A window method for analyzing land surface temperature

Abstract

The application provides a window method for analyzing land surface temperature, comprising: A1 determining the interval of high spatial resolution vegetation coverage in a window area centered on a target pixel; A2 constructing a low spatial resolution vegetation coverage pixel set belonging to the interval of A1, and establishing a window with a data pair of remote sensing land surface temperature and low spatial resolution vegetation coverage of the pixel set; A3 dividing the window of A2 into several sub-windows according to the horizontal axis, and constructing a normal distribution model of land surface temperature and its variation interval in each sub-window; A4 determining the derivative of land surface temperature with respect to vegetation coverage at the target pixel according to the variation interval of land surface temperature in each sub-window of A3; A5 resolving the remote sensing land surface temperature of the target pixel according to the derivative of A4; and A6 spatially downscaling the remote sensing land surface temperature of the target pixel according to the derivative of A4. The application avoids the error caused by using all pixels to calculate the derivative of land surface temperature with respect to vegetation coverage in the feature space method, and realizes the integration of remote sensing land surface temperature decomposition and enhancement.

Description

TECHNICAL FIELD

[0001] The present application relates to a remote sensing land surface temperature change analysis method, and in particular to a remote sensing land surface temperature analysis method considering a local window. BACKGROUND

[0002] Land surface temperature is the most important and most variable land state variable, is the comprehensive effect of land surface radiation energy budget and land energy distribution, is the driving force of land-atmosphere energy exchange, is an important reason for global climate change and water cycle, is the basis for the survival and reproduction of earth's organisms, and plays a crucial role in many fields such as ecological civilization construction and realization of the United Nations development goals.

[0003] Satellite thermal infrared remote sensing is currently the main means of monitoring regional or global land surface temperature. Satellite thermal infrared remote sensing measures land surface long-wave radiation energy to retrieve land surface temperature. Because land surface long-wave radiation has a relatively low energy density, satellite remote sensing thermal infrared band sensors must have a wider instantaneous field of view than visible light band sensors on the same satellite to collect enough land surface long-wave radiation energy to measure land surface temperature. Because satellite remote sensing thermal infrared band sensors have a wider instantaneous field of view, satellite remote sensing thermal infrared band images usually have lower spatial resolution, which limits the application of remote sensing land surface temperature in the fields of ecological protection, environmental remediation and thermal radiation. Therefore, to improve the utilization rate of remote sensing land surface temperature, it is necessary to analyze the land surface temperature.

[0004] Remote sensing land surface temperature contains two important types of information: 1) the rate of change of land surface temperature with the external environment; and 2) the composition of land surface temperature, which is necessary for remote sensing land surface temperature spatial downscaling and decomposition. Current remote sensing land surface temperature downscaling analysis methods mainly include statistical and spectral mixing methods: the basic idea of the statistical method is to apply the relationship between land surface temperature and factors at low spatial resolution to high spatial resolution factors to achieve remote sensing land surface temperature downscaling. This method has a clear principle, but ignores the scale of the relationship between land surface temperature and factors. The spectral mixing method calculates remote sensing land surface temperature according to the visible light band spectrum, without the need to establish a correlation model between land surface temperature and environmental factors, but cannot identify the difference in spectral composition between the thermal infrared band and the visible light band. Remote sensing land surface temperature composition analysis methods include mixed pixel unmixing and feature space method: mixed pixel unmixing decomposes remote sensing land surface temperature with the aid of multispectral images. This method follows the laws of radiation physics, but the required remote sensing information at different angles is currently insufficient. In the feature space method, there is still a lot of uncertainty in determining the four extreme feature points, which largely depends on people's guesses.

[0005] The key to analyzing the above two types of information of remote sensing land surface temperature lies in accurately obtaining the variation law of land surface temperature with environmental factors. In view of the heterogeneity of the spatial distribution of land surface temperature and environmental factors, the component information of remote sensing land surface temperature pixel revealed by the high spatial resolution band of remote sensing image is used, and the variation interval of land surface temperature is estimated by constructing a normal distribution model of land surface temperature under similar environmental conditions. In order to extract the variation information of remote sensing land surface temperature, a window method for analyzing land surface temperature is invented. SUMMARY

[0006] In order to solve the above technical problems, the window method for analyzing land surface temperature is provided, and the technical scheme comprises the following steps:

[0007] A1: determining the interval of high spatial resolution vegetation coverage in the window area centered on the target pixel;

[0008] A2: constructing a low spatial resolution vegetation coverage pixel set belonging to the interval of A1, and establishing a window with vegetation coverage as the horizontal axis, land surface temperature as the vertical axis, and the data pair of remote sensing land surface temperature and low spatial resolution vegetation coverage of the pixel set;

[0009] A3: dividing the A2 window into several sub-windows according to the horizontal axis, and constructing a normal distribution model of land surface temperature and its variation interval in each sub-window;

[0010] A4: determining the derivative of land surface temperature with vegetation coverage at the target pixel according to the variation interval of land surface temperature in each sub-window of A3;

[0011] A5: unmixing the remote sensing land surface temperature of the target pixel according to the derivative of A4;

[0012] A6: unmixing the remote sensing land surface temperature of the target pixel according to the derivative of A4 and the spatial downscaling of high spatial resolution vegetation coverage;

[0013] A7: cyclically executing A1-A5 to unmix the remote sensing land surface temperature of all pixels;

[0014] A8: cyclically executing A1-A4 and A6 to spatially downscale the remote sensing land surface temperature of all pixels.

[0015] As a further technical scheme, the A1 is specifically:

[0016] [F O_min ,F O_max ]={a∈FVC·M O_W :F O_min ≤a≤F O_max}

[0017] In the formula, FVC represents high spatial resolution vegetation coverage, and subscript O represents the target pixel in remote sensing land surface temperature Tm where subscript W denotes the spatial position of T m the length of the pre-set rectangular window at spatial resolution M O_W denotes a mask matrix with 1 for the matrix elements within the rectangular window centered at O with length W and 0 for the rest of the matrix elements, · denotes matrix dot product for corresponding matrix elements, F O_min denotes FVC·M O_W denotes the left end point of the vegetation fraction interval in F O_max denotes FVC·M O_W denotes the right end point of the vegetation fraction interval in a O_min denotes any vegetation fraction in the interval [F O_max ]

[0018] As a further technical solution, the A2 comprises the following steps:

[0019] A21: the spatial resolution of the low spatial resolution vegetation fraction should be equal to the spatial resolution of the remotely sensed land surface temperature, specifically:

[0020]

[0021] where fvc denotes the low spatial resolution vegetation fraction equal to the spatial resolution of the remotely sensed land surface temperature, (x, y) denotes the pixel position, R denotes the ratio of the spatial resolution of the remotely sensed land surface temperature to the spatial resolution of the high spatial resolution vegetation fraction, and Γ denotes a resampling matrix operator with length R;

[0022] A22: constructing a low spatial resolution vegetation fraction pixel set belonging to the membership A1 interval, specifically:

[0023] Ω = {(X, Y) | F O_min ≤ fvc(X, Y) ≤ F O_max}

[0024] where (X, Y) denotes the pixel position of the vegetation fraction belonging to the interval [F O_min , F O_max ] in fvc, and Ω denotes the set of all (X, Y) positions;

[0025] A23: establishing a window with the vegetation fraction as the horizontal axis, the land surface temperature as the vertical axis, and the remotely sensed land surface temperature and low spatial resolution vegetation fraction of the pixel set as the data pair, specifically:

[0026]

[0027] where denotes the position element in any given Ω T mdenotes the remotely sensed land surface temperature, subscript m denotes that the remotely sensed land surface temperature is a mixed temperature, τ denotes the land surface temperature at a location in the fvc, u denotes the vegetation fraction at a location in the fvc, S denotes a window consisting of data pairs (u, τ) of land surface temperature τ and vegetation fraction u at any location in Ω. m

[0028] As a further technical solution, the A3 comprises the following steps:

[0029] A31: dividing the window of A2 into a plurality of sub-windows according to the horizontal axis, specifically:

[0030] S = S1∪S2∪…∪Sn n ∪…∪Sn N-1 ∪Sn N

[0031]

[0032]

[0033] In the formula, N denotes the number of sub-windows into which S is divided, denotes the pixel position in the remotely sensed land surface temperature, s.t. denotes such that holds, S n denotes the nth sub-window of S, consisting of data pairs O_min and land surface temperature O_min belonging to the closed interval [F +(n-1)*binυ, F +n*binυ] of the vegetation fraction; {S n} 1≤n≤N denotes the N sub-windows divided by S;

[0034] A32: constructing a normal distribution model of land surface temperature in each sub-window, specifically:

[0035]

[0036]

[0037]

[0038]

[0039]

[0040] ​​​​

[0041] In the formula, the subscript n represents S n The sequence number, 1≤n≤N, K n S represents n The number of data pairs in the middle, Represents any given S n Data pairs in the middle, This indicates the vegetation cover in the data pair. This represents the surface temperature in the data pair. S represents n The minimum value of surface temperature in all data pairs. S represents n The maximum value of the surface temperature in all data points. Indicates that S n The k-th data point in the middle is the surface temperature. The transformation value that follows a normal distribution after standardization and arcsine transformation by the square root. express The mean, express standard deviation Indicates the transformation variable The normal distribution it follows, i.e.

[0042] A33: Construct the range of surface temperature variation within each sub-window, specifically:

[0043]

[0044] In the formula, z n_l Indicates the transformation variable The left boundary of the value range corresponds to S. n The lower limit of surface temperature variation, z n_r Indicates the transformation variable The right boundary of the value range corresponds to S. n The upper limit of surface temperature variation in China can be determined by the correspondence between the mean, standard deviation and probability of a normal distribution. The mean plus or minus three times the standard deviation can cover 99.7% of the values ​​of the variable.

[0045] As a further technical solution, A4 includes the following steps:

[0046] A41: According to {z} in A3 n_r} 1≤n≤N The equation for determining the upper limit of the change in the transformed variable is as follows:

[0047]

[0048]

[0049] In the formula, β r b represents the slope of the upper bound equation for the change of the variable with vegetation cover. r The intercept of the upper bound equation representing the change of the variable with vegetation cover;

[0050] A42: According to {z} in A3 n_l} 1≤n≤N The equation for determining the lower limit of the change in the transformation variable is as follows:

[0051]

[0052]

[0053] In the formula, β l b represents the slope of the equation representing the lower bound of the variable as a function of vegetation cover. l The intercept of the equation representing the lower limit of the variable as a function of vegetation cover;

[0054] A43: Determine the sub-window where the target pixel is located, specifically:

[0055]

[0056] In the formula, fvc o This represents the vegetation cover of the target pixel, which is equal to the spatial resolution of remotely sensed land surface temperature. Indicates the sub-window number where the target pixel is located;

[0057] A44: Determine the variable that controls the surface temperature of the target pixel, specifically:

[0058]

[0059] In the formula, express The minimum value of surface temperature in all data pairs. express The maximum value of the surface temperature in all data pairs, T m_o This represents the target pixel's remote sensing surface temperature. T represents m_o The variable being transformed;

[0060] A45: Determine the target pixel fvc o The corresponding upper limit value of the transformation variable is as follows:

[0061]

[0062] In the formula, Indicates target cell fvc o The corresponding upper limit value of the transformation variable;

[0063] A46: determining the target pixel fvc o The corresponding transition variable lower limit value is specifically:

[0064]

[0065] In the formula, The target pixel fvc o The corresponding transition variable lower limit value;

[0066] A47: determining the derivative of the transition variable with respect to the vegetation coverage at the target pixel, specifically:

[0067]

[0068] In the formula, β m The derivative of the transition variable with respect to the vegetation coverage at the target pixel;

[0069] A48: determining the intercept of the transition variable with respect to the vegetation coverage at the target pixel, specifically:

[0070]

[0071] In the formula, b m The intercept of the transition variable with respect to the vegetation coverage at the target pixel;

[0072] A49: determining the derivative of the surface temperature with respect to the vegetation coverage at the target pixel, specifically:

[0073]

[0074] In the formula, The derivative of the surface temperature with respect to the vegetation coverage at the target pixel.

[0075] As a further technical solution, the A5 is specifically:

[0076]

[0077]

[0078] In the formula, T v_o The vegetation temperature in the target pixel surface temperature, T s_o The soil temperature in the target pixel surface temperature, ε m The target pixel surface emissivity, ε v The vegetation emissivity in the target pixel, ε s The soil emissivity in the target pixel.

[0079] As a further technical solution, the A6 is specifically:

[0080]

[0081]

[0082]

[0083] In the formula, represents high spatial resolution vegetation coverage corresponding to the spatial range of the target pixel remote sensing ground temperature, represents the high spatial resolution vegetation coverage obtained by using the generated high spatial resolution ground temperature.

[0084] As a further technical solution, the A7 is specifically:

[0085]

[0086]

[0087] In the formula, T v represents the vegetation temperature obtained by cyclically performing A1-A5 to demix all pixel remote sensing ground temperatures, s represents the soil temperature obtained by cyclically performing A1-A5 to demix all pixel remote sensing ground temperatures.

[0088] As a further technical solution, the A8 is specifically:

[0089]

[0090] In the formula, represents the high spatial resolution ground temperature obtained by cyclically performing A1-A4 and A6 to spatially downscale all pixel remote sensing ground temperatures.

[0091] Compared with the prior art, the present application has the beneficial effects that:

[0092] 1) a window framework for expressing the localized relationship between high and low spatial resolution vegetation coverage and remote sensing ground temperature is proposed;

[0093] 2) the derivative of the ground temperature with respect to the vegetation coverage is obtained by using the normal distribution of the ground temperature conversion variable;

[0094] 3) the integration of remote sensing ground temperature component decomposition and spatial enhancement is realized. BRIEF DESCRIPTION OF DRAWINGS

[0095] Figure 1 The technical flowchart for illustrating the embodiments of the present application is shown; DETAILED DESCRIPTION

[0096] The following will be combined with the drawings Figure 1Further illustrate the basic idea, implementation steps of the present application, it is necessary to note that the embodiments shown are only a part of the embodiments of the present application, but should not be as a limitation on the scope of the present application.

[0097] Please refer to Figure 1 , Figure 1 The technical flow chart of the embodiments of the present application, including the following steps:

[0098] A0: obtain the surface emissivity, surface temperature, vegetation coverage and other remote sensing products and pretreatment of the study area;

[0099] A1: determine the high spatial resolution vegetation coverage interval in the window area centered on the target pixel;

[0100] A2: construct the low spatial resolution vegetation coverage pixel set of the membership A1 interval, and establish the window with vegetation coverage as the horizontal axis, surface temperature as the vertical axis, and the remote sensing surface temperature and low spatial resolution vegetation coverage of the pixel set as the data pair;

[0101] A3: divide the A2 window into several sub-windows according to the horizontal axis, and construct the normal distribution model of the surface temperature and its variation interval in each sub-window;

[0102] A4: determine the derivative of the surface temperature with respect to the vegetation coverage at the target pixel according to the variation interval of the surface temperature in each sub-window of A3;

[0103] A5: according to the derivative of A4, the remote sensing surface temperature of the target pixel is solved;

[0104] A6: according to the derivative of A4 and the high spatial resolution vegetation coverage, the remote sensing surface temperature of the target pixel is solved;

[0105] A7: cycle A1-A5 to solve the remote sensing surface temperature of all pixels;

[0106] A8: cycle A1-A4 and A6 to spatially downscale the remote sensing surface temperature of all pixels.

[0107] In one embodiment, in the A0, the surface emissivity remote sensing product of the study area can be downloaded from http: / / modis.gsfc.nasa.gov MOD11A1v061, which covers the whole world, 1km resolution, daily time resolution, stored in 1200km x 1200km slices, the original file is the hdf format of sinusoidal projection, which can be resampled, projected and cut to be consistent with other remote sensing products, and the wide band surface emissivity is obtained by linear fitting, specifically:

[0108] ε=0.261+0.314ε 31 +0.411ε32

[0109] wherein ε 31 , ε 32 are the specific radiance of MODIS 31 band and MODIS 32 band respectively, and ε is the specific radiance of the wide band of the ground surface;

[0110] In the A0, the remote sensing product of the land surface temperature of the study area can be obtained by inverting the thermal infrared band of the remote sensing image. For example, Landsat 8 TIRS band 10 image, single-channel land surface temperature inversion algorithm;

[0111] In the A0, the remote sensing product of the land surface temperature of the study area can also be directly downloaded from the website. For example, ECO2LSTEv001 can be downloaded from https: / / lpdaac.usgs.gov / products / eco2lstev001 / . This product can cover N50°-S50°, with a spatial resolution of 70 meters, a daily time resolution, a time span of 2018-07-09 to the present, a data format of hdf5, and can be resampled, projected and converted, and cut into other remote sensing products by professional tools such as ENVI and ARCGIS;

[0112] In the A0, the remote sensing product of the vegetation coverage of the study area can be calculated by the selected remote sensing image, specifically:

[0113]

[0114] wherein NDVI represents the remote sensing vegetation index of the study area, NDVI s represents the NDVI value of the bare soil or non-vegetation coverage area of the study area, NDVI v represents the NDVI value of the completely vegetation coverage pixel of the study area, and according to the previous research results, NDVI s is selected as 0.25, and NDVI v is selected as 0.85;

[0115] In the A0, the preprocessing is to obtain the pixel component land surface specific radiance of the study area, specifically:

[0116]

[0117]

[0118]

[0119] wherein, represents the number of pixels of the remote sensing land surface temperature of the study area, and fvc represents the vegetation coverage equal to the spatial resolution of the remote sensing land surface temperature, which can be obtained by spatial downscaling of FVC, and εs , ε v respectively represent the ratio of the radiation of the soil and vegetation in the pixel group, represent the average value of the surface radiation ratio, represent the average value of the vegetation coverage.

[0120] In one embodiment, the A1, specifically:

[0121] [F O_min ,F O_max ]={a∈FVC·M O_W :F O_min ≤a≤F O_max}

[0122] In the formula, FVC represents the high spatial resolution vegetation coverage, the subscript O represents the spatial position of the target pixel in the remote sensing surface temperature T m , the subscript W represents the length of the preset rectangular window under the spatial resolution of T m , M O_W represents a mask matrix with matrix elements of 1 and other matrix elements of 0 in the rectangular window centered at O and with a length of W, · represents matrix point multiplication of corresponding matrix elements, F O_min represents the left end point of the vegetation coverage interval in FVC·M O_W , F O_max represents the right end point of the vegetation coverage interval in FVC·M O_W , and a represents any vegetation coverage in the interval [F O_min ,F O_max ].

[0123] In one embodiment, the A2, comprising the following steps:

[0124] A21: The spatial resolution of the low spatial resolution vegetation coverage should be equal to the spatial resolution of the remote sensing surface temperature, specifically:

[0125]

[0126] In the formula, fvc represents the low spatial resolution vegetation coverage equal to the spatial resolution of the remote sensing surface temperature, (x, y) represents the pixel position, R represents the ratio of the spatial resolution of the remote sensing surface temperature to the spatial resolution of the high spatial resolution vegetation coverage, and Γ represents a resampling matrix operator with a length of R;

[0127] A22: Construct a low spatial resolution vegetation coverage pixel set belonging to the interval of A1, specifically:

[0128] Ω={(X,Y)|F O_min ≤fvc(X,Y)≤F O_max}

[0129] where (X, Y) represents the pixel position in fvc where the vegetation cover belongs to [F O_min ,F O_max ], and Ω represents the set of all (X, Y) positions;

[0130] A23: Establish a window with vegetation cover as the horizontal axis, surface temperature as the vertical axis, and the data pairs of the remote sensing surface temperature and low spatial resolution vegetation cover in the pixel set, specifically:

[0131]

[0132] wherein, represents the position element in Ω T m represents the remote sensing surface temperature, and subscript m represents that the remote sensing surface temperature is a mixed temperature, τ represents the surface temperature at T m , υ represents the vegetation cover at in fvc, and S represents the window composed of the data pairs (υ, τ) of the surface temperature τ and the vegetation cover υ at any position in Ω. In one embodiment, the A3 comprises the following steps:

[0133] A31: Divide the A2 window into a plurality of sub-windows according to the horizontal axis, specifically:

[0134] A31: Divide the A2 window into a plurality of sub-windows according to the horizontal axis, specifically:

[0135] S = S1∪S2∪…∪S n ∪…∪S N-1 ∪S N

[0136]

[0137]

[0138] wherein N represents the number of sub-windows into which S is divided, represents the pixel position in the remote sensing surface temperature, and s.t. represents such that is established, and S n represents the nth sub-window of S, which is composed of the data pairs of the vegetation cover belonging to the closed interval [F O_min +(n-1)*binυ, F O_min +n*binυ] and the surface temperature n 1≤n≤N ​​​​​This represents the N sub-windows divided by S;

[0139] A32: Construct a normal distribution model of land surface temperature within each sub-window, specifically as follows:

[0140]

[0141]

[0142]

[0143]

[0144]

[0145]

[0146] In the formula, the subscript n represents S n The sequence number, 1≤n≤N, K n S represents n The number of data pairs in the middle, Represents any given S n Data pairs in the middle, This indicates the vegetation cover in the data pair. This represents the surface temperature in the data pair. S represents n The minimum value of surface temperature in all data pairs. S represents n The maximum value of the surface temperature in all data points. Indicates that S n The k-th data point in the middle is the surface temperature. The transformation value that follows a normal distribution after standardization and arcsine transformation by the square root. express The mean, express standard deviation Indicates the transformation variable The normal distribution it follows, i.e.

[0147] A33: Construct the range of surface temperature variation within each sub-window, specifically:

[0148]

[0149] In the formula, z n_l Indicates the transformation variable The left boundary of the value range corresponds to S. n The lower limit of surface temperature variation, z n_r Indicates the transformation variable the upper boundary of the value of the transition variable, corresponding to S n The upper limit of the change of the land surface temperature is determined according to the correspondence relationship among the mean value, the standard deviation and the probability of the normal distribution. The mean value plus or minus 3 times the standard deviation can cover 99.7% of the values of the transition variable.

[0150] In one embodiment, the A4 comprises the following steps:

[0151] A41: determining the transition variable according to the equation in A3, that is, {z n_r} 1≤n≤N determining the upper limit equation of the change of the transition variable, specifically:

[0152]

[0153]

[0154] wherein, β r represents the slope of the upper limit equation of the change of the transition variable with the vegetation coverage, b r represents the intercept of the upper limit equation of the change of the transition variable with the vegetation coverage;

[0155] A42: determining the transition variable according to the equation in A3, that is, {z n_l} 1≤n≤N determining the lower limit equation of the change of the transition variable, specifically:

[0156]

[0157]

[0158] wherein, β l represents the slope of the lower limit equation of the change of the transition variable with the vegetation coverage, b l represents the intercept of the lower limit equation of the change of the transition variable with the vegetation coverage;

[0159] A43: determining the sub-view window in which the target pixel is located, specifically:

[0160]

[0161] wherein, fvc o represents the vegetation coverage of the target pixel equal to the spatial resolution of the remote sensing land surface temperature, represents the serial number of the sub-view window in which the target pixel is located;

[0162] A44: determining the transition variable of the land surface temperature of the target pixel, specifically:

[0163]

[0164] wherein, represents the minimum value of the land surface temperature in all data pairs, represents the maximum value of the ground temperature in all data pairs, T m_o represents the remote sensing ground temperature of the target pixel, represents T m_o the transition variable of the target pixel;

[0165] A45: determining the upper limit value of the transition variable corresponding to the target pixel fvc o corresponding to the target pixel fvc

[0166]

[0167] In the formula, corresponding to the target pixel fvc o corresponding to the target pixel fvc

[0168] A46: determining the lower limit value of the transition variable corresponding to the target pixel fvc o corresponding to the target pixel fvc

[0169]

[0170] In the formula, corresponding to the target pixel fvc o corresponding to the target pixel fvc

[0171] A47: determining the derivative of the transition variable with respect to the vegetation coverage at the target pixel, specifically:

[0172]

[0173] In the formula, β m corresponding to the target pixel fvc

[0174] A48: determining the intercept of the transition variable with respect to the vegetation coverage at the target pixel, specifically:

[0175]

[0176] In the formula, b m corresponding to the target pixel fvc

[0177] A49: determining the derivative of the ground temperature with respect to the vegetation coverage at the target pixel, specifically:

[0178]

[0179] In the formula, corresponding to the target pixel fvc

[0180] In one embodiment, the A5, specifically:

[0181]

[0182]

[0183] wherein T v_o denotes the vegetation temperature in the target pixel land surface temperature, T s_o denotes the soil temperature in the target pixel land surface temperature, ε m denotes the target pixel land surface emissivity, ε v denotes the vegetation emissivity in the target pixel, ε s denotes the soil emissivity in the target pixel.

[0184] In one embodiment, the A6, specifically:

[0185]

[0186]

[0187]

[0188] wherein T denotes the high spatial resolution vegetation coverage corresponding to the target pixel remote sensing land surface temperature spatial range, denotes the high spatial resolution land surface temperature generated by

[0189] In one embodiment, the A7, specifically:

[0190]

[0191]

[0192] wherein T v denotes the vegetation temperature obtained by cyclically executing A1-A5 to unmix all pixel remote sensing land surface temperatures, T s denotes the soil temperature obtained by cyclically executing A1-A5 to unmix all pixel remote sensing land surface temperatures.

[0193] In one embodiment, the A8, specifically:

[0194]

[0195] wherein T denotes the high spatial resolution land surface temperature obtained by cyclically executing A1-A4 and A6 to spatially downscale all pixel remote sensing land surface temperatures.

[0196] The window method for analyzing land surface temperature provided by the present application has the following characteristics: ​

[0197] 1) The target pixel set similar to the vegetation coverage of the target pixel is screened by the window, thus avoiding the error caused by the feature space method using all pixels to calculate the variation rate of the land surface temperature with the vegetation coverage;

[0198] 2) The mathematical relationship between the land surface temperature and the vegetation coverage is quantified by means of the transition variable of the land surface temperature, thus better meeting the requirement of the mathematical operation on the probability distribution of the input data;

[0199] 3) The present application realizes the integration of the remote sensing land surface temperature component decomposition and the spatial downscaling;

[0200] 4) The present application requires a single type of parameter, does not need the auxiliary data other than the remote sensing product, and is easy to program and realize.

[0201] It should be reminded that other embodiments can be obtained by modifying or replacing some steps of the present embodiment without the creative labor of the ordinary skilled in the art, but these embodiments cannot deviate from the protection scope of the technical solutions of the embodiments of the present application.

Claims

1. A window method of resolving land surface temperature, characterized by, It comprises the following steps: A1: determining the interval of high spatial resolution vegetation coverage in the window area centered on the target pixel; A2: constructing the low spatial resolution vegetation coverage pixel set belonging to the interval of A1, and establishing the window with the vegetation coverage as the horizontal axis, the surface temperature as the vertical axis, and the remote sensing surface temperature and the low spatial resolution vegetation coverage of the pixel set as the data pair; A3: dividing the A2 window into a plurality of sub-windows according to the horizontal axis, and constructing the normal distribution model of the surface temperature and its change interval in each sub-window; A4: determining the derivative of the surface temperature with respect to the vegetation coverage at the target pixel according to the change interval of the surface temperature in each sub-window of A3; A5: solving the remote sensing surface temperature of the target pixel according to the derivative of A4; A6: solving the remote sensing surface temperature of the target pixel according to the derivative of A4 and the spatial downscaling of the high spatial resolution vegetation coverage; A7: cyclically executing A1-A5 to solve the remote sensing surface temperature of all pixels; A8: cyclically executing A1-A4 and A6 to spatially downscale the remote sensing surface temperature of all pixels.

2. The method of claim 1, wherein, The A1 is specifically: [F O_min ,F O_max ] = { a e FVC·M O_W :F O_min ≤ a ≤ F O_max} In the formula, FVC represents high spatial resolution vegetation coverage, subscript O represents the spatial position of the target pixel in remote sensing land surface temperature T m m , subscript W represents the length of the preset rectangular window under the spatial resolution of T m , M O_W represents the length of the rectangular window with O as the center and W as the length, and the matrix elements in the rectangular window are 1 and the rest are 0, · represents the matrix point multiplication of the corresponding matrix elements, F O_min represents FVC·M O_W , F O_max represents the left end point of the vegetation coverage interval in FVC·M O_W , and a represents any vegetation coverage in the interval [F O_min , F O_max ].

3. The method of claim 1, wherein, The A2 comprises the following steps: A21: the spatial resolution of the low spatial resolution vegetation coverage should be equal to the spatial resolution of the remote sensing surface temperature, specifically: In the formula, fvc represents the low spatial resolution vegetation coverage equal to the spatial resolution of the remote sensing surface temperature, (x, y) represents the pixel position, R represents the ratio of the spatial resolution of the remote sensing surface temperature to the spatial resolution of the high spatial resolution vegetation coverage, and Γ represents the resampling matrix operator with a length of R; A22: constructing the low spatial resolution vegetation coverage pixel set belonging to the interval of A1, specifically: Ω = {(X, Y) | F O_min ≤ fvc(X, Y) ≤ F O_max} In the formula, (X, Y) represents the pixel position of the vegetation coverage in the fvc belonging to [F O_min ,F O_max ] and Ω represents a set composed of all (X, Y) positions. A23: establishing the window with the vegetation coverage as the horizontal axis, the surface temperature as the vertical axis, and the remote sensing surface temperature and the low spatial resolution vegetation coverage of the pixel set as the data pair, specifically: wherein represents a position element in Ω T m represents a remotely sensed surface temperature, the subscript m indicates that the remotely sensed surface temperature is a mixed temperature, τ represents the surface temperature at m in Ω represents the fraction of vegetation cover at represents a window of data pairs (υ, τ) consisting of the surface temperature τ and the fraction of vegetation cover υ at any position in Ω 4. The method of claim 1, wherein, The A3 comprises the following steps: A31: dividing the A2 window into a plurality of sub-windows according to the horizontal axis, specifically: S = S1 U S2 U... U S n U... U S N-1 U S N where N denotes the number of sub-windows into which S is divided, denotes the pixel position in the remotely sensed land surface temperature, s.t. denotes such that holds, S n denotes the n-th sub-window of S, by which the vegetation fraction belongs to the closed interval [F O_min +(n-1)*binυ, F O_min +n*binυ], and the land surface temperature constitute the data pair composed of {S n} 1≤n≤N denotes the N sub-windows into which S is divided; A32: constructing the normal distribution model of the surface temperature in each sub-window, specifically: where subscript n denotes the serial number of S n , 1≤n≤N, K n denotes the number of data pairs in S n , denotes an arbitrary given data pair in S n , denotes the vegetation coverage in the data pair, denotes the land surface temperature in the data pair, denotes the minimum value of land surface temperature in all data pairs in S n , denotes the maximum value of land surface temperature in all data pairs in S n , denotes the transformed value of land surface temperature in the kth data pair in S n after normalization and square root inverse sine conversion, which is subject to normal distribution, denotes the mean value of , denotes the standard deviation of , denotes the normal distribution to which the transformed variable is subject, i.e. ​ A33: constructing the change interval of the surface temperature in each sub-window, specifically: In the formula, z n_l represents the lower limit of the value of the transition variable , corresponding to the lower limit of the change in the surface temperature in S n , z n_r represents the upper limit of the value of the transition variable , corresponding to the upper limit of the change in the surface temperature in S n , according to the correspondence between the mean, standard deviation and probability of the normal distribution, the mean plus or minus 3 times the standard deviation can cover 99.7% of the values of the transition variable.

5. The method of claim 1, wherein, The A4 comprises the following steps: A41: The method of A3 wherein the transition variable is a function of the following: {z n_r} 1≤n≤N An upper limit equation for determining a change in the transition variable is determined, specifically: where β r represents the slope of the upper limit equation for the transition variable as a function of vegetation cover, b r represents the intercept of the upper limit equation for the transition variable as a function of vegetation cover; A42: The method of A3 wherein the transition amount is determined by the equation: z = (a + b * x) / (c + d * x) where a, b, c, and d are constants. n_l} 1≤n≤N The lower limit equation of the transition amount change is determined, specifically: where β l represents the slope of the lower bound equation for the transition of the vegetation cover fraction, b l represents the intercept of the lower bound equation for the transition of the vegetation cover fraction; A43: determining the sub-window where the target pixel is located, specifically: In the formula, fvc o represents the vegetation coverage degree of the target pixel equal to the spatial resolution of the remote sensing ground temperature, represents the sub-window serial number where the target pixel is located; A44: determining the transition variable of the surface temperature of the target pixel, specifically: wherein represents the minimum value of the surface temperature in all data pairs, represents the maximum value of the surface temperature in all data pairs, T m_o represents the target pixel remote sensing surface temperature, represents T m_o the transition variable; A45: determining a target pixel fvc o The corresponding transition upper limit value is specifically: In the formula, represents the target pixel fvc o corresponding to the upper limit value of the transition variable; A46: determining a target pixel fvc o The corresponding transition quantity lower limit value is specifically: In the formula, represents the target pixel fvc o corresponding to the lower limit value of the transition variable A47: determining the derivative of the transition variable with respect to the vegetation coverage at the target pixel, specifically: where β m denotes the derivative of the equation of the transition quantity as a function of the vegetation cover at the target pixel; A48: determining the intercept of the transition variable with respect to the vegetation coverage at the target pixel, specifically: where b m represents the intercept of the equation of the transition variable as a function of the vegetation cover at the target pixel; A49: determining the derivative of the surface temperature with respect to the vegetation coverage at the target pixel, specifically: wherein represents the derivative of the surface temperature at the target pixel with respect to the vegetation fraction.

6. The method of claim 1, wherein, The A5 is specifically: where T v_o represents vegetation temperature in the target pixel land surface temperature, T s_o represents soil temperature in the target pixel land surface temperature, ε m represents the target pixel land surface emissivity, ε v represents vegetation emissivity in the target pixel, ε s represents soil emissivity in the target pixel.

7. The method of claim 1, wherein, The A6 is specifically: In the formula, represents high spatial resolution vegetation coverage corresponding to the spatial range of the remote sensing surface temperature of the target pixel, represents high spatial resolution vegetation coverage corresponding to the spatial range of the remote sensing surface temperature of the target pixel, generated high spatial resolution surface temperature.

8. The method of claim 1, wherein, The A7 is specifically: In the formula, T v represents the vegetation temperature obtained by unmixing all pixel remote sensing land surface temperatures by executing A1-A5 in a loop. s represents the soil temperature obtained by unmixing all pixel remote sensing land surface temperatures by executing A1-A5 in a loop.

9. The method of claim 1, wherein, The A8 is specifically: In the formula, represents the high spatial resolution land surface temperature obtained by cyclically performing A1-A4 and A6 spatial downscaling of all pixels of the remote sensing land surface temperature.

Images