CN117540530A

CN117540530A - Urban earth surface temperature downscaling method and device based on high-resolution satellite images

Info

Publication number: CN117540530A
Application number: CN202311308737.5A
Authority: CN
Inventors: 周会珍; 李冬冬; 张婷; 汤伟干; 文强; 朱菊蕊; 陈晨
Original assignee: Twenty First Century Aerospace Technology Co ltd
Current assignee: Twenty First Century Aerospace Technology Co ltd
Priority date: 2023-10-10
Filing date: 2023-10-10
Publication date: 2024-02-09
Anticipated expiration: 2043-10-10
Also published as: CN117540530B

Abstract

The invention discloses a city ground surface temperature scaling method and device based on high-resolution satellite images, relates to the technical field of remote sensing image data processing, and mainly aims to solve the problem that the ground surface temperature of a water body and a shadow area is abnormal in a city ground surface temperature scaling result and improve applicability. The main technical scheme of the invention is as follows: processing multi-source remote sensing data by utilizing a preprocessing rule, and inverting by utilizing a splitting window algorithm to obtain surface temperature data with medium-low spatial resolution; performing automatic classification of the two types of ground objects of the high-spatial resolution multispectral remote sensing image by using a maximum inter-class variance method, and calculating different spatial resolution downscaling parameter factors of water body and shadow ground object type areas and non-water body ground object type areas; based on the different spatial resolution downscaling parameter factors, a corresponding surface downscaling model is built by utilizing the sample points of the target area and used for calculating high spatial resolution surface temperature data of the target area. The method is used for satellite image processing.

Description

Urban earth surface temperature downscaling method and device based on high-resolution satellite images

Technical Field

The invention relates to the technical field of remote sensing image data processing, in particular to a city ground surface temperature downscaling method and device based on high-resolution satellite images.

Background

Urban thermal environments directly affect people's quality of life and are closely related to urban climates. The earth surface temperature LST (Land Surface Temperature) is one of key variables of urban thermal environment research, and the satellite remote sensing technology has the characteristics of wide monitoring range, strong timeliness and the like, can acquire large-scale earth surface temperature data based on the surface scale level, and provides powerful data support for thermal environment research. Because thermal infrared image data in the global area is limited by the performance of the sensor, the problem of lower spatial resolution exists, and therefore, the space detail information of the ground surface temperature of the urban area with extremely complex underlying surface and large space heterogeneity cannot be monitored. Therefore, the ground surface temperature data with high spatial resolution are acquired by adopting a ground surface temperature downscaling method at present, and the definition degree of the urban ground surface temperature data is improved.

At present, the common ground surface temperature downscaling method comprises an image space-time fusion method and a statistical regression method; the image space-time fusion method is to obtain LST data with continuous time sequence and high spatial resolution by carrying out fusion processing on the remote sensing image with low spatial resolution and high temporal resolution and the remote sensing image with high spatial resolution and low temporal resolution. The statistical regression method is to complete the whole process of downscaling by establishing a statistical regression relation between the earth surface temperature and the earth surface parameter multiple factors and assuming that the relation has space scale invariance.

However, the image space-time fusion method has strong dependence on the quality of historical data (no cloud in sunny days), cannot generally explain the physical mechanism of surface temperature inversion, and has low applicability in areas with large spatial heterogeneity (such as cities). The statistical regression method for the ground surface temperature is mainly based on medium-low spatial resolution remote sensing images, so that the spatial resolution of the ground surface temperature downscaling result is low, abnormal values exist in inversion results of a water body and a shadow area, the application effect is poor, the overall accuracy is low, and the statistical regression method is not suitable for urban areas with large spatial heterogeneity.

Disclosure of Invention

In view of the above problems, the invention provides a method and a device for reducing the urban surface temperature scale based on high-resolution satellite images, which mainly aims to realize obtaining urban high-resolution surface temperature data, effectively improve abnormal surface temperature of water bodies and shadow areas in cities and improve applicability.

In order to solve the technical problems, the invention provides the following scheme:

in a first aspect, the present invention provides a method for reducing the temperature of an urban surface based on high-resolution satellite images, the method comprising:

acquiring multi-source remote sensing data of a target area, and processing the multi-source remote sensing data by utilizing a preprocessing rule to obtain preprocessed multi-source remote sensing data; the multi-source remote sensing data at least comprises a middle-low spatial resolution thermal infrared remote sensing image, a middle-low spatial resolution multispectral remote sensing image and a high spatial resolution multispectral remote sensing image; the preprocessing rules at least comprise radiometric calibration, atmospheric correction, resampling and coordinate system conversion;

Based on the preprocessed middle-low spatial resolution thermal infrared remote sensing image and corresponding water vapor total amount data, inversion is carried out by utilizing a window splitting algorithm to obtain middle-low spatial resolution ground surface temperature data serving as to-be-downscaled data;

based on the preprocessed high-spatial-resolution multispectral remote sensing image, performing automatic classification of the two types of ground objects of the high-spatial-resolution multispectral remote sensing image by using a maximum inter-class variance method to obtain a water body and shadow ground object type region and a non-water body ground object type region corresponding to the target region;

calculating different spatial resolution downscaling parameter factors of the water body and shadow ground feature type region and the non-water body ground feature type region based on the preprocessed middle-low spatial resolution multispectral remote sensing image and the preprocessed high spatial resolution multispectral remote sensing image; the downscaling parameter factors comprise normalized difference water body indexes NDWI and normalized vegetation indexes NDVI, wherein the normalized difference water body indexes NDWI are downscaling parameter factors of the water body and shadow ground object type areas, and the normalized vegetation indexes NDVI are downscaling parameter factors of the non-water body ground object type areas;

Based on the to-be-downscaled data and the different spatial resolution downscaling parameter factors, constructing an earth surface temperature downscaling model corresponding to the water body, the shadow ground feature type region and the non-water body ground feature type region respectively by using the number of sample points of the target region; wherein,

the surface temperature scale-down model of the non-water body ground object type region comprises:

LST high score a11×ndv high score + ×b+b (1)

The ground surface temperature scale-down model of the water body and shadow ground object type region comprises the following steps of:

LST _{high score} ＝a ₁ *a ₂ *NDWI _{High score} +a ₁ *b ₂ +b ₁ (2)

In formulas (1) - (2), LST _{High score} NDVI for high spatial resolution surface temperature data _{High score} Normalizing vegetation index for high spatial resolution, NDWI _{High score} Normalizing differential water index for high spatial resolution, a ₁ 、b ₁ 、a ₂ And b ₂ Are all coefficients;

and calculating high-spatial-resolution surface temperature data of the target area by using the surface temperature downscaling model respectively corresponding to the water body and shadow surface feature type area and the non-water body surface feature type area to obtain a surface temperature downscaling result.

In a second aspect, the present invention provides a city ground surface temperature downscaling apparatus based on high resolution satellite images, comprising:

the preprocessing unit is used for acquiring multi-source remote sensing data of the target area, and processing the multi-source remote sensing data by utilizing a preprocessing rule to obtain preprocessed multi-source remote sensing data; the multi-source remote sensing data at least comprises a middle-low spatial resolution thermal infrared remote sensing image, a middle-low spatial resolution multispectral remote sensing image and a high spatial resolution multispectral remote sensing image; the preprocessing rules at least comprise radiometric calibration, atmospheric correction, resampling and coordinate system conversion;

The inversion unit is used for inverting the ground surface temperature data with the middle and low spatial resolutions by using a window splitting algorithm based on the preprocessed middle and low spatial resolution thermal infrared remote sensing image and corresponding water vapor total amount data to obtain the data to be downscaled;

the dividing unit is used for automatically dividing the two types of ground objects of the high-spatial-resolution multispectral remote sensing image by using a maximum inter-class variance method based on the preprocessed high-spatial-resolution multispectral remote sensing image to obtain a water body and shadow ground object type area and a non-water body ground object type area corresponding to the target area;

the first calculation unit is used for calculating different spatial resolution downscaling parameter factors of the water body and shadow ground object type area and the non-water body ground object type area based on the preprocessed middle-low spatial resolution multispectral remote sensing image and the preprocessed high spatial resolution multispectral remote sensing image; the downscaling parameter factors comprise normalized difference water body indexes NDWI and normalized vegetation indexes NDVI, wherein the normalized difference water body indexes NDWI are downscaling parameter factors of the water body and shadow ground object type areas, and the normalized vegetation indexes NDVI are downscaling parameter factors of the non-water body ground object type areas;

The construction unit is used for constructing an earth surface temperature downscaling model corresponding to the water body and shadow ground feature type area and the non-water body ground feature type area respectively by utilizing the number of sample points of the target area based on the to-be-downscaling data and the different spatial resolution downscaling parameter factors; wherein, the surface temperature of the non-water body ground object type area reduces the scale model:

LST _{high score} ＝a ₁ *a ₂ *NDVI _{High score} +a ₁ *b ₂ +b ₁ (1)

LST _{high score} ＝a ₁ *a ₂ *NDWI _{High score} +a ₁ *b ₂ +b ₁ (2)

the second calculation unit is used for calculating the high-spatial-resolution surface temperature data of the target area by using the surface temperature downscaling model corresponding to the water body and shadow surface feature type area and the non-water body surface feature type area respectively, so as to obtain a surface temperature downscaling result.

In order to achieve the above object, according to a third aspect of the present invention, there is provided a storage medium, which includes a stored program, wherein the program, when executed, controls a device in which the storage medium is located to perform the urban surface temperature downscaling method based on high-resolution satellite images according to the first aspect.

To achieve the above object, according to a fourth aspect of the present invention, there is provided an electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, the processor implementing all or part of the steps of the urban surface temperature downscaling apparatus for high resolution satellite image based according to the second aspect when the program is executed.

By means of the technical scheme, the urban surface temperature downscaling method and device based on the high-resolution satellite image are low in overall accuracy and are not suitable for urban areas with large space heterogeneity due to the fact that abnormal values exist in water bodies and shadow areas in the existing downscaling results. The method comprises the steps of carrying out preprocessing such as radiometric calibration, atmospheric correction, resampling, coordinate system conversion and the like on multi-source remote sensing data of a target area to obtain preprocessed multi-source remote sensing data; acquiring surface temperature data with middle and low spatial resolutions as to-be-downscaled data based on the preprocessed multi-source remote sensing data, and then automatically dividing the two types of ground objects of the high spatial resolution multi-spectral remote sensing image by using a maximum inter-class variance method based on the preprocessed high spatial resolution multi-spectral remote sensing image to obtain a water body and shadow ground object type area and a non-water body ground object type area corresponding to the target area; calculating different spatial resolution downscaling parameter factors of the water body and shadow ground feature type region and the non-water body ground feature type region based on the preprocessed middle-low spatial resolution multispectral remote sensing image and the preprocessed high spatial resolution multispectral remote sensing image; based on the to-be-downscaled data and the different spatial resolution downscaling parameter factors, constructing an earth surface temperature downscaling model respectively corresponding to the water body, the shadow ground feature type region and the non-water body ground feature type region by using the number of sample points of the target region; the problem of abnormal values of the downscaling results of the water body and the shadow area can be solved, and the overall accuracy is improved; according to the method, the high-spatial resolution surface temperature data of the target area are calculated by using the surface temperature scale-down model corresponding to the water body and shadow surface feature type area and the non-water body surface feature type area respectively, so that the surface temperature scale-down result is obtained, urban high-resolution surface temperature data can be obtained quickly, the abnormal surface temperature of the water body and shadow area in the city is effectively improved, the applicability in the urban area is high, and the method has high application value.

The foregoing description is only an overview of the present invention, and is intended to be implemented in accordance with the teachings of the present invention in order that the same may be more clearly understood and to make the same and other objects, features and advantages of the present invention more readily apparent.

Drawings

Various other advantages and benefits will become apparent to those of ordinary skill in the art upon reading the following detailed description of the preferred embodiments. The drawings are only for purposes of illustrating the preferred embodiments and are not to be construed as limiting the invention. Also, like reference numerals are used to designate like parts throughout the figures. In the drawings:

FIG. 1 shows a flow chart of a method for reducing urban surface temperature based on high-resolution satellite images, which is provided by the embodiment of the invention;

FIG. 2 shows a flowchart of another method for reducing the urban surface temperature based on high-resolution satellite images according to an embodiment of the invention;

FIG. 3 illustrates a model verification flow chart provided by an embodiment of the present invention;

FIG. 4 shows a block diagram of an urban surface temperature downscaling device based on high resolution satellite images according to an embodiment of the present invention;

FIG. 5 shows a block diagram of another urban surface temperature downscaling device based on high resolution satellite images according to an embodiment of the present invention;

FIG. 6 shows a model verification device provided by an embodiment of the present invention;

FIG. 7 shows an experimental sample area Landsat-8 image provided by an embodiment of the present invention;

fig. 8 shows a beijing No. two satellite image of an experimental sample area according to an embodiment of the present invention;

FIG. 9 is a diagram showing different coverage types of a target area according to an embodiment of the present invention;

FIG. 10 shows a graph of results provided by an embodiment of the present invention for the class II ground object classification (water/non-water) of FIG. 9;

FIG. 11 shows a graph of regression equations for a water body region (shaded) 30mNDWI and 30mLST provided by an embodiment of the present invention;

FIG. 12 shows a graph of regression equations for 30m NDVI and 30m LST for a non-water body region provided by an embodiment of the present invention;

FIG. 13 shows a regression equation diagram of a water body region (containing shadows) 3 mNDWI and 30mNDWI provided by an embodiment of the present invention;

FIG. 14 shows a regression equation diagram of 3m NDVI and 30m NDVI of a non-water body region provided by an embodiment of the present invention;

FIG. 15 shows a medium-low spatial resolution (30 meters) surface temperature map provided by an embodiment of the present invention;

fig. 16 shows a high spatial resolution (3 m) surface temperature map provided by an embodiment of the present invention.

Detailed Description

Exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be embodied in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.

For the problems that the existing ground surface temperature downscaling method has abnormality in water and shadow areas, the overall accuracy is low, and the method is not suitable for urban areas with large space heterogeneity. The inventor thinks of constructing a downscaling factor, and simultaneously adopts a bipartite ground object method to construct a downscaling model aiming at a water body (containing shadow) and a non-water body area respectively.

Therefore, the embodiment of the invention provides a city surface temperature downscaling method based on high-resolution satellite images, which is used for obtaining city high-resolution surface temperature data, effectively improving the abnormal surface temperature of water bodies and shadow areas in cities and improving the applicability, and specifically comprises the following implementation steps as shown in fig. 1:

101. and acquiring multi-source remote sensing data of the target area, and processing the multi-source remote sensing data by utilizing a preprocessing rule to obtain preprocessed multi-source remote sensing data.

The multi-source remote sensing data at least comprises a middle-low spatial resolution thermal infrared remote sensing image, a middle-low spatial resolution multispectral remote sensing image and a high spatial resolution multispectral remote sensing image. The preprocessing rules include at least radiometric scaling, atmospheric correction, resampling, and coordinate system conversion.

Specifically, the target area is an area in which the surface temperature is reduced, and mainly refers to a city area, for example: beijing city area. The multi-source remote sensing data are remote sensing image data from different data sources, as shown in table 1; the data sources may be Landsat-8 satellite, MOD05 (moisture content product) and Beijing No. two satellite, and the embodiment is not particularly limited. After remote sensing data of a certain area are acquired from the plurality of data sources, preprocessing such as radiometric calibration, atmospheric correction, resampling, coordinate system conversion and the like are sequentially carried out, so that preprocessed multi-source remote sensing data are obtained.

It should be noted that the mid-low spatial resolution thermal infrared remote sensing image and the mid-low spatial resolution multispectral remote sensing image are remote sensing data from the same data source in the same time; the high spatial resolution multispectral remote sensing image is a high spatial resolution multispectral satellite remote sensing image from a synchronous passing or small time difference (within 3 days) different from the data source.

Illustrating:

TABLE 1

Data type	Data source	Spatial resolution	Use of the same
				Thermal infrared remote sensing image	Landsat-8	Middle and low score	Calculating middle-low score LST
Total amount of water vapor	MOD05	Middle and low score	Calculating middle-low score LST
				Multispectral remote sensing image	Landsat-8	Middle and low score	Calculating middle-low earth surface parameters
Multispectral remote sensing image	Beijing No. two	High score	Calculating high-resolution surface parameters

102. And (3) based on the preprocessed middle-low spatial resolution thermal infrared remote sensing image and corresponding water vapor total amount data, inversion is performed by utilizing a splitting window algorithm to obtain middle-low spatial resolution ground surface temperature data serving as the to-be-reduced scale data.

After the preprocessed multi-source remote sensing data are obtained from the step 101, a preprocessed middle-low spatial resolution thermal infrared remote sensing image is further obtained, and then the ground surface temperature data with middle-low spatial resolution is obtained through inversion by using a splitting window algorithm through a water vapor total amount product (MOD 05) in the same region and the same time period as the multi-source remote sensing data and is used as the to-be-reduced scale data.

103. And based on the preprocessed high-spatial-resolution multispectral remote sensing image, automatically dividing the two types of ground objects of the high-spatial-resolution multispectral remote sensing image by using a maximum inter-class variance method to obtain a water body and shadow ground object type area and a non-water body ground object type area corresponding to the target area.

After the preprocessed multi-source remote sensing data are obtained from the step 101, a preprocessed high-spatial-resolution multi-spectrum remote sensing image is further obtained, a reflectivity value K corresponding to a near-infrared reflectivity maximum variance category of the near-infrared band reflectivity is calculated by adopting an Otsu algorithm (maximum inter-class variance method) proposed by Japanese scholars in Dajinzhen, and the high-spatial-resolution multi-spectrum remote sensing image is automatically divided into 2 types of ground object types, namely a water body (containing shadow) and a non-water body, so that a water body and shadow ground object type region and a non-water body ground object type region corresponding to a target region are obtained.

104. And calculating different spatial resolution downscaling parameter factors of the water body, the shadow ground feature type region and the non-water body ground feature type region based on the preprocessed middle-low spatial resolution multispectral remote sensing image and the preprocessed high spatial resolution multispectral remote sensing image.

The downscaling parameter factors comprise a normalized difference water body index NDWI (Normalized Difference Water Index) and a normalized vegetation index NDVI (Normalized Difference Vegetation Index), the normalized difference water body index NDWI is the downscaling parameter factor of the water body and shadow ground object type area, and the normalized vegetation index NDVI is the downscaling parameter factor of the non-water body ground object type area.

Combining the water body (containing shadow) and non-water body land coverage types in the urban area and the research basis of index factors commonly used in high-spatial resolution image ground feature classification and identification, and finally respectively selecting normalized difference water body index NDWI as the surface temperature scale-down factor of the water body area (containing shadow); and selecting the normalized vegetation index NDVI as the surface temperature scale-down factor of the non-water body area.

And (3) obtaining automatically divided water body, shadow ground object type areas and non-water body ground object type areas from the step (103), and obtaining preprocessed middle-low spatial resolution multispectral remote sensing images and preprocessed high spatial resolution multispectral remote sensing images respectively corresponding to the water body, shadow ground object type areas and non-water body ground object type areas based on the preprocessed middle-low spatial resolution multispectral remote sensing images and the preprocessed high spatial resolution multispectral remote sensing images obtained from the step (101).

And then calculating a middle-low spatial resolution downscaling factor NDVI based on the multispectral remote sensing image using the middle-low spatial resolution _{Middle and low score} And NDWI _{Middle and low score} The method comprises the steps of carrying out a first treatment on the surface of the Calculating high spatial resolution downscaling factor NDVI using high spatial resolution multispectral remote sensing images _{High score} And NDWI _{High score} 。

105. And constructing a ground surface temperature downscaling model corresponding to the water body and shadow ground object type region and the non-water body ground object type region respectively by utilizing the number of sample points of the target region based on the to-be-downscaling data and different spatial resolution downscaling parameter factors.

Wherein, the surface temperature of the non-water body ground object type area reduces the scale model:

LST _{high score} ＝a ₁ *a ₂ *NDVI _{High score} +a ₁ *b ₂ +b ₁

LST _{high score} ＝a ₁ *a ₂ *NDWI _{High score} +a ₁ *b ₂ +b ₁

In the formula, LST _{High score} NDVI for high spatial resolution surface temperature data _{High score} Normalizing vegetation index for high spatial resolution, NDWI _{High score} Normalizing differential water index for high spatial resolution, a ₁ 、b ₁ A2 and b ₂ Are coefficients.

The data to be downscaled can be obtained from step 102, and different spatial resolution downscaling parameter factors, namely medium and low spatial resolution downscaling parameter factors NDVI, can be obtained from step 104 _{Middle and low score} And NDWI _{Middle and low score} High spatial resolution downscaling parameter factor NDVI _{High score} And NDWI _{High score} 。

Firstly, setting the number of sample points of a target area, wherein the number or the proportion of the sample points can be directly set, for example: the total number of grids is 250000, the number of sample points is 0.01, and the embodiment is not particularly limited. Obtaining random points of the shadow water body ground feature type area and the non-water body ground feature type area and data of points where the random points are located by utilizing the sample points of the target area; and constructing a ground surface temperature downscaling model corresponding to the water body shadow ground feature type region and the non-water body ground feature type region respectively through a linear regression relation between the downscaling parameter factors and ground surface temperature data and a linear regression relation between the downscaling parameter factors of different scales.

Illustrating:

(1) According to the total number N of medium-low spatial resolution image grids _p Determining a satisfactory duty ratio R by adopting a random point duty ratio principle _s And calculating to obtain the total number C of random point selection. Dividing the random point into a water body region (containing shadow) sample point number C by utilizing the ground object classification result obtained in the step 103 _{Water and its preparation method} And non-water body region sample number C _Non-aqueous 。

(2) Sample point C for non-water body region _Non-aqueous The NDVI is adopted as a ground surface temperature downscaling parameter factor, the obtained 4/5 sampling points are used as model training sampling points, 1/5 sampling points are used as model verification sampling points, and the construction process of the ground surface temperature downscaling model of the non-water body area is as follows:

LST _{middle and low score} ＝a ₁ *NDVI _{Middle and low score} +b ₁ (1)

NDVI _{Middle and low score} ＝a ₂ *NDVI _{High score} +b ₂ (2)

Carrying out formula (2) into formula (1) to obtain a ground surface temperature scale-down model of the non-water body region:

LST _{high score} ＝a ₁ *a ₂ *NDWI _{High score} +a ₁ *b ₂ +b ₁ (3)

Sample point C for water body region (containing shadow) _{Water and its preparation method} NDWI is adopted as a ground surface temperature downscaling parameter factor, the obtained 4/5 sampling points are used as model training sampling points, and 1/5 sampling points are used as model training sampling pointsThe model verification sample point and the surface temperature scale-down model process of the water body region (containing shadow) are as follows:

LST _{middle and low score} ＝a ₁ *NDWI _{Middle and low score} +b ₁ (4)

NDWI _{Middle and low score} ＝a ₂ *NDWI _{High score} +b ₂ (5)

Bringing the formula (5) into the formula (4) to obtain a surface temperature downscaling model of the water body region (containing shadows):

LST _{High score} ＝a ₁ *a ₂ *NDWI _{High score} +a ₁ *b ₂ +b ₁ (6)

106. And calculating high-spatial-resolution surface temperature data of the target area by using a surface temperature downscaling model respectively corresponding to the water body, the shadow surface feature type area and the non-water body surface feature type area to obtain a surface temperature downscaling result.

After the surface temperature downscaling model corresponding to the water body and shadow ground object type region and the non-water body ground object type region is obtained from the step 105, the model is utilized to calculate and obtain the whole surface temperature data with high spatial resolution in the target region, the surface temperature downscaling result is obtained, the downscaling result is graded, and a thematic map is manufactured.

Based on the implementation manner of the embodiment of fig. 1, the invention provides a city ground surface temperature downscaling method based on high-resolution satellite images, which constructs downscaling parameter factors based on the high-spatial-resolution images and improves the spatial resolution of ground surface temperature downscaling results; meanwhile, a two-part ground object method is adopted to respectively construct a scale-reducing model aiming at a water body (containing shadow) and a non-water body area, so that the problem of abnormal values of scale-reducing results of the water body and the shadow area is solved, and the overall accuracy is improved. The method is applied to urban areas with complex earth surfaces, so that urban high-resolution earth surface temperature data can be obtained quickly, abnormal earth surface temperatures of water bodies and shadow areas in cities can be effectively improved, applicability in urban areas is high, and the method has high application value.

Further, as a refinement and expansion of the embodiment shown in fig. 1, the embodiment of the present invention further provides another urban ground surface temperature downscaling method based on high-resolution satellite images, as shown in fig. 2, which specifically includes the following steps:

201. and acquiring multi-source remote sensing data of the target area, and processing the multi-source remote sensing data by utilizing a preprocessing rule to obtain preprocessed multi-source remote sensing data.

This step is described in conjunction with step 101 in the above method, and the same contents are not repeated here.

Acquiring the middle-low spatial resolution remote sensing image of the target region in a preset time period, wherein the middle-low spatial resolution remote sensing image comprises a thermal infrared remote sensing image and a multispectral remote sensing image; acquiring the high-spatial-resolution multispectral remote sensing image of the target area according to a preset time difference based on the preset time period corresponding to the medium-low spatial-resolution remote sensing image; performing pretreatment of radiation calibration, atmosphere correction, resampling and coordinate system conversion on the middle-low spatial resolution remote sensing image to obtain a pretreated middle-low spatial resolution thermal infrared remote sensing image and a pretreated middle-low spatial resolution multispectral image; the preprocessed related data of the mid-low spatial resolution multispectral image comprises reflectivity data. Performing pretreatment of radiation calibration, atmosphere correction, resampling and coordinate system conversion on the high-spatial-resolution multispectral remote sensing image to obtain a pretreated high-spatial-resolution multispectral image; the preprocessed related data of the high-spatial-resolution multispectral image comprises reflectivity data, wherein resampling of the high-spatial-resolution multispectral image is according to the principle of maintaining integer multiple relation with the spatial resolution of the medium-low-spatial-resolution remote sensing image.

Illustrating:

(1) And (3) data acquisition: acquiring remote sensing images with 30m spatial resolution of Landsat8 in Beijing city at 9/20/2020, including thermal infrared and multispectral images, as shown in FIG. 7; the multi-spectral remote sensing image with the spatial resolution of Beijing No. two 3.2m in 9 months of 2020 and 19 days closest to the imaging time of the image is acquired, the requirement of image acquisition within 3 days is met, as shown in fig. 8, and the target area is a black dotted line area in fig. 7 and 8. The data of the specific wavelength band used is shown in table 2.

TABLE 2

Time	Data source	Wave band	Resolution ratio
				2020.09.20	Landsat-8	Thermal infrared	30m
2020.09.20	Landsat-8	Multispectral	30m
				2020.09.20	MOD05	Total amount of water vapor	1000 meters
2020.09.19	Beijing No. two	Multispectral	3.2 meters

(2) Pretreatment: and respectively carrying out pretreatment such as radiometric calibration, atmospheric correction, resampling, coordinate system conversion and the like on Landsat-8 and Beijing two-number data. It should be noted that, in order to improve the matching degree between grid pixels of the downscale model, a nearest neighbor method is required to resample the 3.2m Beijing No. two multispectral image to 3 m according to implementation steps.

202. And (3) based on the preprocessed middle-low spatial resolution thermal infrared remote sensing image and corresponding water vapor total amount data, inversion is performed by utilizing a splitting window algorithm to obtain middle-low spatial resolution ground surface temperature data serving as the to-be-reduced scale data.

This step is described in conjunction with step 102 in the above method, and the same contents are not repeated here.

Illustrating:

the data of the 10 th and 11 th waves Duan Regong of Landsat8 subjected to 201 pretreatment and the data of the total amount of MOD05 water vapor are calculated by adopting a window splitting algorithm, so that the surface temperature LST with medium-low spatial resolution is obtained _30m And taking the data as the data to be downscaled.

203. And calculating the reflectivity value corresponding to the near infrared reflectivity maximum variance category by adopting a maximum inter-category variance method based on the preprocessed high-spatial resolution multispectral remote sensing image.

This step is described in conjunction with step 103 in the above method, and the same contents are not repeated here.

204. And (3) based on the reflectivity value, automatically dividing the pretreated high-spatial-resolution multispectral image into water body, shadow ground object types and non-water body ground object types, and obtaining water body, shadow ground object type areas and non-water body ground object type areas.

Illustrating:

the maximum inter-class variance method is adopted to automatically divide water bodies (containing shadows) and non-water bodies respectively for near infrared bands of the 3m space resolution Beijing No. two multispectral remote sensing image (shown in fig. 9) preprocessed in the step 201, and finally the rho is determined _NIR The result of the threshold divided water body (containing shadows) and non-water body =0.2 is optimal, and therefore, it is regarded as water body (containingShadow) and the end result of the non-water body automatic partitioning (as shown in fig. 10).

205. And calculating different spatial resolution downscaling parameter factors of the water body, the shadow ground feature type region and the non-water body ground feature type region based on the preprocessed middle-low spatial resolution multispectral remote sensing image and the preprocessed high spatial resolution multispectral remote sensing image.

Illustrating:

respectively utilizing 30m space resolution Landsat8 multispectral remote sensing images and 3m space resolution Beijing No. two multispectral remote sensing images after preprocessing in the step 201 to calculate downscale parameter factors of different scales:

(1) calculation of NDVI using 3 rd and 4 th bands of Landsat8 _30m Calculating NDWI using the 2 nd and 4 th bands _30m ；

(2) NDVI calculation using No. 3 and No. 4 bands of Beijing No. two satellites _3m Calculation of NDWI with 2 nd and 4 th bands _3m 。

The calculated NDWI _30m 、NDWI _3m 、NDVI _30m And NDVI _3m Respectively used as the surface temperature scale-down parameter factors of the water body (containing shadow) and the non-water body area in the target area.

206. And constructing a ground surface temperature downscaling model corresponding to the water body and shadow ground object type region and the non-water body ground object type region respectively by utilizing the number of sample points of the target region based on the to-be-downscaling data and different spatial resolution downscaling parameter factors.

This step is described in conjunction with step 104 in the above method, and the same contents are not repeated here.

Step one: and obtaining the number of samples of each ground object type area of the target area.

Wherein the ground object type region includes the water body and shadow ground object type region and the non-water body ground object type region.

Specifically, according to the total number of grids of the medium-low spatial resolution remote sensing image of the target area, determining the required duty ratio number by adopting a random point duty ratio principle, and calculating to obtain the random point selection total number; and obtaining the number of the sample points of each ground object type area by utilizing the random point selection total number.

Illustrating:

assuming that the number of columns of the medium-low spatial resolution image of the target area is 500 x 500, the total number of grids is N _p 250000, R is determined according to the random point duty ratio principle _s 0.01, and further the total number of random spots C was 2500. 2500 points are randomly selected in the target area, and according to the result of the two-way ground feature of the high-spatial-resolution remote sensing image, 2315 points in total fall on the water (shadow-containing) area and 185 points in total fall on the non-water area.

Step two: and respectively dividing the sample number of each ground object type region into corresponding model training sample number and model verification sample number according to a preset proportion.

Illustrating:

taking the obtained 4/5 sampling points as model training sampling points and taking 1/5 sampling points as model verification sampling points.

Step three: and obtaining a first type of linear regression equation between the medium-low spatial resolution surface temperature data of each ground feature type region and the corresponding medium-low spatial resolution downscaling parameter factors and a second type of linear regression equation between different spatial resolution downscaling parameter factors by utilizing the number of the model training sample points.

Specifically, the third step is further divided into the following steps:

(1) Obtaining random points corresponding to the water body and shadow ground object type areas and the non-water body ground object type areas by utilizing the number of points of the model training samples;

(2) Acquiring medium-low spatial resolution surface temperature data, medium-low spatial resolution normalized difference water body index NDWI and high spatial resolution normalized difference water body index NDWI of points where the random points of the water body and shadow ground object type areas are located, medium-low spatial resolution surface temperature data, medium-low spatial resolution normalized vegetation index NDVI and high spatial resolution normalized vegetation index NDVI of points where the random points of the non-water body ground object type areas are located;

(3) Based on the medium-low spatial resolution surface temperature data of the point positions of the random points of the water body and shadow ground object type region and the medium-low spatial resolution normalized difference water body index NDWI, constructing a first type linear regression equation of the water body and shadow ground object type region: LST (least squares) _{Middle and low score} ＝a ₁ *NDWI _{Middle and low score} +b ₁ In the LST _{Middle and low score} For the medium-low spatial resolution surface temperature data, NDWI is divided into medium-low spatial resolution normalized difference water indexes, and a1 and b1 are coefficients:

illustrating:

and in the water body area, selecting NDWI as an earth surface temperature scale-down parameter factor. For the random points of the screened water body (containing shadow) area, the method comprises the following steps of 8: and 2, constructing a regression model. Determining LST in a Water (shadow-containing) region _30m With NDWI _a0m Is a regression equation of (2). FIG. 11 is an LST _30m With NDWI _30m Is used for determining LST finally _30m With NDWI _30m The linear regression relationship between:

LST _30m ＝6.07*NDWI _30m +29.72 (7)

(4) Based on the medium-low spatial resolution surface temperature data of the point position where the random point of the non-water body ground feature type region is located and the medium-low spatial resolution normalized vegetation index NDVI, constructing a first type linear regression equation of the non-water body ground feature type region: LST (least squares) _{Middle and low score} ＝a ₁ *NDVI _{Middle and low score} +b ₁ In the LST _{Middle and low score} NDVI for medium and low spatial resolution surface temperature data _{Middle and low score} Normalizing vegetation index for medium and low spatial resolution, a ₁ And b ₁ Are all coefficients;

illustrating:

in a non-water area, NDVI is used as a downscaled surface parameter. Aiming at the screened non-water body area random points, the method comprises the following steps of: and 2, constructing a regression model. Determining LST in non-water body region _30m With NDVI _30m Is a regression equation of (2). FIG. 12 is LST _30m With NDVI _30m Is used for determining LST finally _30m With NDVI _30m The linear regression relationship between:

LST _30m ＝-9.29*NDVI _30m +31.40 (8)

(5) Based on the normalized difference water body index NDWI of middle and low spatial resolutions and the normalized difference water body index NDWI of high spatial resolutions of the water body and shadow feature type region, constructing a second type linear regression equation of the water body and shadow feature type region: NDWI _{Middle and low score} ＝a ₂ *NDWI _{High score} +b ₂ In NDWI _{Middle and low score} NDWI for normalizing differential water index for medium and low spatial resolution _{High score} Normalizing differential water index for high spatial resolution, a ₂ And b ₂ Are all coefficients;

illustrating:

as in example (3), NDWI is determined in the water (shadow-containing) region as shown in FIG. 13 _3m And NDWI _30m The regression equation of (2) is:

NDWI _30m ＝0.21*NDWI _3m -0.22 (9)

(6) Based on the medium-low spatial resolution normalized vegetation index NDVI and the high spatial resolution normalized vegetation index NDVI of the non-water body ground object type region, constructing a second type linear regression equation of the non-water body ground object type region: NDVI _{Middle and low score} ＝a ₂ *NDVI _{High score} +b ₂ Wherein, NDVI _{Middle and low score} Normalized vegetation index for medium and low spatial resolution, NDVI _{High score} Normalizing vegetation index for high spatial resolution, a ₂ And b ₂ Are coefficients.

Illustrating:

as in example (4), NDVI is determined in a non-water region as shown in FIG. 14 _3m And NDVI _30m The regression equation of (2) is:

NDVI _30m ＝0.54*NDVI _3m -0.20 (10)

step four: and acquiring a ground surface temperature scale reduction model corresponding to each ground object type region based on the first type linear regression equation and the second type linear regression equation.

Illustrating:

according to formulas (1) - (3) and formulas (8), (10), the 3m NDVI-3m LST (DEG C) model of the non-water body area is finally obtained as follows: LST (least squares) _3m ＝-5.02*NDVI _3m +33.26 (11), where NDVI _3m NDVI data representing 3m spatial resolution of beijing satellite No. two; LST (least squares) _3m LST representing 3m spatial resolution of beijing satellite No. two.

According to formulas (4) - (6) and formulas (7) and (9), the model of the water body region 3m NDWI-3m LST (DEG C) is finally obtained as follows: LST (least squares) _3m ＝1.27*NDWI _3m +28.38 (12), where NDWI _3m NDWI data representing 3m spatial resolution of beijing satellite No. two; LST (least squares) _3m Ground surface temperature data representing 3m spatial resolution of Beijing No. two satellites.

207. And calculating high-spatial-resolution surface temperature data of the target area by using a surface temperature downscaling model respectively corresponding to the water body, the shadow surface feature type area and the non-water body surface feature type area to obtain a surface temperature downscaling result.

Illustrating:

the constructed urban surface temperature downscaling model (the surface temperature downscaling model corresponding to the water body and shadow ground object type region and the non-water body ground object type region respectively) is applied to the remote sensing image, and the inversion is carried out to obtain the surface temperature spatial distribution result with 3m spatial resolution in the whole target region, wherein the overall verification precision is better than 95%, and the practical application requirement is met. And then grading the surface temperature in the surface temperature spatial distribution result and manufacturing a thematic map, wherein the final inversion result is shown in fig. 15 and 16.

Based on the implementation mode of the above-mentioned figure 2, it can be seen that the invention provides a city ground surface temperature downscaling method based on high-resolution satellite images, and the invention adopts resampling according to the principle that the spatial resolution of the medium-low spatial resolution image and the spatial resolution of the high-spatial resolution image need to maintain an integer multiple relation, so that the matching degree between grid pixels of the downscaling model can be improved; the target area is divided into a water body (containing shadow) and a non-water body area by adopting a bipartite ground object method, the downscaling model construction is respectively carried out, the problem of abnormal values of downscaling results of the water body and the shadow area is solved, and the overall accuracy is improved. The method solves the problems of abnormal values, poor application effect and the like of inversion results of the existing ground surface temperature downscaling method in water and shadow areas, and has strong applicability in urban areas with strong space heterogeneity.

After the earth surface temperature downscaling model is constructed, verification is needed, so that the embodiment shown in fig. 1 and 2 is supplemented, and the embodiment of the invention also provides a model verification method, as shown in fig. 2, which comprises the following specific steps:

301. and verifying the number of sample points by using the model, and obtaining random points corresponding to the water body and shadow ground feature type region and the non-water body ground feature type region respectively.

And (3) performing accuracy verification on the downscaling result by using LST to-be-downscaling data with medium-low spatial resolution as an indirect verification basis by adopting a cross verification method, namely adopting a model construction method with the ratio of 8:2 of the total number of random points when constructing the model in a water body (including shadow) and a non-water body region, wherein 80% of the LST to-be-downscaling data is used for model construction based on downscaling parameter factors, and 20% of the LST to-be-downscaling data is used for accuracy verification of the downscaling model inversion result.

Illustrating:

assuming that the total number of random sample points C of the target area is 2500, the total number of the point positions falling on the water body (shadow-containing) area is 2315, the total number of the point positions falling on the non-water body area is 185, 4/5 sample points are taken as model training sample points, 1/5 sample points are taken as model verification sample points, the number of the model verification sample points falling on the water body (shadow-containing) area is 2315/5=463, and the number of the model verification sample points falling on the non-water body area is 185/5=37; and randomly selecting random points in the respective areas based on the model verification sample points, and further obtaining the point positions of the random points.

302. And based on the to-be-downscaled data, acquiring the surface temperature data of the middle-low spatial resolution of the point where the random point is located by a value extraction method, and taking the surface temperature data as a precision verification true value.

After obtaining the random points corresponding to the water body and shadow ground object type region and the non-water body ground object type region respectively from step 301, based on the to-be-downscaled data (the ground surface temperature data with middle and low spatial resolutions) and the point positions where the random points are located, obtaining the ground surface temperature data with middle and low spatial resolutions of the point positions where the random points are located through a value extraction method, and using the ground surface temperature data as a precision verification true value for model precision verification.

303. Inversion is carried out by using a ground surface temperature scale-down model to obtain ground surface temperature data with high spatial resolution of the point where the random point is located.

After obtaining the random points corresponding to the water body and shadow ground object type region and the non-water body ground object type region respectively from step 301, inversion is performed by using the ground surface temperature scale-down model to obtain the ground surface temperature data with high spatial resolution of the point where the random points are located.

304. And respectively calculating the scale reduction accuracy of the surface temperature data of the water body, the shadow ground feature type region and the non-water body ground feature type region by using a preset accuracy calculation formula based on the surface temperature data of the medium and low spatial resolutions and the surface temperature data of the high spatial resolutions of the point positions where the random points are located.

The preset precision calculation formula is as follows:

in the formula, delta is the accuracy of the down scale of the surface temperature data of the point where the random point is located, and LST _{High score} LST for high spatial resolution surface temperature data _{Middle and low score} Is the surface temperature data with medium and low spatial resolution.

Obtaining surface temperature data of medium-low spatial resolution of the point location of the random point from the step 302; obtaining high spatial resolution surface temperature data of the point location of the random point from step 303; and then respectively calculating the ground surface temperature data downscaling precision of the water body shadow ground object type region and the non-water body ground object type region by using a preset precision calculation formula.

Illustrating:

LST calculated based on the data in Table 2 _{High score} And LST _{Middle and low} The city surface temperature downscaling inversion precision based on the dichotomous ground objects is calculated by a preset precision calculation formula and is as follows: the verification average precision of the non-water area is as follows: 95.33%; the average accuracy of the verification of the water (shadow-containing) area is 96.73%.

305. And calculating the average value of the ground surface temperature data downscaling precision of the water body and shadow ground object type areas and the non-water body ground object type areas to obtain the overall precision of the downscaling model of the target area.

Step 304, respectively solving LST downscaling precision delta of the water body (containing shadow) by a preset precision calculation formula _{Water and its preparation method} And LST downscaling accuracy delta for non-water body region _Non-aqueous Then, the step of averaging the two to obtain the overall accuracy delta of the urban surface temperature downscaling model _{Total (S)} . For example: (95.33% + 96.73%)/2= 96.03%.

Based on the implementation manner of fig. 3, it can be seen that the invention provides a model verification method for verifying the accuracy of a constructed model, and the invention can ensure that the accuracy of the constructed model reaches the standard by establishing verification steps, and ensure that the output result of the model is more accurate when the model is applied, thereby solving the problem of abnormal values of the downscaling results of water and shadow areas and improving the overall accuracy.

Furthermore, as an implementation of the method shown in fig. 1, the embodiment of the invention further provides an urban ground surface temperature downscaling device based on the high-resolution satellite image, which is used for implementing the method shown in fig. 1. The embodiment of the device corresponds to the embodiment of the method, and for convenience of reading, details of the embodiment of the method are not repeated one by one, but it should be clear that the device in the embodiment can correspondingly realize all the details of the embodiment of the method. As shown in fig. 4, the apparatus includes:

A preprocessing unit 41, configured to obtain multi-source remote sensing data of a target area, and process the multi-source remote sensing data by using a preprocessing rule to obtain preprocessed multi-source remote sensing data; the multi-source remote sensing data at least comprises a middle-low spatial resolution thermal infrared remote sensing image, a middle-low spatial resolution multispectral remote sensing image and a high spatial resolution multispectral remote sensing image; the preprocessing rules at least comprise radiometric calibration, atmospheric correction, resampling and coordinate system conversion;

an inversion unit 42, configured to invert the obtained surface temperature data with middle and low spatial resolutions by using a window splitting algorithm based on the preprocessed middle and low spatial resolution thermal infrared remote sensing image obtained from the preprocessing unit 41 and corresponding total amount of water vapor, as to-be-downscaled data;

a dividing unit 43, configured to automatically divide the two types of ground objects of the high-spatial resolution multispectral remote sensing image by using a maximum inter-class variance method based on the preprocessed high-spatial resolution multispectral remote sensing image obtained from the preprocessing unit 41, so as to obtain a water body and shadow ground object type area and a non-water body ground object type area corresponding to the target area;

A first calculating unit 44, configured to calculate different spatial resolution downscaling parameter factors of the water body and shadow feature type region and the non-water body feature type region based on the preprocessed mid-low spatial resolution multispectral remote sensing image and the preprocessed high spatial resolution multispectral remote sensing image obtained from the preprocessing unit 41; the downscaling parameter factors comprise normalized difference water body indexes NDWI and normalized vegetation indexes NDVI, wherein the normalized difference water body indexes NDWI are downscaling parameter factors of the water body and shadow ground object type areas, and the normalized vegetation indexes NDVI are downscaling parameter factors of the non-water body ground object type areas;

a construction unit 45, configured to construct an earth surface temperature downscaling model corresponding to the water body and shadow ground feature type region and the non-water body ground feature type region respectively according to the to-be-downscaling data obtained from the inversion unit 42 and the different spatial resolution downscaling parameter factors obtained from the first calculation unit 44, by using the number of sample points of the target region; wherein,

LST _{High score} ＝a ₁ *a ₂ *NDVI _{High score} +a ₁ *b ₂ +b ₁ (1)

LST _{high score} ＝a ₁ *a ₂ *NDWI _{High score} +a ₁ *b ₂ +b ₁ (2)

and a second calculation unit 46, configured to calculate high-spatial resolution surface temperature data of the target area by using the surface temperature scale-down model corresponding to the water body and shadow surface feature type area and the non-water body surface feature type area obtained from the construction unit 45, so as to obtain a surface temperature scale-down result.

Furthermore, as an implementation of the method shown in fig. 2, the embodiment of the invention further provides another urban ground surface temperature downscaling device based on the high-resolution satellite image, which is used for implementing the method shown in fig. 2. The embodiment of the device corresponds to the embodiment of the method, and for convenience of reading, details of the embodiment of the method are not repeated one by one, but it should be clear that the device in the embodiment can correspondingly realize all the details of the embodiment of the method. As shown in fig. 5, the apparatus includes:

LST _{High score} ＝a ₁ *a ₂ *NDVI _{High score} +a ₁ *b ₂ +b ₁ (1)

LST _{high score} ＝a ₁ *a ₂ *NDWI _{High score} +a ₁ *b ₂ +b ₁ (2)

Further, the construction unit 45 includes:

a first obtaining module 451, configured to obtain the number of samples of each ground object type region of the target region; wherein the ground object type region includes the water body and shadow ground object type region and the non-water body ground object type region:

the distribution module 452 is configured to divide the number of samples of each of the ground object type areas obtained from the first obtaining module 451 into a corresponding number of model training samples and a corresponding number of model verification samples according to a preset ratio;

The second obtaining module 453 is configured to obtain a first type of linear regression equation between the medium-low spatial resolution surface temperature data and the corresponding medium-low spatial resolution downscaling parameter factors and a second type of linear regression equation between different spatial resolution downscaling parameter factors of each ground object type region by using the model training sample points obtained from the allocating module 452;

the third obtaining module 454 is configured to obtain a ground surface temperature scale-down model corresponding to each ground feature type area based on the first type linear regression equation and the second type linear regression equation obtained from the second obtaining module 453.

Further, the second obtaining module 453 includes:

the first obtaining submodule 4531 is configured to obtain random points corresponding to the water body and shadow ground feature type region and the non-water body ground feature type region by using the number of points of the model training sample;

a second obtaining sub-module 4532, configured to obtain medium-low spatial resolution surface temperature data, medium-low spatial resolution normalized difference water body index NDWI, high spatial resolution normalized difference water body index NDWI, medium-low spatial resolution surface temperature data, medium-low spatial resolution normalized vegetation index NDVI, and high spatial resolution normalized vegetation index NDVI of the point where the random point of the water body and shadow feature type region is located, which are obtained from the first obtaining sub-module 4531;

The first construction submodule 4533 is configured to construct a first type of linear regression equation of the water body and shadow feature type region based on the medium-low spatial resolution surface temperature data of the point where the random point of the water body and shadow feature type region is located and the medium-low spatial resolution normalized difference water body index NDWI obtained from the second obtaining submodule 4532: LST (least squares) _{Middle and low score} ＝a ₁ *NDWI _{Middle and low score} +b ₁ In the LST _{Middle and low score} NDWI for medium and low spatial resolution surface temperature data _{Middle and low score} Normalizing the differential water index for medium and low spatial resolution, a ₁ And b ₁ Are all coefficients;

the second construction submodule 4534 is configured to construct a first type of linear regression equation of the non-water feature type region based on the medium-low spatial resolution surface temperature data of the point where the random point of the non-water feature type region is located and the medium-low spatial resolution normalized vegetation index NDVI, which are obtained from the second obtaining submodule 4532: LST (least squares) _{Middle and low score} ＝a ₁ *NDVI _{Middle and low score} +b ₁ In the LST _{Middle and low score} NDVI for medium and low spatial resolution surface temperature data _{Middle and low score} For medium and low spatial resolutionNormalized vegetation index, a ₁ And b ₁ Are all coefficients;

a third construction sub-module 4535, configured to construct a second type linear regression equation of the water body and shadow feature type region based on the normalized difference water body index NDWI of the middle-low spatial resolution and the normalized difference water body index NDWI of the high spatial resolution obtained from the second obtaining sub-module 4532: NDWI _{Middle and low score} ＝a ₂ *NDWI _{High score} +b ₂ In NDWI _{Middle and low score} NDWI for normalizing differential water index for medium and low spatial resolution _{High score} Normalizing differential water index for high spatial resolution, a ₂ And b ₂ Are all coefficients;

a fourth construction sub-module 4536, configured to construct a second type linear regression equation of the non-water body ground object type region based on the medium-low spatial resolution normalized vegetation index NDVI and the high spatial resolution normalized vegetation index NDVI of the non-water body ground object type region obtained from the second obtaining sub-module 4532: NDVI _{Middle and low score} ＝a ₂ *NDVI _{High score} +b ₂ Wherein, NDVI _{Middle and low score} Normalized vegetation index for medium and low spatial resolution, NDVI _{High score} Normalizing vegetation index for high spatial resolution, a ₂ And b ₂ Are coefficients.

Further, the first obtaining module 451 includes:

a calculating submodule 4511, configured to determine a ratio number meeting requirements according to a ratio principle of random points according to a total number of grids of the medium-low spatial resolution remote sensing image in the target area, and calculate to obtain a total number of random point selections;

and an obtaining sub-module 4512, configured to obtain the number of samples of each ground object type region by using the random point selection total obtained from the calculating sub-module 4511.

Further, the preprocessing unit 41 includes:

a first obtaining module 411, configured to obtain the middle-low spatial resolution remote sensing image of the target area in a preset period, where the middle-low spatial resolution remote sensing image includes a thermal infrared remote sensing image and a multispectral remote sensing image;

a second obtaining module 412, configured to obtain the high spatial resolution multispectral remote sensing image of the target area according to a preset time difference based on the preset time period corresponding to the middle-low spatial resolution remote sensing image obtained from the first obtaining module 411;

a first preprocessing module 413, configured to perform preprocessing of radiation calibration, atmospheric correction, resampling, and coordinate system conversion on the mid-low spatial resolution remote sensing image obtained from the first obtaining module 411, so as to obtain the mid-low spatial resolution thermal infrared remote sensing image and the mid-low spatial resolution multispectral image after preprocessing;

a second preprocessing module 414, configured to perform preprocessing of radiation calibration, atmospheric correction, resampling, and coordinate system conversion on the high spatial resolution multispectral remote sensing image obtained by the second obtaining module 412, so as to obtain a preprocessed high spatial resolution multispectral; the resampling of the high-spatial-resolution multispectral image is based on the principle of keeping integer multiple relation with the spatial resolution of the medium-low spatial-resolution remote sensing image.

Further, the dividing unit 43 includes:

the calculating module 431 is configured to calculate, based on the preprocessed high spatial resolution multispectral remote sensing image, a reflectivity value corresponding to a near infrared reflectivity maximum variance category by using a maximum inter-category variance method;

the dividing module 432 is configured to automatically divide the preprocessed high-spatial-resolution multispectral image into a water body, a shadow ground object type and a non-water body ground object type based on the reflectance value obtained from the calculating module 431, so as to obtain the water body, the shadow ground object type region and the non-water body ground object type region.

Furthermore, as an implementation of the method shown in fig. 3, the embodiment of the invention further provides another urban ground surface temperature downscaling device based on the high-resolution satellite image, which is used for implementing the method shown in fig. 3. The embodiment of the device corresponds to the embodiment of the method, and for convenience of reading, details of the embodiment of the method are not repeated one by one, but it should be clear that the device in the embodiment can correspondingly realize all the details of the embodiment of the method. As shown in fig. 6, the apparatus includes:

A first obtaining unit 51, configured to obtain random points corresponding to the water body and shadow feature type region and the non-water body feature type region respectively by using the model verification sample number;

a second obtaining unit 52, configured to obtain, based on the to-be-downscaled data, the surface temperature data of medium-low spatial resolution of the point location where the random point is located obtained from the first obtaining unit 51 by using a value extraction method, as an accuracy verification true value;

a third obtaining unit 53, configured to obtain, by using the ground surface temperature downscaling model, ground surface temperature data with high spatial resolution of a point location where the random point is located, where the data is obtained from the first obtaining unit 51;

a third calculation unit 54, configured to calculate, based on the medium-low spatial resolution surface temperature data of the point where the random point is located obtained from the second obtaining unit 52 and the high spatial resolution surface temperature data obtained from the third obtaining unit 53, scale-down accuracy of the surface temperature data of the water body and shadow feature type region and the non-water body feature type region respectively using a preset accuracy calculation formula; the preset precision calculation formula is as follows:

In the formula, delta is the accuracy of the down scale of the surface temperature data of the point where the random point is located, and LST _{High score} LST for high spatial resolution surface temperature data _{Middle and low score} Is the surface temperature data with medium-low spatial resolution;

and a fourth calculation unit 55, configured to calculate an average value of the scale-down accuracy of the surface temperature data of the water body and shadow feature type region and the non-water body feature type region obtained from the third calculation unit 54, so as to obtain an overall accuracy of the scale-down model of the target region.

Further, an embodiment of the present invention further provides a processor, where the processor is configured to execute a program, where the program executes the method for downscaling urban surface temperature based on high-resolution satellite images described in fig. 1-2.

Further, an embodiment of the present invention further provides a storage medium, where the storage medium is configured to store a computer program, where the computer program controls a device where the storage medium is located to execute the above-mentioned urban surface temperature downscaling method based on high-resolution satellite images in fig. 1-2.

In the foregoing embodiments, the descriptions of the embodiments are emphasized, and for parts of one embodiment that are not described in detail, reference may be made to related descriptions of other embodiments.

It will be appreciated that the relevant features of the methods and apparatus described above may be referenced to one another. In addition, the "first", "second", and the like in the above embodiments are for distinguishing the embodiments, and do not represent the merits and merits of the embodiments.

It will be clear to those skilled in the art that, for convenience and brevity of description, specific working procedures of the above-described systems, apparatuses and units may refer to corresponding procedures in the foregoing method embodiments, which are not repeated herein.

The algorithms and displays presented herein are not inherently related to any particular computer, virtual system, or other apparatus. Various general-purpose systems may also be used with the teachings herein. The required structure for a construction of such a system is apparent from the description above. In addition, the present invention is not directed to any particular programming language. It will be appreciated that the teachings of the present invention described herein may be implemented in a variety of programming languages, and the above description of specific languages is provided for disclosure of enablement and best mode of the present invention.

Furthermore, the memory may include volatile memory, random Access Memory (RAM) and/or nonvolatile memory, such as Read Only Memory (ROM) or flash memory (flash RAM), in a computer readable medium, the memory including at least one memory chip.

It will be appreciated by those skilled in the art that embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.

The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.

These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.

These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.

In one typical configuration, a computing device includes one or more processors (CPUs), input/output interfaces, network interfaces, and memory.

The memory may include volatile memory in a computer-readable medium, random Access Memory (RAM) and/or nonvolatile memory, etc., such as Read Only Memory (ROM) or flash RAM. Memory is an example of a computer-readable medium.

Computer readable media, including both non-transitory and non-transitory, removable and non-removable media, may implement information storage by any method or technology. The information may be computer readable instructions, data structures, modules of a program, or other data. Examples of storage media for a computer include, but are not limited to, phase change memory (PRAM), static Random Access Memory (SRAM), dynamic Random Access Memory (DRAM), other types of Random Access Memory (RAM), read Only Memory (ROM), electrically Erasable Programmable Read Only Memory (EEPROM), flash memory or other memory technology, compact disc read only memory (CD-ROM), digital Versatile Discs (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices, or any other non-transmission medium, which can be used to store information that can be accessed by a computing device. Computer-readable media, as defined herein, does not include transitory computer-readable media (transmission media), such as modulated data signals and carrier waves.

It should also be noted that the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one … …" does not exclude the presence of other like elements in a process, method, article or apparatus that comprises an element.

It will be appreciated by those skilled in the art that embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.

The foregoing is merely exemplary of the present application and is not intended to limit the present application. Various modifications and changes may be made to the present application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc. which are within the spirit and principles of the present application are intended to be included within the scope of the claims of the present application.

Claims

1. An urban earth surface temperature downscaling method based on high-resolution satellite images, which is characterized by comprising the following steps of:

LST _{high score} ＝a ₁ *a ₂ *NDVI _{High score} +a ₁ *b ₂ +b ₁ (1)

IST _{high score} ＝a ₁ *a ₂ *NDWI _{High score} +a ₁ *b ₂ +b ₁ (2)

2. The method according to claim 1, wherein constructing the surface temperature downscaling model of the water body and shadow feature type region and the non-water body feature type region respectively based on the to-be-downscaling data and the different spatial resolution downscaling parameter factors by using the number of sample points of the target region comprises:

Obtaining the number of sample points of each ground object type area of the target area; wherein the ground object type region comprises the water body and shadow ground object type region and the non-water body ground object type region;

dividing the number of the samples of each ground object type area into corresponding model training sample numbers and model verification sample numbers according to a preset proportion;

obtaining a first type of linear regression equation between the medium-low spatial resolution surface temperature data of each ground feature type region and the corresponding medium-low spatial resolution downscaling parameter factors and a second type of linear regression equation between different spatial resolution downscaling parameter factors by utilizing the number of the model training sample points;

and acquiring a ground surface temperature scale reduction model corresponding to each ground object type region based on the first type linear regression equation and the second type linear regression equation.

3. The method according to claim 2, wherein the training the number of sample points using the model to obtain a first type of linear regression equation between the medium-low spatial resolution surface temperature data and the corresponding medium-low spatial resolution downscaling parameter factors and a second type of linear regression equation between different spatial resolution downscaling parameter factors for each of the ground object type regions includes:

Obtaining random points corresponding to the water body and shadow ground object type areas and the non-water body ground object type areas by utilizing the number of points of the model training samples;

acquiring medium-low spatial resolution surface temperature data, medium-low spatial resolution normalized difference water body index NDWI and high spatial resolution normalized difference water body index NDWI of points where the random points of the water body and shadow ground object type areas are located, medium-low spatial resolution surface temperature data, medium-low spatial resolution normalized vegetation index NDVI and high spatial resolution normalized vegetation index NDVI of points where the random points of the non-water body ground object type areas are located;

based on the medium-low spatial resolution surface temperature data of the point positions of the random points of the water body and shadow ground object type region and the medium-low spatial resolution normalized difference water body index NDWI, constructing a first type linear regression equation of the water body and shadow ground object type region: LST (least squares) _{Middle and low score} ＝a ₁ *NDWI _{Middle and low score} +b ₁ In the LST _{Middle and low score} NDWI for medium and low spatial resolution surface temperature data _{Middle and low score} Normalizing the differential water index for medium and low spatial resolution, a ₁ And b ₁ Are all coefficients;

based on the medium-low spatial resolution surface temperature data of the point position where the random point of the non-water body ground feature type region is located and the medium-low spatial resolution normalized vegetation index NDVI, constructing a first type linear regression equation of the non-water body ground feature type region: LST (least squares) _{Middle and low score} ＝a ₁ *NDVI _{Middle and low score} +b ₁ In the LST _{Middle and low score} NDVI for medium and low spatial resolution surface temperature data _{Middle and low score} Normalizing vegetation index for medium and low spatial resolution, a ₁ And b ₁ Are all coefficients;

based on the normalized difference water body index NDWI of middle and low spatial resolutions and the normalized difference water body index NDWI of high spatial resolutions of the water body and shadow feature type region, constructing a second type linear regression equation of the water body and shadow feature type region: NDWI _{Middle and low score} ＝a ₂ *NDWI _{High score} +b ₂ In NDWI _{Middle and low score} NDWI for normalizing differential water index for medium and low spatial resolution _{High score} Normalizing differential water index for high spatial resolution, a ₂ And b ₂ Are all coefficients；

Based on the medium-low spatial resolution normalized vegetation index NDVI and the high spatial resolution normalized vegetation index NDVI of the non-water body ground object type region, constructing a second type linear regression equation of the non-water body ground object type region: NDVI _{Middle and low score} ＝a ₂ *NDVI _{High score} +b ₂ Wherein, NDVI _{Middle and low score} Normalized vegetation index for medium and low spatial resolution, NDVI _{High score} Normalizing vegetation index for high spatial resolution, a ₂ And b ₂ Are coefficients.

4. A method according to claim 3, wherein the obtaining the number of samples for each of the feature type areas of the target area comprises:

Determining the required duty ratio number by adopting a random point duty ratio principle according to the total number of grids of the medium-low spatial resolution remote sensing image of the target area, and calculating to obtain the random point selection total number;

and obtaining the number of the sample points of each ground object type area by utilizing the random point selection total number.

5. The method according to any one of claims 2-4, wherein after the constructing the surface temperature downscaling model corresponding to the water and shadow feature type region and the non-water feature type region respectively using the number of sample points of the target region based on the to-be-downscaled data and the different spatial resolution downscaling parameter factors, the method further comprises:

verifying the number of sample points by using the model, and acquiring random points respectively corresponding to the water body and shadow ground object type areas and the non-water body ground object type areas;

based on the to-be-downscaled data, acquiring the surface temperature data of middle and low spatial resolutions of the point where the random point is located by a value extraction method, and taking the surface temperature data as a precision verification true value;

inversion is carried out by utilizing the earth surface temperature scale-down model to obtain earth surface temperature data with high spatial resolution of the point position where the random point is located;

Based on the surface temperature data with medium and low spatial resolutions and the surface temperature data with high spatial resolutions of the point positions of the random points, respectively calculating the surface temperature data scale-reducing precision of the water body and shadow ground feature type areas and the non-water body ground feature type areas by using a preset precision calculation formula; the preset precision calculation formula is as follows:

and calculating the average value of the downscaling precision of the surface temperature data of the water body and shadow ground object type areas and the non-water body ground object type areas to obtain the overall precision of the downscaling model of the target area.

6. The method of claim 1, wherein the acquiring the multi-source remote sensing data of the target area and processing the multi-source remote sensing data using the preprocessing rule to obtain the preprocessed multi-source remote sensing data comprises:

acquiring the middle-low spatial resolution remote sensing image of the target region in a preset time period, wherein the middle-low spatial resolution remote sensing image comprises a thermal infrared remote sensing image and a multispectral remote sensing image;

Acquiring the high-spatial-resolution multispectral remote sensing image of the target area according to a preset time difference based on the preset time period corresponding to the medium-low spatial-resolution remote sensing image;

performing pretreatment of radiation calibration, atmosphere correction, resampling and coordinate system conversion on the middle-low spatial resolution remote sensing image to obtain a pretreated middle-low spatial resolution thermal infrared remote sensing image and a pretreated middle-low spatial resolution multispectral image;

performing pretreatment of radiation calibration, atmosphere correction, resampling and coordinate system conversion on the high-spatial-resolution multispectral remote sensing image to obtain a pretreated high-spatial-resolution multispectral image; the resampling of the high-spatial-resolution multispectral image is based on the principle of keeping integer multiple relation with the spatial resolution of the medium-low spatial-resolution remote sensing image.

7. The method according to claim 6, wherein the automatically dividing the two types of ground objects of the high-spatial-resolution multispectral remote sensing image by using a maximum inter-class variance method based on the preprocessed high-spatial-resolution multispectral remote sensing image to obtain a water body and shadow ground object type area and a non-water body ground object type area corresponding to the target area, comprises:

Calculating a reflectivity value corresponding to a near infrared reflectivity maximum variance category by adopting a maximum inter-category variance method based on the preprocessed high spatial resolution multispectral remote sensing image;

and based on the reflectivity value, automatically dividing the pretreated high-spatial-resolution multispectral image into water body, shadow ground object types and non-water body ground object types, and obtaining the water body, shadow ground object type areas and non-water body ground object type areas.

8. Urban earth surface temperature scale reduction device based on high-resolution satellite images, which is characterized by comprising:

the construction unit is used for constructing an earth surface temperature downscaling model corresponding to the water body and shadow ground feature type area and the non-water body ground feature type area respectively by utilizing the number of sample points of the target area based on the to-be-downscaling data and the different spatial resolution downscaling parameter factors; wherein,

LST _{high score} ＝a ₁ *a ₂ *NDVI _{High score} +a ₁ *b ₂ +b ₁ (1)

LST _{high score} ＝a ₁ *a ₂ *NDWI _{High score} +a ₁ *b ₂ +b ₁ (2)

In formulas (1) - (2), LST _{High score} Is a high spatial divisionResolution surface temperature data, NDVI _{High score} Normalizing vegetation index for high spatial resolution, NDWI _{High score} Normalizing differential water index for high spatial resolution, a ₁ 、b ₁ 、a ₂ And b ₂ Are all coefficients;

9. A storage medium comprising a stored program, wherein the program, when run, controls a device in which the storage medium is located to perform the urban surface temperature downscaling method based on high resolution satellite imagery according to any one of claims 1 to 7.

10. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor implements the high resolution satellite image based urban surface temperature downscaling method of any one of claims 1 to 7 when the program is executed.