A Spatial Downscaling Method for Land Surface Temperature of Radiation Balance

Abstract

The present invention relates to a method for spatially downscaling the surface temperature of radiation balance, belonging to the technical field of surface temperature image enhancement. It includes: S1 spatially upscaling the vegetation coverage in the visible light band to the spatial resolution of the thermal infrared band, and screening the pixel set where the vegetation coverage is between the minimum and maximum values of the vegetation coverage of each pixel in the visible light band corresponding to the target pixel in the thermal infrared band; S2 re-screening the pixel set screened in S1 according to the mean and variance of the surface temperature of the pixel set screened in S1 according to the normal distribution; S3 sorting the pixel set screened in S2 in ascending order of the upscaled vegetation coverage, and using the sorting result to calculate the change rate and intercept of the surface temperature with respect to the vegetation coverage; S4 spatially downscaling the surface temperature of the thermal infrared band according to the change rate and intercept obtained in S3 and the vegetation coverage of each pixel in the visible light band; S5 adjusting the spatially downscaled surface temperature according to the radiation balance so that the surface temperature before and after spatial downscaling satisfies the radiation balance.

Description

Technical Field

[0001] The present invention relates to a method for enhancing the spatial details of land surface temperature in the thermal infrared band of remote sensing, and particularly to a method for downscaling the spatial land surface temperature that follows the law of land surface radiation energy balance. Background Art

[0002] As one of the most important variables in land surface processes, land surface temperature is an important indicator for characterizing land surface energy budget, land-atmosphere energy exchange, land-atmosphere water and heat exchange, and climate change research. Monitoring the land surface temperature at the regional or global scale in the thermal infrared band of remote sensing based on Planck's law is currently the only feasible means. However, due to the relatively large instantaneous field of view of satellite thermal infrared band payloads, the spatial resolution of the land surface temperature retrieved in the thermal infrared band is often lower than that of visible light payloads. Therefore, how to improve the spatial resolution of the land surface temperature in the thermal infrared band has become a research hotspot in the fields of thermal infrared remote sensing applications, data fusion, data analysis, etc. How to conveniently, quickly, and accurately obtain the land surface temperature over a large area with high precision has become a technical problem in dealing with issues such as climate warming, environmental degradation, and ecological health.

[0003] Using visible light band information to enrich the spatial details of land surface temperature in the thermal infrared band is an important approach. Currently, the methods based on this idea are mainly divided into three categories: statistical, modulation allocation, and spectral mixture models. The statistical method assumes that "the relationship spatial scale is invariant" and applies the statistical relationship between low-spatial-resolution land surface temperature and the regression kernel to high spatial resolution. However, due to the strong spatial heterogeneity of land surface temperature, "the relationship spatial scale is invariant" is basically not satisfied in practice. Modulation allocation distributes the land surface temperature in the thermal infrared band to each sub-pixel according to a certain proportion, and the distribution factors include the panchromatic band, emissivity, land surface temperature of other sensors in the same season, etc. However, this method cannot cover the change rate of land surface temperature with respect to the distribution factors. The spectral mixture model directly correlates high- and low-spatial-resolution land surface temperatures based on linear spectral mixture, and then regresses to solve the high-spatial-resolution land surface temperature. However, this method cannot distinguish the difference between the relationship between land surface temperature in the thermal infrared band and high-spatial-resolution information. These three categories of methods usually ignore the clear physical background and quantitative requirements of thermal infrared remote sensing, and cannot guarantee the consistency of radiation energy before and after downscaling the spatial land surface temperature in the thermal infrared band.

[0004] The limitations of the above three categories of methods are mainly due to the lack of pertinence in solving the relationship between land surface temperature in the thermal infrared band and the land surface temperature after spatial downscaling, and the insufficient consideration of the differences in the variation laws of land surface temperature in different environments. In order to fully explore the connection methods between different spatial scales of land surface temperature in the thermal infrared band by means of visible light band information, to reveal the local differentiation law of land surface temperature with environmental changes by combining different wavelength electromagnetic spectrum information, and to enhance the clarity of land surface temperature in the thermal infrared band, a method for downscaling the spatial land surface temperature with radiation balance is invented. Summary of the Invention

[0005] To solve the above technical problems, the present invention provides a method for spatially downscaling the land surface temperature of radiation balance.

[0006] To achieve the above object, the technical solution adopted by the present invention is as follows:

[0007] A method for spatially downscaling the land surface temperature of radiation balance, comprising the following steps:

[0008] S1: Spatially upscale the vegetation coverage in the visible light band to obtain the upscaled vegetation coverage with the spatial resolution of the thermal infrared band, and screen out the pixel set with the vegetation coverage between the minimum and maximum values of the vegetation coverage of each pixel in the visible light band corresponding to the target pixel in the thermal infrared band on the upscaled vegetation coverage;

[0009] S2: Rescreen the pixel set screened in S1 according to the mean and standard deviation of the land surface temperature of the pixel set screened in S1, that is, only retain the pixels with the land surface temperature between plus or minus 1.5 times the standard deviation of the mean according to the normal distribution;

[0010] S3: Sort the land surface temperature and the upscaled vegetation coverage of the pixel set screened in S2 in ascending order of the upscaled vegetation coverage, and calculate the change rate and intercept of the land surface temperature with respect to the vegetation coverage at the thermal infrared target pixel using the sorted land surface temperature and upscaled vegetation coverage;

[0011] S4: Calculate the spatially downscaled land surface temperature with the spatial resolution of the visible light band corresponding to the thermal infrared target pixel according to the change rate and intercept obtained in S3 and the vegetation coverage of each pixel in the visible light band corresponding to the thermal infrared target pixel;

[0012] S5: Adjust the spatially downscaled land surface temperature obtained in S4 according to the radiation balance to make the land surface temperature at the thermal infrared target pixel satisfy the radiation balance with the land surface temperature of the corresponding spatially downscaled pixel.

[0013] Further, the S1 is specifically:

[0014] P T ={p|fvc Omin ≤FVC p ≤fvc Omax}

[0015]

[0016] FVC = Γ(fvc)

[0017] In the formula, fvc is the vegetation coverage in the visible light band, Γ represents the spatial upscaling model, FVC represents the vegetation coverage obtained by upscaling fvc to the spatial resolution of the thermal infrared band, the subscript T represents the thermal infrared band resolution, the subscript O represents the target pixel in the land surface temperature (LST) of the thermal infrared band, R represents the ratio of the LST spatial resolution to the fvc spatial resolution, R×R represents the number of pixels corresponding to the target pixel O in fvc, and fvc Oij|1≤i , j≤R represents the vegetation coverage of the pixel in the i-th row and j-th column corresponding to the target pixel O in fvc, and fvc Omin represents the minimum value of the vegetation coverage of the R×R pixels corresponding to the target pixel O in fvc, and fvc Omax represents the maximum value of the vegetation coverage of the R×R pixels corresponding to the target pixel O in fvc, and FVC p represents the vegetation coverage of the pixel p in FVC that lies between fvc Omin and fvc Omax , and P T represents the set composed of all pixels p.

[0018] Furthermore, the specific steps of the said S2 are as follows:

[0019] S21: Calculate the mean and variance of the land surface temperature in the thermal infrared band of the pixel set P T screened by S1, specifically:

[0020]

[0021]

[0022] In the formula, q is the pixel in the pixel set P T , LST q is the land surface temperature in the thermal infrared band of the pixel q, and FVC q is the vegetation coverage of the pixel q, is the weighted mean of the land surface temperatures in the thermal infrared band of all pixels in P T , and is the weighted standard deviation of the land surface temperatures in the thermal infrared band of all pixels in P T ;

[0023] S22: According to and σ PT , screen the pixel set P T screened by S1 again, that is, only retain the pixels whose land surface temperature is between the mean plus or minus 1.5 times the standard deviation according to the normal distribution, specifically:

[0024]

[0025] ​Where κ is a preset multiple of 1.5, and Q T represents the set composed of the pixels that satisfy the condition in P T and meet the requirements in .

[0026] Furthermore, S3 includes the following steps:

[0027] S31: Sort the land surface temperature and upscaled vegetation cover of the pixel set screened in S2 in ascending order of upscaled vegetation cover. Specifically:

[0028]

[0029]

[0030] Where A represents the ascending sorting operator, represents the sequence sorted in ascending order of upscaled vegetation cover on the pixel set Q T , represents the ω-th pixel of, FVC ω represents the upscaled vegetation cover value of, FVC Ω is the set composed of the upscaled vegetation cover of the pixels in the sequence sorted in ascending order of upscaled vegetation cover for , LST ω represents the land surface temperature value of, LST Ω is the set composed of the land surface temperatures of the pixels in the sequence sorted in ascending order of upscaled vegetation cover for ;

[0031] S32: Use the sorted land surface temperature and upscaled vegetation cover to calculate the change rate and intercept of the land surface temperature with respect to vegetation cover at the thermal infrared target pixel. Specifically:

[0032] [β, d] = M(LST Ω , FVC Ω )

[0033] Where M represents the model for obtaining the change rate of land surface temperature with respect to vegetation cover, β is the change rate of land surface temperature with respect to vegetation cover, and d is the intercept of the land surface temperature with respect to vegetation cover.

[0034] Furthermore, S4 is specifically:

[0035]

[0036] Where {T Oij} 1≤i,j≤RRepresents the spatially downscaled land surface temperature with the spatial resolution of the visible light band corresponding to the land surface temperature target pixel O in the thermal infrared band.

[0037] Further, the S5 is specifically:

[0038]

[0039]

[0040] In the formula, 1 ≤ i, j ≤ R, γ Oij Represents the spatially downscaled land surface temperature at pixel (i, j) after radiation balance adjustment according to the spatially downscaled land surface temperature {T Oij}, 1≤i,j≤R the land surface temperature LST of the target pixel in the thermal infrared band, O the emissivity ε of the target pixel in the thermal infrared band, O and the emissivities {ε Oij} 1≤i,j≤R of each pixel in the visible light band corresponding to the target pixel in the thermal infrared band.

[0041] Compared with the prior art, the beneficial effects of the present invention are:

[0042] 1) Developed a refined model of land surface temperature in the thermal infrared band by combining local detail information of vegetation coverage and global change information of land surface temperature;

[0043] 2) Proposed a localization scheme for quantifying the change of land surface temperature with vegetation coverage;

[0044] 3) The radiation can maintain balance before and after the spatial downscaling of land surface temperature in the thermal infrared band. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 Is the flowchart of the method in the embodiment provided by the present invention; DETAILED DESCRIPTION OF THE EMBODIMENTS

[0046] To make the technical features, implementation schemes and beneficial effects of the present invention clearer and easier to understand, the purpose and technical features of the present invention are further introduced in detail below in combination with the embodiments and the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them, and should not be regarded as limiting the scope of the present invention.

[0047] Please refer to Figure 1 , Figure 1 which is the technical flowchart in the embodiment provided by the present invention, and includes the following steps:

[0048] S1 Obtain remote sensing images in the visible light band and remote sensing images in the thermal infrared band, remote sensing products of atmospheric column water vapor content and remote sensing products of land surface emissivity and preprocess them;

[0049] The vegetation coverage in the visible light band with S2 spatial upscaling is used to obtain the upscaled vegetation coverage with the spatial resolution of the thermal infrared band. On the upscaled vegetation coverage, a pixel set is selected where the vegetation coverage is between the minimum and maximum values of the vegetation coverage of each pixel in the visible light band corresponding to the target pixel in the thermal infrared band.

[0050] S3 Re-screen the pixel set selected in S2 based on the mean and standard deviation of the land surface temperature of the pixel set selected in S2, that is, only retain the pixels where the land surface temperature is between plus and minus 1.5 times the standard deviation of the mean according to the normal distribution.

[0051] S4 Sort the land surface temperature and the upscaled vegetation coverage of the pixel set selected in S3 in ascending order of the upscaled vegetation coverage, and calculate the change rate and intercept of the land surface temperature with respect to the vegetation coverage at the thermal infrared target pixel using the sorted land surface temperature and upscaled vegetation coverage.

[0052] S5 Calculate the spatially downscaled land surface temperature with the spatial resolution of the visible light band for the target pixel in the corresponding thermal infrared band based on the change rate and intercept obtained in S4 and the vegetation coverage of each pixel in the visible light band corresponding to the target pixel in the thermal infrared band.

[0053] S6 Adjust the spatially downscaled land surface temperature obtained in S5 according to the radiation balance so that the land surface temperature at the thermal infrared target pixel and the land surface temperature of the corresponding spatially downscaled pixel satisfy the radiation balance.

[0054] In S1, the spatial resolution of the remote sensing image in the visible light band is higher than that of the remote sensing image in the thermal infrared band. The two remote sensing images can be from the same satellite in the same period or different satellites in the same period or the same satellite in different periods or different satellites in different periods.

[0055] Exemplarily, the remote sensing image in the visible light band can be selected from the Landsat 8 OLI land imager sensor or the first to third sensors of ASTER or the HJ-1B satellite CCD camera, and the thermal infrared band image can be selected from the Landsat 8 TIRS thermal infrared sensor or the ASTER thermal infrared sensor or the HJ-1B satellite infrared camera; the remote sensing product of the atmospheric column water vapor content is selected from the MOD021KM-Level 1B product that can estimate the atmospheric moisture content; the remote sensing product of the land surface emissivity can be the ASTER GED v3 dataset, including the emissivity of 5 ASTER bands in the thermal infrared range.

[0056] In S1, the preprocessing includes the calculation of the vegetation index, the calculation of the vegetation coverage, the calculation of the atmospheric column water vapor content, the calculation of the land surface emissivity, and the inversion of the land surface temperature in the thermal infrared band.

[0057] The calculation of the vegetation index is preferably:

[0058]

[0059] In the formula, NIR and R respectively represent remote sensing images in the near-infrared band and the red band, and NDVI represents the vegetation index;

[0060] The calculation of the vegetation coverage is preferably:

[0061]

[0062] In the formula, fvc is the vegetation coverage in the visible light band, NDVI bare and NDVI veg are the NDVI values of pixels when the bare soil and the vegetation are fully covered respectively. According to previous studies, the representative value of NDVI bare is 0.3, and the representative value of NDVI veg is 0.8;

[0063] The calculation of the atmospheric column water vapor content is preferably:

[0064] w = {[0.02 - ln(ρ 19 / ρ 2 )] / 0.651} 2

[0065] In the formula, ρ 19 , ρ 2 are the apparent reflectances of the 19th and 2nd bands in the MOD021KM-Level 1B product respectively, and w is the atmospheric column water vapor content (g·cm -2 );

[0066] The calculation of the surface emissivity is preferably:

[0067] ε = fvcε veg + (1 - fvc)ε bare

[0068] ε bare = 0.682ε 13 + 0.2578ε 14 + 0.0584

[0069]

[0070] In the formula, ε 13A and ε 14A are the original emissivities of band 13 and band 14 in the ASTER GEDv3 product respectively, FVC a represents the vegetation coverage calculated using the ASTER near-infrared band and red band images, and ε is the surface emissivity in the thermal infrared band;

[0071] The retrieval of the land surface temperature in the thermal infrared band is preferably a single-channel algorithm, specifically:

[0072]

[0073]

[0074]

[0075] In the formula, LST is the land surface temperature in the thermal infrared band (K), γ(λ, T sr ) and δ(λ, T sr ) are variables related to the linear expansion of Planck's formula, λ is the equivalent wavelength in the thermal infrared band (μm), T sr is the brightness temperature in the thermal infrared band (K), ε is the land surface emissivity in the thermal infrared band, L sr is the thermal radiation brightness received by the satellite sensor at the top of the atmosphere (W·m -2 ·sr -1 ·μm -1 ), ψ 1 (λ, w), ψ 2 (λ, w), ψ 3 (λ, w) are variables dependent on wavelength and water vapor content, c 1 =1.19104×10 8 W·μm 4 ·m -2 ·sr -1 , c 2 =14388μm·K, ψ 1 (λ, w), ψ 2 (λ, w), ψ 3 (λ, w) is obtained by fitting a second-order polynomial of w and can be expressed as the product of the coefficient matrix ψ and [w 2 w 1]′, and the specific form of ψ is:

[0076]

[0077] In S2, the vegetation coverage in the visible light band at the spatial upscaling is used to obtain the upscaled vegetation coverage at the spatial resolution of the thermal infrared band. On the upscaled vegetation coverage, a pixel set is selected where the vegetation coverage is between the minimum and maximum vegetation coverages of each pixel in the visible light band corresponding to the target pixel in the thermal infrared band, specifically:

[0078] P T ={p|fvc Omin ≤FVC p ≤fvc Omax}

[0079]

[0080] FVC = Γ(fvc)

[0081] Wherein, Γ represents a spatial upscaling model, FVC represents the vegetation coverage obtained by upscaling fvc to the spatial resolution of the thermal infrared band, the subscript T represents the resolution of the thermal infrared band, the subscript O represents the target pixel in LST, R represents the ratio of the LST spatial resolution to the fvc spatial resolution, R×R represents the number of pixels corresponding to the target pixel O in fvc, fvc Oij|1≤i,j≤R represents the vegetation coverage of the pixel at the i-th row and j-th column corresponding to the target pixel O in fvc, fvc Omin represents the minimum value of the vegetation coverage of the R×R pixels corresponding to the target pixel O in fvc, fvc Omax represents the maximum value of the vegetation coverage of the R×R pixels corresponding to the target pixel O in fvc, FVC p represents the vegetation coverage of the pixel p in FVC that lies between fvc Omin and fvc Omax ; P T represents the set composed of all pixels p in FVC;

[0082] Exemplarily, the spatial upscaling model Γ can be optionally one of local averaging, nearest neighbor method, bilinear interpolation, cubic convolution, median sampling, point spread function, wavelet transform method, empirical regression, or fractal scale conversion.

[0083] In S3, the set of pixels screened according to S2 is screened again based on the mean and standard deviation of the land surface temperature of the set of pixels screened in S2, that is, only the pixels whose land surface temperature lies between plus and minus 1.5 times the standard deviation of the mean are retained according to the normal distribution. The specific steps are as follows:

[0084] S31: Calculate the mean and standard deviation of the land surface temperature in the thermal infrared band of the set of pixels P T screened in S2. Specifically:

[0085]

[0086]

[0087] Wherein, q is the pixel in the set of pixels P T , LST q is the land surface temperature in the thermal infrared band of the pixel q, FVC q is the vegetation coverage of the pixel q, is the weighted mean of the land surface temperatures in the thermal infrared band of all pixels in P T , is the weighted standard deviation of the land surface temperatures in the thermal infrared band of all pixels in P T ;

[0088] S32: According to and re-screen the pixel set P screened by S2 T again, that is, only retain the pixels whose land surface temperature is between the mean plus or minus 1.5 times the standard deviation according to the normal distribution. Specifically:

[0089]

[0090] In the formula, Q T represents the set composed of the pixels in P T that satisfy , κ is the preset multiple 1.5, and it can also be determined according to the expected number of pixels in Q T after screening. If the expected number of pixels in Q T is large, the value of κ can tend to be large; if the expected number of pixels in Q T is small, the value of κ can tend to be small.

[0091] In the above-mentioned S4, the land surface temperature and the upscaled vegetation coverage of the pixel set screened by S3 are sorted in ascending order according to the upscaled vegetation coverage, and the change rate and intercept of the land surface temperature at the thermal infrared target pixel with respect to the vegetation coverage are calculated by using the sorted land surface temperature and upscaled vegetation coverage, including the following steps:

[0092] S41: Sort the land surface temperature and the upscaled vegetation coverage of the pixel set screened by S3 in ascending order according to the upscaled vegetation coverage. Specifically:

[0093]

[0094]

[0095] In the formula, A represents the ascending sorting operator, represents the sequence sorted in ascending order according to the upscaled vegetation coverage on the pixel set Q T , represents the ω-th pixel of A(FVC QT ), FVC ω represents the value of the upscaled vegetation coverage of Ω , FVC is the set composed of the upscaled vegetation coverages of the sequence pixels sorted in ascending order according to the upscaled vegetation coverage of ω , LST represents Ω the land surface temperature value of , LST Ω is the set composed of the land surface temperatures of the sequence pixels sorted in ascending order according to the upscaled vegetation coverage of ;

[0096] Exemplarily, A can be optionally one of bubble sort, simple selection sort, direct insertion sort, binary insertion sort, or Shell sort;

[0097] S42: Calculate the change rate and intercept of the surface temperature with respect to the vegetation coverage at the thermal infrared target pixel using the sorted surface temperature and the upscaled vegetation coverage, specifically:

[0098] [β, d] = M(LST Ω , FVC Ω )

[0099] In the formula, M represents the model for obtaining the change rate of the surface temperature with respect to the vegetation coverage, β is the change rate of the surface temperature with respect to the vegetation coverage, and d is the intercept of the surface temperature with respect to the vegetation coverage change;

[0100] Exemplarily, the model M can be optionally one of least squares regression, random forest, or support vector regression.

[0101] In step S5, calculate the spatially downscaled surface temperature with the spatial resolution of the visible light band corresponding to the thermal infrared band target pixel according to the change rate and intercept obtained in S4 and the vegetation coverage of each pixel in the visible light band corresponding to the thermal infrared band target pixel, specifically:

[0102]

[0103] In the formula, {T Oij} 1≤i,j≤R represents the spatially downscaled surface temperature with the spatial resolution of the visible light band corresponding to the thermal infrared band surface temperature target pixel O.

[0104] In step S6, adjust the spatially downscaled surface temperature obtained in S5 based on radiation balance to satisfy the radiation balance between the surface temperature of the thermal infrared band target pixel and the surface temperature of the corresponding spatially downscaled pixel, specifically:

[0105]

[0106]

[0107] In the formula, 1 ≤ i, j ≤ R, γ Oij represents according to the spatially downscaled surface temperature {T Oij}, the surface temperature LST of the thermal infrared band target pixel, 1≤i,j≤R the emissivity ε of the thermal infrared band target pixel, O and the emissivity {ε O} of each pixel in the visible light band corresponding to the thermal infrared band target pixel Oij} 1≤i,j≤RThe spatially downscaled land surface temperature at pixel (i, j) after radiative balance adjustment.

[0108] A method for spatially downscaling land surface temperature of radiative balance proposed by the present invention has the following characteristics:

[0109] 1) Only the red band and near-infrared band of remote sensing images are required to enhance the spatial details of land surface temperature in the thermal infrared band, and the demand for input data is small.

[0110] 2) By using the secondary screening of pixels, the change rate of land surface temperature with the environment can be extracted only by using the remote sensing information of vegetation coverage, and the operation is simple and easy.

[0111] 3) The radiative energy before and after the spatial downscaling of land surface temperature in the thermal infrared band is basically consistent, and it has a high spatial downscaling accuracy.

[0112] 4) The proposed method can also be used for the spatial downscaling of other land surface parameters, such as soil moisture, etc. after appropriate modification.

[0113] It should be noted that the present invention may have many other implementation manners. For those of ordinary skill in the art, other embodiments can be obtained by referring to the embodiments of the present invention without creative work, but these embodiments cannot depart from the protection scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for spatially downscaling the surface temperature of radiative balance, characterized in that, it includes the following steps: S1: Spatially upscale the vegetation coverage in the visible light band to obtain the upscaled vegetation coverage with the spatial resolution of the thermal infrared band. On the upscaled vegetation coverage, select the pixel set where the vegetation coverage is between the minimum and maximum values of the vegetation coverage of each pixel in the visible light band corresponding to the target pixel in the thermal infrared band. Specifically: P T = {p | fvc Omin ≤ FVC p ≤ fvc Omax} FVC = Γ(fvc) In the formula, fvc is the vegetation coverage in the visible light band, Γ represents the spatial upscaling model, FVC represents the vegetation coverage obtained by upscaling fvc to the spatial resolution of the thermal infrared band, the subscript T represents the thermal infrared band resolution, the subscript O represents the target pixel in the land surface temperature (LST) of the thermal infrared band, R represents the ratio of the LST spatial resolution to the fvc spatial resolution, R×R represents the number of pixels corresponding to the target pixel O in fvc, and fvc Oij | 1≤i,j≤R represents the vegetation coverage of the pixel in the i-th row and j-th column corresponding to the target pixel O in fvc, and fvc Omin represents the minimum value of the vegetation coverage of the R×R pixels corresponding to the target pixel O in fvc, and fvc Omax represents the maximum value of the vegetation coverage of the R×R pixels corresponding to the target pixel O in fvc, and FVC p represents the vegetation coverage of the pixel p between fvc Omin and fvc Omax in FVC, and P T represents the set composed of all pixels p; S2: Re-select the pixel set screened in S1 according to the mean and standard deviation of the surface temperature of the pixel set screened in S1, that is, only retain the pixels where the surface temperature is between plus and minus 1.5 times the standard deviation of the mean according to the normal distribution. The specific steps are: S21: Calculate the mean and mean square deviation of the land surface temperature of the thermal infrared band of the pixel set P screened by S1, specifically: T ​ Where q is the pixel in the pixel set P T , LST q is the land surface temperature of the thermal infrared band of pixel q, and FVC q is the vegetation coverage of pixel q, is the weighted mean of the land surface temperatures of the thermal infrared bands of all pixels in P T , and is the weighted standard deviation of the land surface temperatures of the thermal infrared bands of all pixels in P T ; S22: According to and re-screen the pixel set P screened by S1 T That is, according to the normal distribution, only retain the pixels whose land surface temperature is between 1.5 times the mean plus or minus the standard deviation. Specifically: where κ is a preset multiple of 1.5, and Q T represents the set composed of the pixels that satisfy the condition in P T and meet the requirement in the set; S3: Sort the surface temperature and upscaled vegetation coverage of the pixel set screened in S2 in ascending order of the upscaled vegetation coverage, and calculate the change rate and intercept of the surface temperature with respect to the vegetation coverage at the thermal infrared target pixel by using the sorted surface temperature and upscaled vegetation coverage. It includes the following steps: S31: Sort the surface temperature and upscaled vegetation coverage of the pixel set screened in S2 in ascending order of the upscaled vegetation coverage. Specifically: In the formula, A represents the ascending sorting operator, represents the sequence sorted in ascending order according to the upscaled vegetation coverage on the pixel set Q T and represents the ω-th pixel of , and FVC ω represents the upscaled vegetation coverage value of , and FVC Ω is the set composed of the upscaled vegetation coverages of the pixels in the sequence sorted in ascending order according to the upscaled vegetation coverage of , and LST ω represents the land surface temperature value of , and LST Ω is the set composed of the land surface temperatures of the pixels in the sequence sorted in ascending order according to the upscaled vegetation coverage of ; S32: Calculate the change rate and intercept of the surface temperature with respect to the vegetation coverage at the thermal infrared target pixel by using the sorted surface temperature and upscaled vegetation coverage. Specifically: [β,d] = M(LST Ω , FVC Ω ) In the formula, M represents the model for obtaining the change rate of the surface temperature with respect to the vegetation coverage, β is the change rate of the surface temperature with respect to the vegetation coverage, and d is the intercept of the change of the surface temperature with respect to the vegetation coverage; S4: Calculate the spatially downscaled surface temperature with the spatial resolution of the visible light band corresponding to the thermal infrared target pixel according to the change rate and intercept obtained in S3 and the vegetation coverage of each pixel in the visible light band corresponding to the thermal infrared target pixel; S5: Adjust the spatially downscaled surface temperature obtained in S4 according to the radiative balance to make the surface temperature of the thermal infrared target pixel satisfy the radiative balance with the surface temperature of the corresponding spatially downscaled pixel.

2. The method according to claim 1, characterized in that, for S4, specifically: In the formula, {T Oij} 1≤i,j≤R represents the spatially downscaled land surface temperature with the spatial resolution of the visible light band corresponding to the land surface temperature target pixel O in the thermal infrared band.

3. The method according to claim 1, characterized in that, for S5, specifically: Where 1≤i,j≤R, γ Oij represents the spatially downscaled land surface temperature {T Oij} 1≤i,j≤R , the land surface temperature LST of the target pixel in the thermal infrared band O , the emissivity ε of the target pixel in the thermal infrared band O and the emissivities {ε Oij} 1≤i,j≤R of each pixel in the visible light band corresponding to the target pixel in the thermal infrared band. The spatially downscaled land surface temperature at pixel (i,j) after radiation balance adjustment.

Images