finally, the actual shortwave radiation data NSSR after eliminating the remote sensing observation effect can be obtained. actual .
Citation Information
Patent Citations
Sun-cloud-satellite observation geometry-based under-cloud surface temperature estimation method
CN114564767A
All-weather surface temperature near-real-time inversion method fused with multi-source satellite remote sensing
CN113158570A
Method for responding to surface temperature of afforestation based on multi-source remote sensing and energy balance model
CN113343408A
Under-cloud surface temperature reconstruction method based on surface temperature annual change model
CN114218756A
KR20230112887A
