Surface temperature inversion method and equipment based on remote sensing satellite data and medium

Abstract

The embodiment of the invention discloses a surface temperature inversion method and device based on remote sensing satellite data, and a medium, and relates to the technical field of remote sensing, and the method comprises the steps: receiving a heat island effect analysis demand triggered by a user, obtaining satellite remote sensing data of a target region and a corresponding metadata file based on the heat island effect analysis demand, and storing the satellite remote sensing data in the target region; processing the satellite remote sensing data to determine earth surface radiation parameters; the atmospheric features and the earth surface features of the target area are evaluated through the metadata file and a pre-obtained normalized vegetation index, a dynamic adjustment factor corresponding to the target area is determined, a preset single-window algorithm model is optimized through the dynamic adjustment factor, and an optimized single-window algorithm model is determined; and inputting the surface emissivity, the atmospheric transmittance, the brightness temperature value and a pre-acquired atmospheric average action temperature into an optimized single window algorithm model for calculation, and determining a surface temperature distribution diagram of the target area for heat island effect analysis.

Description

Technical Field

[0001] This specification relates to the field of remote sensing technology, and in particular to a method, equipment and medium for retrieving land surface temperature based on remote sensing satellite data. Background Technology

[0002] Land surface temperature is a key parameter for studying global climate change, urban thermal environment, and surface energy balance. Traditional methods for observing land surface temperature, such as glass liquid thermometers and platinum resistance thermometry, are single-point, local measurements. While these methods offer high accuracy, the data they acquire is spatially discontinuous, making it difficult to directly apply to regional-scale studies. Extending discrete-point data to a regional scale using spatial interpolation methods not only introduces significant accuracy errors but also fails to accurately reflect the spatial heterogeneity of land surface temperature, making it unsuitable for large-scale environmental monitoring.

[0003] With the development of remote sensing technology, using satellite data to retrieve land surface temperature has become an effective means of obtaining regional-scale temperature information. Among these, the Landsat series satellites, especially the thermal infrared sensor carried by Landsat-8, are widely used due to their good data continuity and high spatial resolution. Currently, land surface temperature retrieval methods based on this type of data generally employ single-window or split-window algorithms based on physical models. However, these existing methods still have significant limitations in practical implementation. First, the calculation of key parameters in the algorithms (such as surface emissivity and atmospheric transmittance) is often oversimplified. For example, surface emissivity is often set to a fixed value or estimated only through a single index, failing to fully consider the influence of mixed pixels under complex land cover, leading to inaccurate parameter characterization. Second, the retrieval algorithms themselves usually adopt a fixed form, and their model parameters are not dynamically adjusted according to the atmospheric conditions and surface characteristics of different regions. This results in poor universality and unstable retrieval accuracy when applied to different geographical environments and seasons.

[0004] Therefore, in the process of detailed analysis of the urban heat island effect, when using existing remote sensing technology to retrieve regional surface temperature, there are problems with limited accuracy and reliability of the retrieval results due to inaccurate calculation of key parameters and insufficient adaptability of the retrieval algorithm, which leads to deviations in the final heat island effect analysis conclusions. Summary of the Invention

[0005] This specification provides one or more embodiments of a method, device, and medium for land surface temperature inversion based on remote sensing satellite data, which is used to solve the following technical problem: In the process of detailed analysis of urban heat island effect, when using existing remote sensing technology to invert regional land surface temperature, due to the inaccurate calculation of key parameters and the insufficient adaptability of the inversion algorithm, there are problems with limited accuracy and reliability of the inversion results, which leads to deviations in the final heat island effect analysis conclusions.

[0006] One or more embodiments of this specification employ the following technical solutions: This specification provides one or more embodiments of a method for retrieving land surface temperature based on remote sensing satellite data. The method includes: receiving a user-triggered request for urban heat island effect analysis; acquiring satellite remote sensing data and corresponding metadata files for a target area based on the request; processing the satellite remote sensing data to determine land surface radiation parameters, wherein the land surface radiation parameters include land surface emissivity, atmospheric transmittance, and brightness temperature; evaluating the atmospheric and land surface characteristics of the target area using the metadata files and pre-acquired normalized vegetation index to determine a dynamic adjustment factor corresponding to the target area; optimizing a preset single-window algorithm model using the dynamic adjustment factor to determine an optimized single-window algorithm model; and inputting the land surface emissivity, atmospheric transmittance, brightness temperature, and pre-acquired average atmospheric temperature into the optimized single-window algorithm model for calculation to determine a land surface temperature distribution map of the target area for urban heat island effect analysis.

[0007] This specification provides one or more embodiments of a surface temperature retrieval device based on remote sensing satellite data, comprising: At least one processor; and, A memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, which, when executed by the at least one processor, enables the at least one processor to perform the above-described method.

[0008] This specification provides one or more embodiments of a non-volatile computer storage medium storing computer-executable instructions configured to perform the above-described method.

[0009] The above-mentioned at least one technical solution adopted in the embodiments of this specification can achieve the following beneficial effects: By integrating the geographic spatiotemporal information in the metadata with the surface feature index obtained by remote sensing inversion, a dynamic adjustment factor that can perceive the regional environmental characteristics in real time is constructed. The macroscopic atmospheric background (such as the atmospheric profile characteristics that change with latitude and season) and the microscopic surface attributes (such as NDVI reflecting the vegetation cover) are coupled and analyzed, so that the single-window algorithm model of the inversion algorithm has the ability to self-adjust and optimize, and can automatically adapt to the inversion needs of different underlying surface types and different seasonal climate conditions, such as from arid areas to humid areas, from urban core areas to suburbs, and fundamentally overcome the drawbacks of the traditional static model's one-size-fits-all approach. Secondly, by treating the optimized single-window algorithm model as a whole computational unit, the accurately calculated surface emissivity, atmospheric transmittance, brightness temperature, and average atmospheric temperature are processed collaboratively. The parameters within this optimized model are not simply linearly superimposed, but constitute a highly nonlinear functional relationship based on the physical process of thermal infrared radiation transmission. The surface emissivity accurately characterizes the emissivity of the ground object itself, the atmospheric transmittance quantifies the attenuation effect of the atmosphere on thermal infrared radiation, the brightness temperature reflects the original radiation signal received by the sensor, and the average atmospheric temperature represents the contribution of downward atmospheric radiation. The introduction of a dynamic adjustment factor allows the model to fine-tune the calculation results according to the real-time environmental context, thereby significantly reducing the fundamental errors caused by the rigidity of the model structure. Ultimately, the entire chain of technological innovation, from data preprocessing and precise parameter calculation to dynamic algorithm optimization, ensures that the generated surface temperature distribution map not only has higher absolute accuracy but also truly reflects the spatial details and spatiotemporal evolution of the surface thermal field. This provides an unprecedented data quality foundation for the analysis of the urban heat island effect, enabling researchers to more accurately identify the spatial pattern of the heat island, quantify its intensity, and analyze its driving factors, greatly enhancing the practicality of this technical solution in operational applications. Attached Figure Description

[0010] To more clearly illustrate the technical solutions in the embodiments or prior art of this specification, the drawings used in the description of the embodiments or prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this specification. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. In the drawings: Figure 1 A flowchart illustrating a method for retrieving land surface temperature based on remote sensing satellite data, provided as an embodiment of this specification; Figure 2 This is a schematic diagram of a surface temperature inversion device based on remote sensing satellite data, provided as an embodiment of this specification. Detailed Implementation

[0011] To enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this specification, and not all embodiments. Based on the embodiments of this specification, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of this specification.

[0012] This specification provides a method for retrieving land surface temperature based on remote sensing satellite data. It should be noted that the execution entity in this specification can be a server or any device with data processing capabilities. Figure 1 A flowchart illustrating a method for retrieving land surface temperature based on remote sensing satellite data, as provided in this specification, is shown below. Figure 1 As shown, the main steps include the following: Step S101: Receive the user-triggered heat island effect analysis request, and based on the heat island effect analysis request, obtain satellite remote sensing data and corresponding metadata files of the target area, process the satellite remote sensing data, and determine the surface radiation parameters.

[0013] Among them, the surface radiation parameters include surface emissivity, atmospheric transmittance, and brightness temperature. In one embodiment of this specification, a user-triggered request for heat island effect analysis of a specific geographical area is received through a human-computer interaction interface. First, the spatial boundary information of the target area input by the user is standardized. This spatial boundary information is usually provided in the form of administrative boundary coordinates or custom polygon geographic coordinates. Then, it is spatiotemporally matched with pre-stored satellite orbit parameters to determine the optimal imaging time period and specific path row number of the satellite remote sensing data product to be called.

[0014] Next, it automatically connects to a remote geospatial data cloud platform or satellite data archive center through the built-in data access interface, and initiates a search request for Landsat-8 satellite OLI_TIRS sensor digital products by matching the calculated spatiotemporal parameters as query conditions. The request will clearly specify key filtering indicators such as the product level of the required data, cloud cover threshold, and imaging date range. After a dataset that meets the criteria is retrieved, the data download process will be automatically triggered. This will not only acquire the master data file of the satellite image containing multiple spectral bands, but also simultaneously download the corresponding metadata file. The metadata file is a text file called _MTL.txt, which systematically records a series of parameters that are crucial to the subsequent physical inversion algorithm, such as the radiometric calibration parameters of the image (including the gain value RADIANCE_MULT_BAND_x and offset RADIANCE_ADD_BAND_x for each band), brightness temperature inversion constants (K1_CONSTANT_BAND_10 and K2_CONSTANT_BAND_10), the precise timestamp of image acquisition, solar elevation angle, satellite azimuth angle, and geographic coordinates of the center point of the image coverage area.

[0015] All downloaded raw data is automatically transferred to the designated computing storage space and organized and archived according to preset naming rules. At the same time, the integrity of the data and the completeness of the metadata fields are verified to ensure that subsequent processing steps can seamlessly read and correctly parse this basic data. This complete data acquisition and preparation process provides an indispensable data input foundation for a series of automated processing steps such as radiometric calibration, atmospheric correction, parameter calculation and temperature inversion. This ensures that the entire technical solution, from user demand triggering to the final generation of heat island effect analysis results, can be successfully launched and reliably executed.

[0016] The satellite remote sensing data is processed to determine the surface radiation parameters, specifically including: performing radiometric calibration on the satellite remote sensing data to obtain the radiance value and surface reflectance data of the target area; calculating the normalized difference vegetation index and vegetation cover of the target area based on the surface reflectance data, and calculating the surface emissivity of the target area based on the normalized difference vegetation index and vegetation cover; calculating the atmospheric transmittance of the target area based on the imaging time and geographical location information in the metadata file; and calculating the brightness temperature value of the target area at the satellite sensor altitude using the radiance value and the brightness temperature inversion constant obtained from the metadata file.

[0017] In one embodiment of this specification, after data acquisition, the downloaded Landsat-8 OLI_TIRS satellite raw digital product undergoes radiometric calibration, converting the raw dimensionless digital quantization values ​​recorded by the sensor into radiometric parameters with explicit physical meaning. First, the _MTL.txt metadata file, which is strictly matched with the image, is automatically parsed to precisely extract the radiometric calibration gain coefficient and offset parameters corresponding to each band. These parameters are explicitly identified in the metadata file by the tags RADIANCE_MULT_BAND_x and RADIANCE_ADD_BAND_x, where x represents a specific band number. Next, based on these parameters, radiometric calibration calculations are performed on the raw DN value of each pixel in the image. This involves multiplying the DN value of each pixel by the corresponding gain coefficient and adding the offset to obtain the radiance value of that pixel at the top of the atmosphere. This value accurately represents the spectral radiation energy received at the sensor's entrance pupil, and its unit is W·m². -2 ·sr -1 ·μm -1 .

[0018] While performing radiance calculations, atmospheric correction is further executed to obtain accurate surface reflectance data, achieved by invoking the integrated 6S atmospheric radiative transfer model. The apparent reflectance image obtained after radiocalibration, along with the precise imaging time, solar zenith angle, satellite observation angle, and geographic coordinates of the image center point recorded in the metadata file, are automatically input into the 6S model as core parameters. Based on the imaging time and geographic location, the model automatically matches or allows professionals to manually set the atmospheric mode and aerosol type best suited to the actual conditions of the day, using a built-in lookup table. For example, for continental regions, the continental aerosol mode is selected by default, and initial aerosol optical thickness parameters are set based on meteorological auxiliary data or visibility observation information. Furthermore, the pixel-by-pixel aerosol optical thickness inversion function is enabled to refine the radiative distortion caused by atmospheric scattering and absorption. After the complex radiative transfer calculations of the 6S model, atmospherically corrected true surface reflectance data is output. This data effectively eliminates the influence of atmospheric effects such as Rayleigh scattering, aerosol scattering, and water vapor absorption, more accurately reflecting the spectral characteristics of the Earth's surface.

[0019] Finally, the radiance data generated by radiometric calibration and the surface reflectance data generated after atmospheric correction are recombined and output according to a unified spatial reference and pixel size to form an intermediate data product with spatial registration and clear physical meaning. This provides an accurate and reliable data foundation for subsequent calculation of key inversion parameters such as normalized vegetation index, vegetation cover, surface emissivity, and brightness temperature.

[0020] Based on the surface reflectance data, the normalized vegetation index (NVI) and vegetation cover of the target area are calculated. Specifically, this includes: extracting near-infrared and red band reflectance values ​​from the surface reflectance data; calculating the NVI value for each pixel using the NVI calculation formula based on the near-infrared and red band reflectance values; statistically analyzing the NVI values ​​of all pixels within the target area to determine the cumulative distribution statistics of the NVI; determining the minimum and maximum values ​​of the NVI based on the cumulative distribution statistics; and calculating the vegetation cover of each pixel using the vegetation cover calculation formula based on the NVI value, the minimum value, and the maximum value.

[0021] In one embodiment of this specification, after obtaining atmospherically corrected surface reflectance data, a calculation process for the normalized vegetation index and vegetation cover is executed. First, reflectance values ​​in the near-infrared and red bands are accurately extracted from the multi-band surface reflectance data. The near-infrared band corresponds to the fifth band of the Landsat-8 OLI sensor, and the red band corresponds to the fourth band. The spectral response characteristics of these two bands are highly sensitive to vegetation pigment content and cell structure. By reading the reflectance dataset with a unified geocoding generated in the preprocessing stage, a band extraction algorithm is used to automatically locate and load the reflectance raster layers for these two specific bands, ensuring that each pixel at a spatial location has complete dual-band reflectance information.

[0022] After data preparation is completed, the normalized vegetation index is calculated for each pixel. This calculation follows a standardized formula, which is to divide the difference between the near-infrared band reflectance value and the red band reflectance value by their sum. This normalization process can effectively enhance the vegetation signal while suppressing the influence of environmental factors such as light conditions and terrain shadows, so that the calculation results can stably reflect the physiological state and spatial distribution characteristics of vegetation.

[0023] After completing the full-scene NDVI calculation, a global statistical analysis is performed on the NDVI values ​​of all valid pixels within the target area, generating a statistical feature report describing the spatial distribution of NDVI values, including the numerical range, frequency distribution histogram, and cumulative distribution function curve. Based on this cumulative distribution statistic, a non-parametric statistical method is used to automatically determine the confidence interval boundary values ​​of NDVI. Specifically, the NDVI values ​​corresponding to the cumulative percentage reaching the low quantile and high quantile thresholds are located and defined as the minimum and maximum effective values ​​of the NDVI statistical distribution in that area, respectively. This method effectively eliminates the interference of abnormal pixels on the statistical results. Finally, based on the NDVI value of each pixel and the statistically obtained global minimum and maximum values, pixel-level calculations are performed using the vegetation cover calculation formula. This formula calculates the difference between the current pixel's NDVI value and the global minimum value, then divides it by the difference between the global maximum and minimum values ​​to obtain a continuous numerical value representing the vegetation cover ratio. The calculation result is normalized to a numerical range between zero and one, fully representing the continuous change from completely no vegetation cover to completely covered vegetation. The final vegetation cover raster data and the intermediate NDVI raster data are stored together in a designated data warehouse, and standardized inputs are provided for the subsequent surface emissivity calculation module. These refined parameters will provide crucial surface feature parameter support for the subsequent temperature inversion algorithm.

[0024] Traditional methods typically use fixed thresholds or empirical formulas to estimate vegetation cover, failing to fully consider the spatial heterogeneity of vegetation distribution and phenological variations across different regions and seasons, leading to systematic biases in parameter calculations. Compared to conventional methods, this technical solution introduces a dynamic statistical analysis method based on the cumulative distribution function, adaptively determining the effective range of NDVI values ​​for each specific scenario. This data-driven approach effectively avoids the problem of overestimating or underestimating vegetation cover in local areas due to the use of a globally fixed threshold. Simultaneously, by establishing a complete processing chain, the process from raw reflectance data to the final vegetation cover product is systematized and automated, ensuring standardized data processing and repeatable results. Since the calculation of surface emissivity highly depends on accurate vegetation cover information, this technical solution indirectly enhances the technical robustness of the entire surface temperature retrieval chain by improving the reliability of vegetation cover calculations. It can automatically adapt to vegetation change characteristics under different geographical environments and seasonal conditions, providing reliable technical support for large-scale, long-term remote sensing monitoring applications.

[0025] The surface emissivity of the target area is calculated based on the normalized vegetation index (NVI) and the vegetation cover. Specifically, this includes: calculating the initial emissivity value of each pixel using a predefined piecewise function based on the NVI value of each pixel; identifying mixed pixels in the target area based on the vegetation cover; for the identified mixed pixels, weighting the initial emissivity of the vegetation component and the initial emissivity of the non-vegetation component using the corresponding vegetation cover value to obtain the surface emissivity of the mixed pixel; and for non-mixed pixels, determining the surface emissivity based on the initial emissivity value.

[0026] In one embodiment of this specification, a predefined piecewise function calculation model is first invoked. This model is built into the system's parameterized algorithm library, and its functional relationship is a physical empirical model constructed based on the typical emissivity characteristics exhibited by different land cover types in different vegetation index intervals. During calculation, the NDVI value of each pixel is read, and the corresponding calculation path is automatically selected according to the specific interval in which the value is located: for pixels with extremely low NDVI values ​​and characterized by typical non-vegetation features, a fixed emissivity value representing bare soil or artificial building material surface type is directly assigned to them; for pixels with NDVI values ​​in the typical vegetation growth interval, a multivariate regression formula preset for that interval is used for calculation, which fits the emissivity variation law of the vegetation canopy through a polynomial combination of NDVI values; for pixels with extremely high NDVI values ​​and characterized by dense vegetation, another set of calculation coefficients specifically optimized for high-density vegetation is used to ensure accurate estimation of the emissivity of the dense vegetation canopy.

[0027] After calculating the initial emissivity values ​​for all pixels, the identification and processing of mixed pixels is performed. The pre-calculated vegetation cover value for each pixel is read, and a judgment is made based on a set threshold range. Pixels with vegetation cover that is neither extremely high nor extremely low—that is, where the surface is composed of both vegetation and non-vegetation components—are identified as mixed pixels. For these identified mixed pixels, based on the principle of a linear fusion model, the previously calculated emissivity value of the pixel under pure vegetation conditions is used as the vegetation component contribution. Simultaneously, the contribution of non-vegetation components is obtained by querying the system's built-in typical non-vegetation surface emissivity lookup table. Then, using the pixel's own vegetation cover as a weight, the emissivity of the vegetation and non-vegetation components is weighted and summed to obtain the final emissivity value that truly reflects the internal compositional relationship of the mixed pixel. For areas identified as non-mixed pixels, including pure vegetation pixels and pure non-vegetation pixels, the initial emissivity value calculated in the first stage is directly used as the final result, without the need for weighted fusion.

[0028] After the entire calculation process is completed, a surface emissivity raster map with the same spatial resolution as the input data is generated. This not only accurately characterizes the emissivity of each pixel in the geothermal infrared band, but also significantly improves the estimation accuracy of emissivity parameters in complex underlying surface areas by differentiating between pure pixels and mixed pixels. This ensures the scientific validity and reliability of the final heat island effect analysis results from the source.

[0029] Compared with conventional methods that use single, fixed empirical values ​​or simple linear relationships to estimate surface emissivity, this technical solution introduces a dynamic piecewise function based on NDVI and a hybrid pixel decomposition model driven by vegetation cover to achieve a more refined estimation of surface emissivity. The piecewise function model can more accurately capture the nonlinear variation of emissivity during the continuous change from bare soil to dense vegetation, rather than simply treating it as a linear relationship. By integrating hybrid pixel decomposition technology, vegetation and non-vegetation components within a pixel are distinguished through rigorous weighted calculations, which fundamentally solves the shortcomings of traditional methods in characterizing the surface thermal radiation characteristics of complex mosaic landscapes.

[0030] Based on the imaging time and geographic location information in the metadata file, the atmospheric transmittance of the target area is calculated. Specifically, this includes: simulating the first atmospheric transmittance using a preset atmospheric radiative transfer model, inputting the imaging time and geographic location information; acquiring atmospheric water vapor content product data corresponding to the target area and imaging time, and inverting to obtain the second atmospheric transmittance; acquiring meteorological station observation data within the target area, and calculating the third atmospheric transmittance; determining corresponding weighting coefficients based on the spatial and temporal resolutions of the first, second, and third atmospheric transmittances; and using these weighting coefficients to perform a weighted average calculation of the first, second, and third atmospheric transmittances to obtain the fused atmospheric transmittance.

[0031] In one embodiment of this specification, a preset atmospheric radiative transfer model is initiated. This model uses mature radiative transfer codes such as MODTRAN or 6S, which have been validated over a long period, as the core of the calculation. The system automatically extracts the precise imaging timestamp and center point geographic coordinates of the image from the metadata file and uses these parameters as the basic inputs to the atmospheric model. Simultaneously, by accessing the global reanalysis database, atmospheric state parameters such as atmospheric temperature and humidity vertical profiles and pressure fields that match the imaging time are obtained. These data are combined with the imaging geographic location information to construct a vertical profile model representing the atmospheric conditions at that time and place. After completing the configuration of all input parameters, atmospheric radiative transfer simulation calculations are performed. By solving the transmission equation of electromagnetic waves in the atmosphere under a specific thermal infrared band, the transmittance of the entire atmospheric layer to thermal infrared radiation under the satellite observation geometry is obtained and determined as the first atmospheric transmittance. Its characteristics are that it has a clear physical mechanism and strictly corresponds to the atmospheric conditions at the time of imaging.

[0032] To compensate for potential biases in single-model simulations, a multi-source data fusion process was initiated in parallel. Atmospheric water vapor content raster product data, strictly matched to the target area and imaging time, was automatically acquired via a data interface. These products typically originate from inversion results from satellites such as MODIS or AMSR. Based on this water vapor content data, a second atmospheric transmittance was derived using an empirical relationship between water vapor absorption bands and thermal infrared transmittance. The advantage of this result is that it directly originates from satellite observations and reflects the spatial distribution characteristics of water vapor. Simultaneously, an IoT interface was used to access the observation network of meteorological stations within and around the target area to acquire surface meteorological elements such as temperature, relative humidity, and air pressure recorded by each station at the imaging time. Atmospheric transmittance at each station location was estimated using empirical formulas based on this surface data. Then, a third atmospheric transmittance field covering the entire target area was generated using spatial interpolation. The advantage of this result is that it includes actual ground observation information and has high temporal matching accuracy.

[0033] After obtaining three transmittance data with different characteristics, an objective evaluation is conducted based on the inherent attributes of each data source. Specifically, the temporal resolution of the reanalysis data on which the first atmospheric transmittance depends is analyzed, the spatial resolution of the water vapor content product corresponding to the second atmospheric transmittance is analyzed, and the effective meteorological station density used to generate the third atmospheric transmittance is statistically analyzed. Then, based on the principle that the higher the resolution, the greater the weight, the corresponding weight coefficients are assigned to the three transmittance results, and the sum of the three weights is 1, to ensure that the data source with rich spatial details and accurate temporal matching plays a greater role in the fusion process.

[0034] Specifically, the first weighting coefficient corresponding to the first atmospheric transmittance is: The second weighting coefficient corresponding to the second atmospheric transmittance is The second weighting coefficient corresponding to the second atmospheric transmittance is R1 is the temporal resolution (hours) of the reanalysis meteorological data used by the atmospheric radiative transfer model, obtained from the metadata of the reanalysis meteorological data; R2 is the spatial resolution (kilometers) of the atmospheric water vapor content product, obtained from the metadata of the atmospheric water vapor content product; N3 is the number of effective meteorological stations in the target area, obtained by statistically analyzing the meteorological station observation data after performing an integrity check.

[0035] Finally, a weighted average fusion calculation is performed to resample the three transmittance data to a unified spatial grid. The data are then summed at the pixel level according to the determined weight coefficients to generate the final atmospheric transmittance product. This product retains the mechanism of the physical model while incorporating the spatial details of satellite observations and the accuracy of ground measurements, providing reliable atmospheric parameter input for subsequent temperature inversion.

[0036] Compared to conventional methods that rely solely on model simulation or simply use fixed values ​​for atmospheric transmittance calculation, this technical solution innovatively integrates transmittance data from three different sources and establishes an intelligent weight allocation mechanism based on the inherent characteristics of the data. This effectively overcomes the limitations of a single data source, fully utilizes the theoretical completeness of the physical model, the spatial continuity of satellite products, and the temporal accuracy of ground observations, resulting in an optimal balance between the spatiotemporal representativeness of the final atmospheric transmittance product. By objectively determining the weight coefficients through analysis of the spatial and temporal resolutions of each data source, the uncertainty of subjective empirical assignments is avoided, making the fusion process more scientific and the results more reliable. By providing more accurate and reliable atmospheric transmittance data, the temperature inversion error caused by insufficient atmospheric correction is significantly reduced, enabling the acquisition of high-precision surface temperature products even under complex atmospheric conditions.

[0037] Step S102: Using metadata files and pre-acquired normalized vegetation index, evaluate the atmospheric and surface characteristics of the target area, determine the dynamic adjustment factor corresponding to the target area, and optimize the preset single-window algorithm model using the dynamic adjustment factor to determine the optimized single-window algorithm model.

[0038] Using the metadata file and pre-acquired normalized vegetation index (NWDI), the atmospheric and surface features of the target area are evaluated to determine the corresponding dynamic adjustment factor. Specifically, this includes: extracting the center latitude information and imaging time month information of the target area from the metadata file to calculate the atmospheric condition adjustment component based on the center latitude information and month information; calculating the average NWDI of all pixels in the target area to calculate the surface feature adjustment component based on the average NWDI; and calculating the dynamic adjustment factor by weighted fusion of the atmospheric condition adjustment component and the surface feature adjustment component.

[0039] In one embodiment of this specification, based on the calculation of surface and atmospheric parameters, the system extracts the center latitude information of the target area from the metadata file. This information accurately records the geographic coordinates of the geometric center point of the image coverage area in degrees. Simultaneously, the system extracts the month information of the imaging time, which accurately reflects the specific month in which the data was acquired. Based on these two fundamental parameters, the system calculates the atmospheric condition adjustment component using the following formula. Where M represents the month and φ represents the center latitude, the design principle of this formula is to use trigonometric functions to construct a comprehensive index that can simultaneously respond to seasonal cycle changes and latitudinal zone changes. The sine function term is used to capture the seasonal fluctuation characteristics of atmospheric conditions in different months, while the cosine function term is used to reflect the gradient change law of atmospheric characteristics in different latitudinal zones. The two are multiplied and then normalized to ensure that the output value is within the standard range of 0 to 1.

[0040] Simultaneously, the surface feature adjustment component is calculated in parallel. First, the normalized vegetation index (NDVI) values ​​of all effective pixels within the target area are statistically averaged. The calculated average NDVI value (NDVI_avg) serves as a comprehensive indicator characterizing the overall land cover status of the area. This indicator effectively reflects the density and spatial distribution characteristics of vegetation within the region. For densely vegetated areas, the average NDVI value is higher, indicating relatively uniform land cover dominated by natural vegetation. Conversely, for densely urbanized areas, the average NDVI value is lower, indicating that the land cover is dominated by artificial building materials and exhibits strong spatial heterogeneity. The calculated average NDVI value is directly used as the surface feature adjustment component, i.e. This design is based on the close relationship between vegetation cover and surface thermal radiation characteristics. Dense vegetation usually has a high emissivity and relatively stable thermal inertia, while artificial surfaces exhibit different thermal radiation characteristics.

[0041] After obtaining the atmospheric condition adjustment component and the surface feature adjustment component, which respectively characterize atmospheric and surface features, the final dynamic adjustment factor is calculated using a weighted fusion formula: α_min and α_max represent the preset upper and lower boundary values ​​of the dynamic adjustment factor, respectively, with values ​​of 0.8 and 1.2. λ is the balance weight coefficient, which is used to adjust the relative contribution of atmospheric conditions and surface features to the final result, and can be set according to the prior knowledge of different regions. α_min and α_max are preset fixed values, taking values ​​of 0.8 and 1.2 respectively, serving as the lower and upper limits of the dynamic adjustment factor. In the calculation formula for F_atm (atmospheric condition adjustment component), M represents the imaging month (ranging from 1 to 12), and φ represents the center latitude (in degrees). Through trigonometric function combination and normalization, the value of F_atm is constrained to the range [0,1]. Simultaneously, F_surf (surface feature adjustment component) is directly taken as the average normalized vegetation index (NDVI_avg) of all pixels in the target area. While the theoretical range of NDVI is [-1,1], its values ​​in actual surface applications are mostly concentrated in the [0,1] interval. Furthermore, NDVI_avg, as a statistical average, also falls within the [0,1] range. In the weighted fusion part, λ is the balancing weight coefficient, ranging from [0,1], ensuring that the weighted sum is also within [0,1]. Therefore, the final calculation result of the dynamic adjustment factor falls within the [0.8, 1.2] interval. It should be noted that the weighting coefficients are set with appropriate weighting schemes according to specific application scenarios and regional characteristics. For example, in areas with drastic changes in vegetation cover, the weight of the surface feature adjustment component can be appropriately increased, while in areas with complex atmospheric conditions, the weight of the atmospheric condition adjustment component can be increased accordingly.

[0042] The entire calculation process fully considers atmospheric conditions and surface characteristics that affect the accuracy of land surface temperature inversion. By establishing quantitative relationships with specific geographic spatiotemporal parameters and remote sensing observations, the dynamic adjustment factor can adaptively respond to the inversion needs under different regions, seasons, and land cover conditions. This provides scientific and reasonable adjustment parameters for subsequent single-window algorithm optimization, effectively improving the adaptability and stability of the land surface temperature inversion model in different application scenarios.

[0043] This technical solution establishes a dynamic adjustment mechanism that can adapt to different geographical environments and imaging conditions by integrating both atmospheric conditions and surface features. It comprehensively considers the main factors affecting the accuracy of temperature inversion, including atmospheric condition information reflecting regional climate background and seasonal changes, as well as vegetation index information characterizing surface cover characteristics. This allows the adjustment factors to more accurately reflect the comprehensive environmental characteristics of a specific region at a specific time period, giving the temperature inversion algorithm stronger environmental adaptability and regional specificity. It can effectively meet the temperature inversion needs under different underlying surface conditions, such as from arid to humid regions, and from urban built-up areas to natural vegetation areas.

[0044] The optimized single-window algorithm model is determined by optimizing the preset single-window algorithm model using the dynamic adjustment factor. Specifically, this includes: obtaining a pre-determined error compensation factor; determining model optimization parameters based on the dynamic adjustment factor and the error compensation factor; and optimizing the single-window algorithm model using these optimization parameters. Obtaining the pre-determined error compensation factor specifically includes: calculating the spatial distribution standard deviation of brightness temperature values ​​within the target region; calculating an initial compensation value based on the spatial distribution standard deviation using a preset mapping function; and applying range constraints to the initial compensation value to ensure it falls within the range of the preset compensation factor value, using the constrained compensation value to determine the error compensation factor.

[0045] In one embodiment of this specification, after obtaining the dynamic adjustment factor, the adaptability and accuracy of the basic algorithm are improved by introducing an error compensation mechanism and a dynamic adjustment mechanism.

[0046] First, the error compensation factor calculation process is initiated. Based on statistical analysis of the spatial distribution characteristics of thermal infrared brightness temperature values ​​within the target area, the calculated brightness temperature raster data for the entire target area is extracted from the preprocessed data. This data represents the equivalent blackbody temperature after Planck's formula transformation of the radiance received by the satellite sensor in the thermal infrared band. Next, spatial statistical analysis is performed to calculate the spatial distribution standard deviation of brightness temperature values ​​for all effective pixels within the target area. This statistic reflects the heterogeneity and dispersion of the surface thermal field in the spatial dimension, and its calculation formula is as follows: , where σ represents the spatial distribution standard deviation of the brightness temperature value, T_i represents the brightness temperature value of the i-th pixel, μ represents the arithmetic mean of the brightness temperature values ​​of all pixels in the target area, and N is the total number of effective pixels.

[0047] After obtaining the spatial distribution standard deviation, the initial compensation value is calculated using a preset mapping function. The mapping function adopts the hyperbolic tangent function form, and its specific expression is as follows: ,in `_init` represents the initial compensation value, `k` is the scaling factor used to control the compensation amplitude, `σ` is the calculated spatial distribution standard deviation, `S` is the normalization coefficient used to adjust the sensitivity of the function to the standard deviation, and `tanh` is the hyperbolic tangent function that smoothly maps the input value to a fixed interval. This function design ensures that the compensation factor is close to the baseline value in regions with relatively homogeneous brightness temperature spatial distribution, while producing an appropriate compensation effect in regions with strong brightness temperature spatial variation.

[0048] Subsequently, the initial compensation value is subjected to range constraints. Conditional judgments and linear pruning are used to ensure that the compensation factor falls within a preset reasonable value range. If _init is less than the preset lower limit of 0.95, then β is taken as that lower limit. If _init is greater than the preset upper limit of 1.05, then take... Set to the upper limit value; otherwise, keep it unchanged. With _init remaining unchanged, this process ultimately yields the error compensation factor that meets the requirements. .

[0049] Simultaneously possesses dynamic adjustment factors and error compensation factor Based on this, model optimization parameters are constructed. These two factors correct the basic algorithm from different dimensions. The dynamic adjustment factor mainly makes macroscopic adjustments based on regional climate background and surface features, while the error compensation factor performs microscopic optimization based on the thermal field spatial characteristics of specific images. These two factors are integrated through a specific combination rule to form complete model optimization parameters. This combination adopts a hybrid form of product and weighting, specifically expressed as follows: ,in This represents the final model optimization parameters. As a dynamic adjustment factor, As an error compensation factor, this combination ensures that the two correction factors function independently yet in coordination with each other.

[0050] Finally, the preset single-window algorithm model is optimized using the optimized parameters of this model. The optimized parameters are then introduced as correction coefficients into the output stage of the basic single-window algorithm. The expression of the optimized single-window algorithm model is as follows: Where Ts represents the retrieved true surface temperature, a and b are empirical constants of the algorithm, namely -67.355351 and 0.458606, respectively, and C is the surface emissivity. With atmospheric transmittance The product of these two variables, D, is an intermediate variable determined by both atmospheric transmittance and surface emissivity. T10 is the brightness temperature value observed at altitude by the satellite sensor, and Ta is the average atmospheric temperature. The model parameters are optimized; this optimization process enables the basic single-window algorithm to have adaptive adjustment capabilities, and can automatically optimize the inversion performance according to the characteristics of specific images and regional environmental conditions, significantly improving the applicability and accuracy of the land surface temperature inversion model under different geographical environments and seasonal conditions.

[0051] Compared with the traditional single-window algorithm, this technical solution significantly improves the performance and applicability of the land surface temperature inversion model by introducing a dual correction mechanism of dynamic adjustment factor and error compensation factor. The traditional single-window algorithm usually uses fixed parameters and forms, which cannot adapt to the spatiotemporal variations of atmospheric conditions and land surface characteristics in different regions and seasons. This rigid model often exhibits problems of unstable accuracy and insufficient adaptability when dealing with complex and diverse real-world scenarios, especially in areas with abnormal atmospheric conditions or complex land cover types, where it is prone to generating large inversion errors.

[0052] This technical solution establishes a scientific dynamic adjustment mechanism and an error compensation mechanism, enabling the single-window algorithm to possess self-optimization and adaptive adjustment capabilities. The dynamic adjustment factor is macroscopically adjusted based on the regional climate background and surface characteristics, ensuring that the algorithm can adapt to the inversion needs under different geographical environments and seasonal conditions. Meanwhile, the error compensation factor microscopically optimizes the thermal field spatial characteristics of specific images, effectively correcting systematic biases caused by abnormal atmospheric conditions or strong surface heterogeneity. This dual correction mechanism ensures that the temperature inversion model maintains the stability of its physical foundation while possessing the flexibility to cope with complex situations. It significantly improves the accuracy and reliability of surface temperature inversion results, providing a higher-quality data foundation for various application scenarios. The optimized single-window algorithm model, while maintaining computational efficiency, greatly enhances its adaptability to different regions, seasons, and surface cover conditions, effectively solving the limitations of traditional methods in terms of regional applicability.

[0053] Step S103: Input the surface emissivity, atmospheric transmittance, brightness temperature, and the pre-acquired average atmospheric temperature into the optimized single-window algorithm model for calculation to determine the surface temperature distribution map of the target area for use in heat island effect analysis.

[0054] In one embodiment of this specification, after completing all preliminary parameter calculations and model optimization, the surface temperature inversion calculation is performed. The system first reads the surface emissivity, atmospheric transmittance, brightness temperature at satellite sensor altitude, and average atmospheric temperature into memory according to a predetermined format, ensuring complete spatial registration of the data. Then, the optimized single-window algorithm model is invoked, and all parameters are substituted into the formula for pixel-level point-by-point calculation. During the calculation, unit consistency checks and numerical validity verification are automatically performed, and outliers exceeding the physical range are marked and replaced. After the inversion calculation is completed, a floating-point raster dataset with the same number of rows and columns as the input data is generated, i.e., a preliminary surface temperature distribution map. The result is then post-processed, including converting the temperature unit from Kelvin to Celsius and applying a spatial filtering algorithm to suppress residual noise, ultimately forming a final surface temperature distribution product available for analysis. This product accurately reveals the spatial distribution details and relative warm and cold patterns of surface temperature in the target area. By overlaying it with land use data, it can automatically identify high-temperature clusters, accurately quantify the intensity of urban heat islands, and generate thematic maps and statistical reports. This provides direct and reliable data support for urban planning, environmental assessment, and climate research, completing a fully automated processing chain from raw satellite data to final application products.

[0055] The technical solutions in the embodiments of this specification integrate geographic spatiotemporal information in metadata with surface feature indices obtained from remote sensing inversion, constructing a dynamic adjustment factor capable of real-time perception of regional environmental characteristics. This couples macroscopic atmospheric background (such as atmospheric profile characteristics that vary with latitude and season) with microscopic surface attributes (such as NDVI reflecting vegetation cover), thereby enabling the single-window algorithm model of the inversion algorithm to have self-adjustment and optimization capabilities. It can automatically adapt to the inversion needs of different underlying surface types and different seasonal climate conditions, from arid to humid areas and from urban core areas to suburbs, fundamentally overcoming the drawbacks of the one-size-fits-all approach of traditional static models. Secondly, by treating the optimized single-window algorithm model as a whole computational unit, the accurately calculated surface emissivity, atmospheric transmittance, brightness temperature, and average atmospheric temperature are processed collaboratively. The parameters within this optimized model are not simply linearly superimposed, but constitute a highly nonlinear functional relationship based on the physical process of thermal infrared radiation transmission. The surface emissivity accurately characterizes the emissivity of the ground object itself, the atmospheric transmittance quantifies the attenuation effect of the atmosphere on thermal infrared radiation, the brightness temperature reflects the original radiation signal received by the sensor, and the average atmospheric temperature represents the contribution of downward atmospheric radiation. The introduction of a dynamic adjustment factor allows the model to fine-tune the calculation results according to the real-time environmental context, thereby significantly reducing the fundamental errors caused by the rigidity of the model structure. Ultimately, the entire chain of technological innovation, from data preprocessing and precise parameter calculation to dynamic algorithm optimization, ensures that the generated surface temperature distribution map not only has higher absolute accuracy but also truly reflects the spatial details and spatiotemporal evolution of the surface thermal field. This provides an unprecedented data quality foundation for the analysis of the urban heat island effect, enabling researchers to more accurately identify the spatial pattern of the heat island, quantify its intensity, and analyze its driving factors, greatly enhancing the practicality of this technical solution in operational applications.

[0056] This specification also provides another method for retrieving land surface temperature based on remote sensing satellite data, which is described below in the corresponding embodiment: Download Landsat-8 OLI_TIRS satellite digital products from platforms such as geospatial data cloud, and select the image data corresponding to the study area. Use the Radiometric Calibration tool in professional software (such as ENVI) to convert the raw DN values ​​of the Landsat-8 data into atmospheric outer surface reflectance or radiance values. The calculation formula is as follows:

[0057] In the formula, This represents the pixel grayscale value DN, whose value is known, and the parameter... and These represent gain and offset, respectively. The offset parameters are obtained from the image header file. (RADIANCE_MULT_BAND_x) and (RADIANCE_ADD_BAND_x) enables radiometric calibration.

[0058] Next, atmospheric correction is performed using a professional software based on the 6S atmospheric radiative transfer model. Apparent reflectance is selected as the data type. The radiometrically calibrated apparent reflectance image file and the _MTL.txt metadata file in the original data folder are input. Parameters such as atmospheric mode (automatically selected by the system or manually set according to actual conditions), aerosol type (continental aerosols are selected by default), and initial visibility (set according to the weather conditions at the time of image capture: 40-1000 km for clear weather, 20-30 km for moderate pollution, and below 15 km for heavy pollution) are set. Whether to invert aerosols pixel-by-pixel (yes is selected by default, performing aerosol optical thickness inversion processing) is also selected. The save path and filename of the generated surface reflectance image are set before atmospheric correction is performed.

[0059] Next, we calculate the key parameters, starting with the brightness temperature. The brightness temperature T refers to the temperature of a blackbody when the observed object and the blackbody radiate equal amounts of energy. It can be used as an indicator of an object's temperature, but it is not the object's true temperature. According to Planck's law, the thermal radiation intensity value is converted into the corresponding brightness temperature. The calculation formula is as follows:

[0060] Radiance value Thermal radiation intensity value (W·m) -2 ·sr -1 ·μm -1 The brightness temperature inversion constants of the TIRS10 band can be obtained from the radiometrically calibrated data; these can be obtained from the image header file. , The values ​​are: K1_CONSTANT_BAND_10 = 774.89, K2_CONSTANT_BAND_10 = 1321.08, and the center wavelength λ is 10.9.

[0061] Secondly, surface emissivity is calculated. Based on a classification-based emissivity method, the remote sensing image is first classified into different land cover types, such as water bodies, vegetation, and urban areas. Based on measured or empirical values, different emissivity values ​​are assigned to each land cover type, thus generating a surface emissivity image. For example, the emissivity of water bodies can be assigned a value of 0.995, while the emissivity of urban areas can be assigned a value between 0.9 and 0.95 based on the characteristics of the building materials.

[0062] Based on the emissivity method of NDVI, the NDVI (Normalized Difference Vegetation Index) of the entire band is calculated. The NDVI tool of professional software is used to calculate the NDVI of the radiometric calibration data of the entire band.

[0063]

[0064] Where NIR represents the pixel value in the near-infrared band, and Red represents the pixel value in the infrared band. Then, vegetation cover (FVC) is calculated based on the NDVI values. The bandmath tool is used to query the radiometrically calibrated full-band vegetation cover. A confidence interval is determined by the cumulative percentage, and the DN values ​​closest to 5% and 95% are taken as the minimum and maximum values, respectively. After statistically analyzing the NDVI values, the Compute Statistics tool is used to obtain relevant data. Finally, the land surface emissivity is calculated using the following formula:

[0065] For mixed pixels, a weighted calculation is performed based on vegetation cover:

[0066] in, Vegetation and Non-vegetation areas were calculated using the formulas described above.

[0067] Finally, atmospheric transmittance is calculated. Atmospheric modeling software (such as MODTRAN) is used to simulate the relationship between atmospheric transmittance and water vapor content. Inputting information such as the time of image formation and the center's latitude and longitude, atmospheric profile parameters are obtained, and then atmospheric transmittance is calculated. Combining existing atmospheric water vapor content products (such as atmospheric water vapor content data provided by NASA) and meteorological station data, atmospheric transmittance is calculated using methods such as weighted averaging. Let the atmospheric transmittance obtained from the atmospheric modeling software simulation be τ1, the atmospheric transmittance obtained from the atmospheric water vapor content product be τ2, and the atmospheric transmittance calculated from the meteorological station data be τ3. Then the fused atmospheric transmittance τ is: τ=w 1 ×τ 1 +w 2 ×τ 2 +w 3 ×τ 3 in, w 1. w 2. w 3 represents the weighting coefficient, which can be determined based on the reliability and accuracy of the data. w 1+ w 2+ w 3 = 1.

[0068] There is a linear relationship between the average atmospheric temperature and the air temperature near the ground (generally 2 m) (T0). The appropriate linear relationship formula is selected for calculation based on the geographical location and season of the study area. For example, for the average tropical atmosphere (15°N, annual average), Ta = 17.9769 + 0.91715T0; for the average summer atmosphere in the mid-latitudes (45°N, July), Ta = 16.0110 + 0.92621T0; and for the average winter atmosphere in the mid-latitudes (45°N, January), Ta = 19.2704 + 0.91118T0.

[0069] An optimized single-window algorithm is used for land surface temperature inversion. The formula for this algorithm is as follows:

[0070] in, The actual surface temperature (K); a and b These are constants, namely -67.355351 and 0.458606; C and D As an intermediate variable, , ,in It is the surface emissivity. τ It is atmospheric transmittance; T 10 It is the brightness temperature (K) of the 10th band pixel detected by the sensor at the satellite's altitude, which can be calculated using the brightness temperature value calculation formula; Ta It is the average atmospheric operating temperature (K); α As a dynamic adjustment factor, it is dynamically adjusted according to the atmospheric conditions and surface characteristics of different regions and seasons, and its value range is [0.8, 1.2]. β This is a compensation factor used to compensate for errors in the algorithm. Its value range is [0.95, 1.05], and it can be optimized and determined according to the actual situation.

[0071] Several representative ground stations were selected within the study area, and measured surface temperature data were obtained using traditional surface temperature observation methods (such as platinum resistance thermometry). The retrieved surface temperature data was compared and analyzed with the measured data, and indices such as root mean square error (RMSE) and mean absolute error (MAE) were calculated to evaluate the accuracy of the retrieval results.

[0072]

[0073]

[0074] in, n For the sample size, TS,反演, i For the first i Inversion of surface temperature for each sample, T S,实测,i For the first i The measured surface temperature of each sample.

[0075] This specification uses Landsat-8 OLI / TIRS satellite imagery provided by the Geospatial Data Cloud website as the data source. Cloudless imagery of XX city is selected, and the map projection is in the UTM-WGS84 projection coordinate system. After system radiometric and geometric correction, the data product is used to invert the surface temperature distribution map of XX city using the methods described above. The surface temperature values ​​are normalized so that the maximum value is 1 and the minimum value is 0. The urban heat island effect density of the study area is divided into five levels according to an arithmetic progression with a tolerance of 0.2. 0–0.2 represents a strong green island area, 0.2–0.4 represents a green island area, 0.4–0.6 represents a normal area, 0.6–0.8 represents a heat island area, and 0.8–1.0 represents a strong heat island area.

[0076] Supervised classification of images of XX City was performed using decision tree classification. The classified images were then filtered, aggregated, and processed to obtain a land use classification map of XX City. The "Class Statistics" function in ENVI software was used to statistically analyze the land use areas of various types—strong heat island zone, heat island zone, normal zone, green island zone, and strong green island zone—to quantitatively study the land use patterns under different heat island effect levels.

[0077] The spatial distribution characteristics of the urban heat island effect in XX City are basically consistent with the outlines of different land use types in the study area. The land use types in the heat island zone and the strong heat island zone are mostly construction land and bare land, meaning that construction land, with impermeable surfaces such as concrete and asphalt as the main underlying surface, is the main contributing factor to the urban heat island. Green island zones and strong green island zones account for 29.6% of the total area of ​​the study area. Water bodies, vegetation, and forest land are the main land types in the green island zones, while water bodies are the dominant land type in the strong green island zones. Therefore, water bodies and forest land with high vegetation coverage can effectively reduce urban surface temperature, thereby weakening the intensity of the urban heat island.

[0078] This specification also provides an embodiment of a surface temperature inversion device based on remote sensing satellite data, such as... Figure 2 As shown, the device includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the above-described method.

[0079] This specification also provides a non-volatile computer storage medium storing computer-executable instructions configured to perform the above-described method.

[0080] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the embodiments of apparatus, devices, and non-volatile computer storage media are basically similar to the method embodiments, so the descriptions are relatively simple; relevant parts can be referred to the descriptions of the method embodiments.

[0081] The above description is merely one or more embodiments of this specification and is not intended to limit this specification. Various modifications and variations can be made to the one or more embodiments of this specification by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principle of one or more embodiments of this specification should be included within the scope of the claims of this specification.

Claims

1. A method for retrieving land surface temperature based on remote sensing satellite data, characterized in that, The method includes: The system receives user-triggered heat island effect analysis requests, acquires satellite remote sensing data and corresponding metadata files for the target area based on these requests, processes the satellite remote sensing data, and determines surface radiation parameters, including surface emissivity, atmospheric transmittance, and brightness temperature. The atmospheric and surface characteristics of the target area are evaluated using the metadata file and the pre-acquired normalized vegetation index. The dynamic adjustment factor corresponding to the target area is determined, and the preset single-window algorithm model is optimized using the dynamic adjustment factor to determine the optimized single-window algorithm model. The surface emissivity, atmospheric transmittance, brightness temperature, and pre-acquired average atmospheric temperature are input into the optimized single-window algorithm model for calculation to determine the surface temperature distribution map of the target area for use in heat island effect analysis.

2. The method for retrieving land surface temperature based on remote sensing satellite data according to claim 1, characterized in that, The satellite remote sensing data is processed to determine the surface radiation parameters, specifically including: The satellite remote sensing data is radiometrically calibrated to obtain the radiance value and surface reflectance data of the target area; Based on the surface reflectance data, the normalized vegetation index and vegetation cover of the target area are calculated, and the surface emissivity of the target area is calculated according to the normalized vegetation index and the vegetation cover. Based on the imaging time and geographic location information in the metadata file, the atmospheric transmittance of the target area is calculated; The brightness temperature of the target area at the satellite sensor altitude is calculated using the radiance value and the brightness temperature inversion constant obtained from the metadata file.

3. The method for retrieving land surface temperature based on remote sensing satellite data according to claim 2, characterized in that, Based on the surface reflectance data, the normalized vegetation index and vegetation cover of the target area are calculated, specifically including: Extract near-infrared and red light band reflectance values ​​from the surface reflectance data; Based on the near-infrared band reflectance value and the red band reflectance value, the normalized vegetation index value of each pixel is calculated using the normalized vegetation index calculation formula. The normalized vegetation index (NDI) values ​​of all pixels within the target area are statistically analyzed to determine the cumulative distribution statistics of the NDI. Based on the cumulative distribution statistics, the minimum and maximum values ​​of the normalized vegetation index are determined. Based on the normalized vegetation index value, the minimum value, and the maximum value, the vegetation coverage of each pixel is calculated using the vegetation coverage calculation formula.

4. The method for retrieving land surface temperature based on remote sensing satellite data according to claim 3, characterized in that, Calculating the surface emissivity of the target area based on the normalized vegetation index and the vegetation cover includes: The initial emissivity value of each pixel is calculated using a predefined piecewise function based on the normalized vegetation index value of each pixel. Based on the vegetation coverage, identify mixed pixels in the target area; For the identified mixed pixels, the initial emissivity of the vegetation component and the initial emissivity of the non-vegetation component are weighted and calculated using the corresponding vegetation coverage value to obtain the surface emissivity of the mixed pixel. For non-mixed pixels, the surface emissivity is determined using the initial emissivity value.

5. The method for retrieving land surface temperature based on remote sensing satellite data according to claim 2, characterized in that, Based on the imaging time and geographic location information in the metadata file, the atmospheric transmittance of the target area is calculated, specifically including: By using a preset atmospheric radiative transfer model, inputting the imaging time and the geographical location information, the first atmospheric transmittance is simulated and obtained; Acquire atmospheric water vapor content product data corresponding to the target area and imaging time, and invert the second atmospheric transmittance to obtain the second atmospheric transmittance. Acquire meteorological station observation data within the target area and calculate the third atmospheric transmittance; Based on the spatial and temporal resolutions of the first, second, and third atmospheric transmittance, the corresponding weighting coefficients are determined. The first atmospheric transmittance, the second atmospheric transmittance, and the third atmospheric transmittance are weighted and averaged using the weighting coefficients to obtain the fused atmospheric transmittance.

6. The method for retrieving land surface temperature based on remote sensing satellite data according to claim 1, characterized in that, The atmospheric and surface characteristics of the target area are evaluated using the metadata file and the pre-acquired normalized vegetation index to determine the dynamic adjustment factor corresponding to the target area, specifically including: Extract the center latitude information of the target area and the month information of the imaging time from the metadata file, and calculate the atmospheric condition adjustment component based on the center latitude information and the month information; Calculate the average normalized vegetation index of all pixels in the target area, and calculate the land feature adjustment component based on the average normalized vegetation index. The dynamic adjustment factor is calculated by weighted fusion of the atmospheric condition adjustment component and the surface feature adjustment component.

7. The method for retrieving land surface temperature based on remote sensing satellite data according to claim 1, characterized in that, The preset single-window algorithm model is optimized using the aforementioned dynamic adjustment factor to determine the optimized single-window algorithm model, specifically including: Obtain a predetermined error compensation factor, and determine the model optimization parameters based on the dynamic adjustment factor and the error compensation factor; The single-window algorithm model is optimized using the model optimization parameters to determine the optimized single-window algorithm model.

8. The method for retrieving land surface temperature based on remote sensing satellite data according to claim 7, characterized in that, Obtain the predetermined error compensation factor, specifically including: Calculate the spatial standard deviation of the brightness temperature values ​​within the target region; Based on the spatial distribution standard deviation, the initial compensation value is calculated using a preset mapping function; The initial compensation value is subjected to range constraint processing to ensure that it falls within the range of the preset compensation factor value, and the error compensation factor is determined based on the compensation value after constraint processing.

9. A surface temperature retrieval device based on remote sensing satellite data, characterized in that, The device includes: At least one processor; and, A memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor to enable the at least one processor to perform the method as described in any one of claims 1-8.

10. A non-volatile computer storage medium storing computer-executable instructions, characterized in that, The computer-executable instructions are configured to perform the method as described in any one of claims 1-8.

Images