Surface temperature satellite remote sensing estimation method and system based on passive microwave single-channel brightness temperature

Abstract

The invention provides an earth surface temperature satellite remote sensing estimation method and system based on passive microwave single-channel brightness temperature, and relates to the technical field of satellite remote sensing, and the method comprises the steps: based on a global monthly scale microwave emissivity climate state data set, aiming at the selected Ku frequency vertical polarization channel brightness temperature, calculating the Ku frequency vertical polarization channel brightness temperature; acquiring Ku frequency vertical polarization climate state emissivity with global spatial heterogeneity characteristics; performing quality control on the brightness temperature of the Ku frequency vertical polarization channel and the Ku frequency vertical polarization climate state emissivity with the global spatial heterogeneity characteristic, and re-projecting and sampling the emissivity data to the same spatial projection and resolution as the brightness temperature data so as to obtain preprocessed brightness temperature and emissivity data; and constructing a land surface temperature estimation model with global universality based on the Ku frequency vertical polarization brightness temperature according to a passive microwave radiation transfer equation with approximate derivation and the preprocessed brightness temperature and emissivity. The fabric can effectively penetrate through cloud layer shielding.

Description

Technical Field

[0001] This invention relates to the field of satellite remote sensing technology, and in particular to a satellite remote sensing estimation method and system for surface temperature based on passive microwave single-channel brightness temperature. Background Technology

[0002] In recent years, a series of passive microwave inversion methods for land surface temperature have been proposed. Based on the number of passive microwave channels used in the inversion process, existing methods can be clearly divided into three categories: single-channel models, dual-channel models, and multi-channel models. Dual-channel and multi-channel models require passive microwave brightness temperature data from two or more channels as input, while single-channel models can perform land surface temperature inversion using only the brightness temperature data from a single passive microwave channel. Therefore, single-channel models have significant advantages in terms of computational efficiency and convenience.

[0003] Current single-channel inversion models primarily use brightness temperature data from a single Ka-frequency vertical polarization channel as the core input, establishing a linear fit with measured land surface temperature. However, the accuracy and stability of this type of method are constrained by multiple factors: First, complex underlying surface features (such as vegetation cover gradients and soil moisture content differences) and atmospheric structures (such as cloud water content and atmospheric water vapor profiles) affect the physical mapping relationship between brightness temperature and land surface temperature; second, insufficient sample size and spatial representativeness during model training lead to regional dependence and application limitations in their inversion performance. These problems directly result in poor portability of existing single-channel models, leading to low accuracy when extended to regional scales and global applications, making it difficult to meet the application requirements of high-precision dynamic land surface temperature monitoring. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide a satellite remote sensing estimation method and system for land surface temperature based on passive microwave single-channel brightness temperature, which has stronger atmospheric penetration capability and anti-interference characteristics and can effectively penetrate cloud cover.

[0005] To solve the above-mentioned technical problems, the technical solution of the present invention is as follows:

[0006] Firstly, a satellite remote sensing estimation method for land surface temperature based on passive microwave single-channel brightness temperature, the method comprising:

[0007] Step 1: Derive the approximate passive microwave radiation transfer equation and establish a simplified relationship between microwave brightness temperature and surface temperature.

[0008] Step 2: Based on the approximate simplification conditions of the derived approximate passive microwave radiative transfer equation, and the analysis results of the sensitivity of microwave brightness temperature to atmospheric influence and the sensitivity of surface temperature and surface emissivity changes, the brightness temperature of the Ku frequency vertical polarization channel is selected as the single-channel input data.

[0009] Step 3: Based on the global monthly microwave emissivity climatological dataset, obtain the Ku-frequency vertical polarization climatological emissivity with global spatial heterogeneity characteristics for the selected Ku-frequency vertical polarization channel brightness temperature.

[0010] Step 4: Perform quality control on the brightness temperature of the Ku-frequency vertical polarization channel and the Ku-frequency vertical polarization climatological emissivity, which has global spatial heterogeneity characteristics. Reproject and sample the emissivity data to the same spatial projection and resolution as the brightness temperature data to obtain preprocessed brightness temperature and emissivity data. Based on the preprocessed brightness temperature and emissivity data, construct a globally universal surface temperature estimation model to calculate the global-scale surface temperature distribution.

[0011] Step 5: Correct the global-scale surface temperature distribution to obtain the true surface temperature.

[0012] Secondly, a satellite remote sensing estimation system for land surface temperature based on passive microwave single-channel brightness temperature includes:

[0013] The derivation module is used to derive the approximate passive microwave radiative transfer equation and establish a simplified relationship between microwave brightness temperature and surface temperature.

[0014] The analysis module is used to select the brightness temperature of the Ku-frequency vertical polarization channel as single-channel input data based on the approximate simplification conditions of the derived approximate passive microwave radiative transfer equation, as well as the analysis results of the sensitivity of microwave brightness temperature to atmospheric influence and the sensitivity of surface temperature and surface emissivity changes.

[0015] The acquisition module is used to obtain the Ku-frequency vertical polarization climatological emissivity with global spatial heterogeneity characteristics based on the global monthly microwave emissivity climatological dataset and the brightness temperature of the selected Ku-frequency vertical polarization channel.

[0016] The processing module performs quality control on the brightness temperature of the Ku-frequency vertical polarization channel and the Ku-frequency vertical polarization climatological emissivity, which has global spatial heterogeneity characteristics. It also reprojects and samples the emissivity data to the same spatial projection and resolution as the brightness temperature data to obtain preprocessed brightness temperature and emissivity data. Based on the preprocessed brightness temperature and emissivity data, a globally universal surface temperature estimation model is constructed to calculate the global-scale surface temperature distribution.

[0017] The correction module is used to correct the global-scale surface temperature distribution to obtain the true surface temperature.

[0018] Thirdly, a computing device includes:

[0019] One or more processors;

[0020] A storage device for storing one or more programs that, when executed by one or more processors, cause the one or more processors to implement the method.

[0021] Fourthly, a computer-readable storage medium storing a program that, when executed by a processor, implements the method.

[0022] The above-described solution of the present invention has at least the following beneficial effects:

[0023] This invention derives an approximate passive microwave radiative transfer equation, selects a satisfactory passive microwave single-channel brightness temperature, obtains emissivity with spatial heterogeneity on a global scale, and constructs a universal passive microwave surface temperature estimation model, thus obtaining a universally applicable satellite-borne passive microwave surface temperature on a global scale. This invention proposes a simple and practical method for calculating surface temperature based solely on the brightness temperature of a single passive microwave channel. Compared to multi-channel brightness temperature methods, it is more convenient and efficient. Furthermore, compared to traditional regionally dependent Ka-frequency vertical polarization single-channel brightness temperature methods, it fully considers the spatial heterogeneity of emissivity on a global scale, thereby reducing the impact of microwave emissivity on the accuracy of surface temperature inversion and achieving universally applicable satellite remote sensing estimation of passive microwave surface temperature. Attached Figure Description

[0024] Figure 1 This is a flowchart illustrating the satellite remote sensing estimation method for land surface temperature based on passive microwave single-channel brightness temperature provided in an embodiment of the present invention.

[0025] Figure 2 This is a schematic diagram of a satellite remote sensing estimation system for surface temperature based on passive microwave single-channel brightness temperature, provided by an embodiment of the present invention. Detailed Implementation

[0026] Exemplary embodiments of the present disclosure will now be described in more detail 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 implemented 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.

[0027] like Figure 1 As shown, embodiments of the present invention propose a satellite remote sensing estimation method for land surface temperature based on passive microwave single-channel brightness temperature, the method comprising the following steps:

[0028] Step 1: Derive the approximate passive microwave radiation transfer equation and establish a simplified relationship between microwave brightness temperature and surface temperature.

[0029] Step 2: Based on the approximate simplification conditions of the derived approximate passive microwave radiative transfer equation, and the analysis results of the sensitivity of microwave brightness temperature to atmospheric influence and the sensitivity of surface temperature and surface emissivity changes, the brightness temperature of the Ku frequency vertical polarization channel is selected as the single-channel input data.

[0030] Step 3: Based on the global monthly microwave emissivity climatological dataset, obtain the Ku-frequency vertical polarization climatological emissivity with global spatial heterogeneity characteristics for the selected Ku-frequency vertical polarization channel brightness temperature.

[0031] Step 4: Perform quality control on the brightness temperature of the Ku-frequency vertical polarization channel and the Ku-frequency vertical polarization climatological emissivity, which has global spatial heterogeneity characteristics. Reproject and sample the emissivity data to the same spatial projection and resolution as the brightness temperature data to obtain preprocessed brightness temperature and emissivity data. Based on the preprocessed brightness temperature and emissivity data, construct a globally universal surface temperature estimation model to calculate the global-scale surface temperature distribution.

[0032] Step 5: Correct the global-scale surface temperature distribution to obtain the true surface temperature.

[0033] In this embodiment of the invention, by selecting a Ku-frequency vertically polarized single channel that is insensitive to atmospheric influences but sensitive to surface temperature, the interference from atmospheric factors such as clouds and water vapor is effectively reduced, ensuring data availability under complex weather conditions. Simultaneously, by combining climatological emissivity data with global spatial heterogeneity, a universally applicable estimation model is constructed, overcoming the instability in accuracy caused by the strong regional sample dependence of traditional single-channel models. This is the first time that stable and reliable surface temperature data can be obtained globally using only a single microwave channel. Compared to traditional single-channel methods relying on Ka frequencies, the Ku frequency selected in this invention has better atmospheric penetration while being less sensitive to emissivity changes caused by drastic variations in surface soil moisture content, resulting in a more stable physical relationship between brightness temperature and surface temperature. Furthermore, through systematic data preprocessing steps (such as quality control to remove water bodies, snow, and precipitation areas), the contamination of brightness temperature signals by non-target features and extreme weather is effectively eliminated, further ensuring the physical consistency and accuracy of the inversion results. This invention creatively introduces a penetration correction step for the Ku frequency. Because Ku frequencies have a certain penetrating power into soil, the temperature obtained through direct inversion is actually the soil temperature at a certain depth below the surface. Through the correction process of this invention, this temperature can be effectively converted into the true surface skin temperature, thus solving the long-standing problem of physical definition deviation in passive microwave remote sensing inversion of surface temperature.

[0034] In a preferred embodiment of the present invention, step 1, deriving an approximate passive microwave radiative transfer equation and establishing a simplified relationship between microwave brightness temperature and Earth's surface temperature, includes:

[0035] Step 11, based on Rayleigh-Jones' law and Planck's radiation law, for the soil-vegetation-atmosphere surface, considering vegetation as a single-scattering layer on rough soil, derive an approximate passive microwave radiative transfer equation. The derivation of the approximate passive microwave radiative transfer equation specifically includes:

[0036] First, referring to the classical theory of passive microwave remote sensing for surface temperature inversion, it is determined that the brightness temperature signal received by the satellite passive microwave sensor is a superposition of contributions from multiple components, including surface radiation, atmospheric radiation, and cosmic background radiation. The specific expression of its original radiative transfer equation is as follows:

[0037] ;

[0038] in, This refers to the passive microwave brightness temperature, which is the radiation temperature actually observed by the satellite sensor. The Earth's surface temperature that can be detected by microwaves; Surface emissivity reflects the Earth's ability to radiate microwave energy outwards from its surface. Atmospheric optical thickness reflects the degree of atmospheric attenuation of microwave radiation. Temperature of the vegetation canopy; The vegetation single-scatter albedo; The temperature of the cosmic microwave background radiation is a contribution from the background radiation from outer space. Indicates upward atmospheric radiation; Indicates the optical thickness of vegetation; Indicates downward atmospheric radiation; It represents the equivalent atmospheric radiation.

[0039] In passive microwave remote sensing, it is assumed that the atmosphere is a homogeneous, non-scattering medium. Atmospheric absorption and emission are mainly affected by atmospheric transmittance and upward and downward atmospheric radiation. The upward and downward atmospheric radiation can be approximated as equal, i.e. .

[0040] Step 12: For the derived approximate passive microwave radiative transfer equation, after neglecting atmospheric effects or correcting for atmospheric effects, assuming that the vegetation canopy temperature and the land surface temperature are approximately equal, the physical relationship between microwave brightness temperature and land surface temperature is simplified into a linear expression to establish a simplified relationship between microwave brightness temperature and land surface temperature. Specifically, this includes:

[0041] Based on the approximate radiative transfer equation obtained in step 11, and considering the actual technical requirements of passive microwave surface temperature inversion, a precondition is introduced that atmospheric influences are not considered or have been corrected for. This condition is based on two points: firstly, subsequent data preprocessing will reduce atmospheric interference by removing precipitation areas and incorporating atmospheric auxiliary data; secondly, to meet the efficiency requirements of single-channel inversion, it is necessary to isolate the complex influence of atmospheric factors on the relationship between brightness temperature and surface temperature, focusing on the mapping patterns between core variables.

[0042] Based on the above premises, the approximate radiative transfer equation derived in step 11 is further simplified. Specifically, since atmospheric effects have been eliminated or corrected, the atmospheric optical thickness... ,at this time And atmospheric equivalent radiation temperature Its contribution to brightness temperature is negligible; cosmic microwave background radiation temperature The value is extremely small, and its contribution to the overall brightness temperature signal is negligible in engineering applications. Furthermore, assuming no vegetation influence, the canopy temperature and the surface temperature are approximately equal, i.e. Therefore, ignoring atmospheric effects Alternatively, after atmospheric correction, assuming that the vegetation canopy temperature and the surface temperature are approximately equal, a simplified passive microwave radiative transfer equation is obtained. The surface emissivity is preserved during the simplification process. Key parameters were optimized to ensure the equations still reflect the core physical mapping relationship between microwave brightness temperature and Earth's surface temperature. After the above simplification process, complex terms in the equations were removed, retaining only microwave brightness temperature. Surface temperature and surface emissivity Three core variables.

[0043] Through the above simplification process, the linear expression between microwave brightness temperature and surface temperature is finally obtained, specifically in the form:

[0044] ;

[0045] in, Brightness temperature data at frequency f observed by the satellite passive microwave sensor (the optimal observation frequency channel will be selected based on this linear relationship in subsequent step 2). for The surface emissivity at a given frequency depends on the observation frequency and polarization mode, and is affected by factors such as soil moisture, surface roughness, and vegetation growth status (step 3 will obtain the corresponding emissivity data for the selected frequency). This represents the microwave-detectable surface temperature. This linear expression clearly establishes the core correlation between microwave brightness temperature and surface temperature at a single frequency, providing a fundamental theoretical basis for subsequent surface temperature estimation models based on single-channel brightness temperature.

[0046] In this embodiment of the invention, the complex physical model of passive microwave radiation transmission is significantly simplified. By reasonably assuming and removing redundant variables, the computational complexity of subsequent single-channel inversion is reduced, meeting the core requirements of efficient and convenient single-channel model operation. A linear correlation between microwave brightness temperature and surface temperature is established, providing a core theoretical basis for subsequent selection of optimal single channels (such as Ku-frequency vertical polarization channels), matching corresponding emissivity data, and constructing estimation models. The simplification process takes into account both physical rationality and engineering practicality, retaining the core mapping relationship between brightness temperature and surface temperature, and laying the foundation for overcoming the regional dependence of traditional single-channel models and achieving universal inversion on a global scale.

[0047] In a preferred embodiment of the present invention, step 2, based on the approximate simplification conditions of the derived approximate passive microwave radiative transfer equation, and the analysis results of the sensitivity of microwave brightness temperature to atmospheric influence and the sensitivity of changes in surface temperature and surface emissivity, selects the brightness temperature of the Ku-frequency vertical polarization channel as the single-channel input data, including:

[0048] Step 21: Based on the derived approximate passive microwave radiative transfer equation, and considering the linear relationship between passive microwave brightness temperature and surface temperature, as well as the close correlation between surface temperature estimation and surface emissivity changes, analyze the approximate simplification preconditions of the passive microwave radiative transfer equation, the sensitivity of passive microwave brightness temperature to surface temperature, and the sensitivity of surface emissivity changes to obtain the analysis results. Specifically, this includes the linear relationship expression obtained in Step 12, which is: (in the formula) The passive microwave brightness temperature at frequency f. Let f be the surface emissivity at frequency f. (This refers to the microwave-detectable surface temperature). Simultaneously, combining this with the approximate passive microwave radiative transfer equation derived in step 11, the equivalent atmospheric radiative temperature is determined. Atmospheric optical thickness The potential impact of atmospheric factors on brightness temperature signals is used to define boundary conditions for subsequent sensitivity analysis.

[0049] Based on the linear relationship expression, the surface emissivity at a fixed frequency f and the corresponding frequency Under the premise of keeping brightness constant, the brightness temperature is analyzed by the method of controlling variables. With surface temperature Analysis revealed a linear relationship between brightness temperature and surface temperature, with the slope of this linear relationship varying across different frequencies and polarization modes. The slope of the vertically polarized channel was generally higher than that of the horizontally polarized channel, indicating that the brightness temperature of the vertically polarized channel was more sensitive to changes in surface temperature.

[0050] Also based on a linear relationship expression, at a fixed frequency f and surface temperature Under the premise of keeping it unchanged, analyze the brightness temperature With surface emissivity The variation patterns of microwave emissivity are investigated. Microwave emissivity depends on frequency and polarization, and is influenced by various surface factors such as soil moisture and surface roughness. The focus is on comparing the degree of interference between emissivity changes and brightness temperature at different frequencies: the emissivity of low-frequency channels (such as L, C, and X) varies significantly with soil moisture, resulting in higher sensitivity of brightness temperature to emissivity changes; the emissivity of high-frequency channels (such as Ka and W) is significantly affected by atmospheric factors, failing to adequately support a linear relationship between brightness temperature and surface temperature. The core analytical results show that the brightness temperature of vertically polarized channels is more sensitive to surface temperature, and there are significant differences in the sensitivity of brightness temperature to atmospheric influences and surface emissivity changes among different frequency channels.

[0051] Step 22: Based on the analysis results, from the relatively low-frequency channels at and below the Ka frequency, select vertical polarization channels that are insensitive to atmospheric influences but sensitive to surface temperature. Simultaneously, exclude frequency channels with excessive penetration and those sensitive to changes in soil moisture content to determine the candidate channel set, specifically including:

[0052] Four screening criteria were established: ① Low sensitivity to atmospheric influences, i.e., selecting relatively low-frequency channels (e.g., Ka frequencies and below); ② Low penetration, i.e., avoiding excessively low-frequency channels (e.g., not selecting L frequencies); ③ Low sensitivity to changes in soil moisture content (e.g., not selecting L, C, or X frequencies); ④ High sensitivity to surface temperature (e.g., selecting vertically polarized channels). Simultaneously, the standard channels of existing international spaceborne passive microwave radiometers were used as the screening range, encompassing vertical (V) and horizontal (H) polarized channels at frequencies such as L, C, X, Ku, K, Ka, and W, ensuring that the selected targets closely match actual satellite observation scenarios.

[0053] The first step, based on condition ① (relatively low frequency channels at Ka frequency and below), eliminates all polarization channels at the high-frequency W frequency, retaining vertical and horizontal polarization channels at L, C, X, Ku, K, and Ka frequencies. The second step, based on condition ④ (vertical polarization channels sensitive to surface temperature), eliminates all horizontal polarization channels, retaining only LV, CV, XV, Ku-V, KV, and Ka-V channels (V represents vertical polarization). The third step, based on condition ② (excessively low frequency channels with strong penetration), and considering the inability to select excessively low frequency channels (e.g., not selecting L frequency), eliminates the LV channel, leaving CV, XV, Ku-V, KV, and Ka-V channels. The fourth step, based on condition ③ (insensitive to soil moisture changes), further analyzes the remaining channels: CV and XV channels are highly sensitive to soil moisture changes, therefore, they are eliminated. Thus, the remaining Ku-V, KV, and Ka-V channels still have screening value and are temporarily retained.

[0054] After the above four rounds of screening, a set of candidate channels that meet all the basic screening criteria was finally obtained, specifically Ku-V, KV, and Ka-V vertical polarization channels.

[0055] Step 23: From the candidate channel set, simulate the difference between the on-board brightness temperature and the surface emission brightness temperature of each channel using the MonoRTM atmospheric radiative transfer model. Analyze and compare the differences in the degree of atmospheric influence on each channel, and finally select the brightness temperature of the Ku-frequency vertically polarized channel as the single-channel input data. Specifically, this includes:

[0056] Based on the analysis results in step 21 showing differences in the sensitivity of brightness temperatures of different frequency channels to atmospheric influence, the degree of atmospheric influence on the channel is used as the core comparison index. This index is also an important condition for deriving the approximate passive microwave radiative transfer equation, i.e., ignoring atmospheric influence or performing atmospheric influence correction. According to the physical law that the longer the wavelength of passive microwave data, the stronger the atmospheric penetration ability and the stronger the anti-interference characteristics, the lower the frequency and the longer the wavelength, the less affected by the atmosphere (clouds, water vapor, etc.), and the higher the inversion stability.

[0057] Simultaneously, by combining the MonoRTM atmospheric radiative transfer model to simulate the difference between the on-board brightness temperature and the surface emission brightness temperature of each channel, the Ku-V, KV, and Ka-V vertical polarization channels in the candidate channel set were ranked by frequency and analyzed for atmospheric influence: the frequencies of the three channels, from low to high, are Ku frequency < K frequency < Ka frequency, with the corresponding atmospheric penetration decreasing sequentially and the degree of influence from the atmosphere (such as cloud water content and atmospheric water vapor profile) increasing sequentially. Among them, the Ka-V channel is a commonly used channel in traditional single-channel models; the KV channel is slightly less affected by the atmosphere than the Ka-V channel, but more affected than the Ku-V channel; the Ku-V channel, as a relatively low-frequency channel, meets the requirement of being less sensitive to atmospheric influence while avoiding the defect of excessive penetration at L frequency.

[0058] Based on the above comparative analysis, and considering the different atmospheric effects on the three, the relatively low-frequency Ku-frequency vertical polarization channel was ultimately selected. The brightness temperature of the Ku-frequency vertical polarization channel was then determined as the single-channel input data (this channel brightness temperature is expressed as...). At the same time, this choice can overcome the shortcomings of traditional Ka frequency single channel. Compared with Ka frequency, Ku frequency maintains strong atmospheric penetration capability while avoiding the defect of excessive low frequency penetration, making the physical relationship between brightness temperature and surface temperature more stable.

[0059] In this embodiment, sensitivity analysis provides scientific theoretical support for channel selection, avoiding the blindness of traditional channel selection and ensuring that the selection direction meets the core requirements of efficient inversion and high precision of single channel. The channel range is precisely narrowed down according to clear conditions, eliminating unsuitable channels such as those with excessive penetration or those that are sensitive to soil moisture content and thus affect emissivity changes, ensuring the basic applicability of the candidate channel set. The Ku frequency vertical polarization channel finally selected in step 23 effectively reduces the impact of atmospheric interference and soil moisture content changes, overcoming the accuracy and stability problems of traditional Ka frequency single channels.

[0060] In a preferred embodiment of the present invention, step 3, based on a global monthly microwave emissivity climatological dataset, obtains the Ku-frequency vertical polarization climatological emissivity with global spatial heterogeneity characteristics for the selected Ku-frequency vertical polarization channel brightness temperature, including:

[0061] Step 31: Select a globally covered monthly microwave emissivity climatological dataset. This dataset contains emissivity climatological information for various frequencies and polarizations, specifically including: a definition of microwave emissivity characteristics. Microwave emissivity is a parameter dependent on frequency and polarization, and influenced by various surface factors such as soil moisture, soil texture, surface roughness, and vegetation growth status. Therefore, its spatial variation is significant. Furthermore, based on the key assumption of this invention that emissivity exhibits spatial heterogeneity under certain climatological conditions but does not change significantly over a certain period, the dataset must meet three core conditions: ① It must have global coverage to support global-scale surface temperature inversion requirements and possess global spatial heterogeneity; ② The time series length must be no less than 3 years, and the temporal resolution must be no less than a monthly scale (if the temporal resolution is daily data at a daily scale, the daily emissivity must be synthesized into monthly average emissivity data), matching the climatological characteristic that emissivity does not change significantly over a certain period; ③ It must contain emissivity climatological information for various frequencies and polarizations, and be able to match the Ku-frequency vertical polarization channel finally selected in Step 2.

[0062] Based on the above conditions, we selected existing mature global microwave emissivity climatological datasets (such as climatological products generated by assimilation of long-term satellite observation data). These datasets should cover the frequencies commonly used by international spaceborne passive microwave radiometers and include vertical V-polarized emissivity data for Ku frequencies. The time series length should be no less than 3 years. In terms of time dimension, monthly average emissivity climatological data from January to December should be provided to ensure that it can meet the needs of different observation times. In terms of spatial dimension, it should have a global land coverage pattern and be able to reflect the spatial heterogeneity of emissivity on a global scale (such as the differences in emissivity of different underlying surfaces such as tropical forests, temperate grasslands, and arid deserts).

[0063] After screening, a global monthly microwave emissivity climatological dataset that meets all the above conditions was selected as the emissivity data source of this invention. The core advantage of this dataset is that it can provide frequency and polarization emissivity information that matches the channel selected in step 2, and it can take into account the global spatial heterogeneity and temporal stability of emissivity through the characteristics of climatological data, which meets the core requirement of this invention to obtain emissivity with spatial heterogeneity on a global scale.

[0064] Step 32: Based on the finally selected Ku-frequency vertical polarization channel, extract the monthly average emissivity data corresponding to the Ku-frequency vertical polarization from the global monthly microwave emissivity climatological dataset, specifically including:

[0065] Using the Ku-frequency vertical polarization channel finally selected in step 23 as the core matching benchmark, the key parameters for data extraction are determined: the frequency is the Ku band, and the polarization mode is vertical polarization V, ensuring that the extracted emissivity data matches the brightness temperature channel selected in step 2. They are completely consistent in frequency and polarization, which is the basis for subsequent construction of surface temperature estimation models. (The foundation of)

[0066] Based on the matching rules described above, the global monthly microwave emissivity climatological dataset selected in step 31 is filtered and extracted. Specifically, the following steps are taken: In the frequency dimension of the dataset, all data layers corresponding to Ku frequencies are located and filtered; within the Ku frequency data layers, the emissivity data layers corresponding to vertical polarization V are further filtered, and horizontal polarization (H) data are removed; all monthly average emissivity climatological data from January to December in this data layer are retained to form a Ku frequency vertical polarization monthly average emissivity dataset. This dataset maintains a globally covered spatial pattern and can reflect the spatial heterogeneity characteristics of Ku frequency vertical polarization emissivity in different regions.

[0067] After extraction, the data validity was verified to confirm that: the extracted data frequency and polarization completely matched the Ku frequency vertical polarization channel selected in step 2, with no channel mismatch issues; the extracted data contained complete monthly-scale information from January to December, with no missing time-dimensional data; the data had complete spatial coverage, with no spatial gaps globally, ensuring that it could support global-scale land surface temperature inversion. After verification, the next step was initiated.

[0068] Step 33: Based on the month corresponding to the current observation time, select the emissivity data for the corresponding month from the extracted monthly average emissivity data of Ku-frequency vertical polarization to obtain the Ku-frequency vertical polarization climatological emissivity with global spatial heterogeneity characteristics, specifically including:

[0069] Obtain the satellite observation time (accurate to the month, such as July 2023) of the Ku-frequency vertical polarization channel brightness temperature data in step 2, and determine that the month for matching emissivity data is July. This operation is based on the assumption of this invention that the emissivity under climatological conditions does not change much over a certain period of time. The difference between the climatological emissivity of the same month and the emissivity of the actual observed month is small, which can meet the accuracy requirements for surface temperature estimation.

[0070] Based on a specific observation month (e.g., July), the monthly average emissivity dataset of Ku-frequency vertical polarization extracted in step 32 is filtered by month. Specifically, a climatological layer of monthly average emissivity data corresponding only to July is selected from this dataset, regardless of the year. This data layer preserves the spatial distribution characteristics of July emissivity of Ku-frequency vertical polarization globally.

[0071] After filtering, Ku-frequency vertical polarization climatological emissivity data matching the current observation month are obtained (represented as...). This data possesses two core characteristics: its frequency and polarization perfectly match the brightness temperature channel selected in step 2, meeting the input requirements of the estimation model; and it exhibits global spatial heterogeneity, enabling it to adapt to surface temperature inversion scenarios across different underlying surfaces globally, providing key emissivity input parameters for the subsequent construction of a globally universal surface temperature estimation model.

[0072] In a preferred embodiment of the present invention, step 4 involves quality control of the brightness temperature of the Ku-frequency vertical polarization channel and the Ku-frequency vertical polarization climatological emissivity, which exhibits global spatial heterogeneity. The emissivity data is then reprojected and sampled to the same spatial projection and resolution as the brightness temperature data to obtain preprocessed brightness temperature and emissivity data. Based on the preprocessed brightness temperature and emissivity data, a globally universal surface temperature estimation model is constructed to calculate the global-scale surface temperature distribution, including:

[0073] Step 41: Based on global land cover type information, identify and remove areas covered by oceans, inland waters, wetlands, and snow and ice, and extract Ku-frequency vertical polarization channel brightness temperature data over the land surface, specifically including:

[0074] Based on the key conclusion that passive microwave brightness temperature is significantly affected by oceans, inland waters, wetlands, and snow, the core processing principle of this step is determined: non-land areas are excluded, and effective land-based brightness temperature data is retained. The input data includes two types: the Ku-frequency vertical polarization channel brightness temperature data finally obtained in step 23 (represented as...). This data has global coverage and includes brightness temperature information for various regions such as land, ocean, and snow. The global land cover type dataset (such as internationally used land cover products) needs to clearly distinguish different land cover types such as ocean, inland water, wetland, snow, forest, grassland, and desert, and the spatial resolution should be compatible with the brightness temperature data.

[0075] Based on a global land cover type dataset, a mask for non-target areas is constructed through attribute filtering. Specifically, in the land cover data, the spatial extent of all land cover types—marine / inland water bodies (e.g., lakes, rivers, reservoirs), wetlands (e.g., swamps, mudflats), and snow and ice cover areas (e.g., polar ice caps, alpine glaciers, seasonal snow cover)—is located and marked as invalid areas. Other land cover types (e.g., forests, grasslands, deserts, farmland) are marked as valid land areas, forming a binary mask (valid areas are assigned a value of 1, invalid areas are assigned a value of 0).

[0076] Spatially overlaying the constructed non-target area mask with Ku-frequency vertical polarization channel brightness temperature data is performed. Masking operations retain valid land area brightness temperature data with a mask value of 1, while invalid area data such as oceans, inland waters, wetlands, and snow / ice areas with a mask value of 0 are removed. After the operation, Ku-frequency vertical polarization channel brightness temperature data covering only the global land area is obtained, which has preliminarily eliminated contamination of the brightness temperature signal by non-target features.

[0077] Step 42: Based on global atmospheric precipitation information, identify and remove precipitation areas, and filter out the Ku-frequency vertical polarization channel brightness temperature data of non-precipitation areas from the land surface to obtain preprocessed brightness temperature data, specifically including:

[0078] This step uses the global land area Ku-frequency vertical polarization channel brightness temperature data extracted in step 41 as the processing object. It eliminates the technical requirement of extreme weather to contaminate the brightness temperature signal and focuses on the interference of precipitation weather on the brightness temperature data. Precipitation processes will change the atmospheric microwave radiation characteristics, resulting in brightness temperature signal distortion. Therefore, it is necessary to remove data from precipitation areas. The input auxiliary data is a global atmospheric precipitation dataset (such as satellite remote sensing precipitation products). This dataset needs to have a temporal resolution (such as daily or real-time precipitation data) and spatial resolution that match the brightness temperature data, and can accurately identify precipitation areas worldwide.

[0079] Based on a global atmospheric precipitation dataset, a precipitation threshold is set (e.g., precipitation intensity ≥ 0.1 mm / h is considered a precipitation area). All spatial regions within the land area extracted in step 41 that meet the precipitation threshold are located and marked. These regions are then marked as invalid precipitation areas, and the remaining non-precipitated areas are marked as non-valid precipitation areas. A binary mask is also created (non-precipitated areas are assigned a value of 1, and precipitation areas are assigned a value of 0). It should be noted that the spatial range of this mask is strictly limited to the valid land area of ​​step 41 to avoid duplicate processing of already excluded areas such as oceans.

[0080] The precipitation area identification mask is spatially overlaid with the land brightness temperature data obtained in step 41. Brightness temperature data from non-precipitation areas with a mask value of 1 are retained, while those from precipitation areas with a mask value of 0 are discarded. After the operation, preprocessed Ku-frequency vertical polarization channel brightness temperature data is obtained. This data only covers non-precipitation areas of global land, completely eliminating contamination of the brightness temperature signal by non-target features such as oceans, water bodies, wetlands, snow, and extreme precipitation weather, effectively ensuring the physical consistency and accuracy of the brightness temperature data.

[0081] Step 43: For Ku-frequency vertical polarization climatological emissivity data with global spatial heterogeneity, geographic masking is applied to remove ocean, inland water, wetland, snow-covered areas, and precipitation areas, as well as outliers with emissivity greater than 1, to obtain quality-controlled emissivity data, specifically including:

[0082] The core objective and input data for data processing are defined as follows: The core objective is to ensure that the emissivity data and the preprocessed brightness temperature data are completely consistent in spatial range, in order to facilitate the subsequent construction of a land surface temperature estimation model. The system provides matching input parameters. If the spatial ranges of the two parameters do not match, it will lead to data misalignment or invalid values ​​during model calculation. The input data includes two types: Ku-frequency vertical polarization climatological emissivity data with global spatial heterogeneity characteristics obtained in step 33 (represented as...). The data covers the entire globe; the non-target area mask constructed in step 41 and the precipitation area identification mask constructed in step 42 ensure that the emissivity and brightness temperature of the rejected areas are completely synchronized.

[0083] The non-target area mask from step 41 (excluding ocean, inland water, wetland, and snow) is overlaid and fused with the precipitation area identification mask from step 42 (excluding land precipitation areas) to construct a comprehensive processing mask. Fusion rules: Only spatial regions that simultaneously satisfy a non-target area mask value of 1 (land) and a precipitation area mask value of 1 (non-precipitation) are retained; all other regions (ocean, water, wetland, snow, and land precipitation areas) are marked as invalid regions, forming a binary comprehensive mask that is completely consistent with the brightness and temperature preprocessing range.

[0084] The constructed integrated processing mask is spatially superimposed with the Ku-frequency vertical polarization climatological emissivity data obtained in step 33. Effective emissivity data with a mask value of 1 are retained, while emissivity data from all invalid regions are removed, along with outliers greater than 1. After the calculation, preprocessed Ku-frequency vertical polarization climatological emissivity data is obtained, achieving a complete spatial match with the preprocessed brightness temperature data. This avoids inversion errors caused by spatial misalignment of data in subsequent model construction, providing crucial matching emissivity parameters for building a globally applicable land surface temperature estimation model.

[0085] In a preferred embodiment of the present invention, step 44, reprojecting the quality-controlled emissivity data to the same spatial projection system and resolution as the brightness temperature data, to obtain preprocessed brightness temperature and emissivity data, specifically includes:

[0086] Confirm the original spatial resolution of the two sets of core data: Assuming the preprocessing in step 42... The data spatial resolution is 25km (the standard resolution for international spaceborne passive microwave sensors), and the data is preprocessed in step 43. The data spatial resolution is 0.25°, and reprojection sampling is performed to ensure accurate spatial pixel matching. The input data is the data after preprocessing in step 43. Data and the data after preprocessing in step 42 Data (as a resolution matching benchmark).

[0087] Based on the data characteristics used in this embodiment and general data processing requirements, the key resampling parameters are set as follows: The TELSEM2 climatological emissivity data used in this embodiment has an original spatial resolution of 0.25° (approximately 25km, consistent with the conventional brightness temperature resolution of international spaceborne passive microwave sensors), and is compared with the preprocessed data from step 42. The data has a 25km spatial resolution match, so no additional interpolation processing is required for this specific data.

[0088] Considering the universality of the method, if other types of climatological emissivity data are used in practical applications, their original spatial resolution may differ from that of the satellite microwave brightness temperature data (e.g., higher or lower than 25km). In this case, it is necessary to unify the spatial resolution to a spatial scale consistent with the brightness temperature data through reprojection sampling. The core parameters are set as follows: Target spatial resolution: strictly matching the reference data (after preprocessing in step 42). Data (i.e., 25km); Interpolation method: Bilinear interpolation is used. This method calculates the weighted average of the four neighboring valid pixels around the pixel to be interpolated. This method can ensure the spatial smoothness of the emissivity data and avoid abrupt numerical jumps. It can also better preserve the spatial heterogeneity of surface emissivity and fit the spatial distribution law of surface parameters; Boundary processing: The spatial range of the original valid area is strictly preserved during the resampling process. For invalid areas (such as ocean, water bodies, wetlands, snow and ice, and land precipitation areas marked by comprehensive processing masks), they are still uniformly assigned invalid values ​​to ensure that the validity label of the processed data is consistent with the original state.

[0089] Based on the above parameters, the quality control results are... The data underwent bilinear interpolation resampling. After the operation, spatial superposition and comparison were used for verification: it was confirmed that the spatial resolution of the resampled emissivity data accurately matched 25km, the effective area range was completely consistent with the preprocessed brightness temperature data, and the spatial heterogeneity of the original emissivity was not destroyed, thus obtaining the resampled emissivity data.

[0090] Step 45: Based on the preprocessed brightness temperature and emissivity data, construct a globally applicable land surface temperature estimation model to calculate the global-scale land surface temperature distribution, specifically including:

[0091] Based on the core linear relationship obtained in step 12: without considering atmospheric effects or after completing atmospheric effect correction, the linear relationship between the brightness temperature at satellite-observed frequency f and the surface temperature is as follows: (in the formula) The brightness temperature is at frequency f. For frequency f, (For surface temperature). In accordance with the technical solution of this invention, the f frequency is the Ku frequency and the polarization is vertical polarization; therefore, this linear relationship needs to be converted to... The expression for the dependent variable serves as the core form of the estimation model.

[0092] For linear relations After shifting and transforming the terms, we obtain the formula for estimating land surface temperature: Based on the specific data of this invention, substitute the corresponding parameters: The brightness temperature of the Ku-frequency vertical polarization channel after preprocessing in step 42 ( ), The resampled Ku-frequency vertical polarization climatological emissivity ( The final surface temperature estimation model was determined as follows: .

[0093] The core reason for the model's global applicability is that the input brightness temperature data comes from the Ku-frequency vertical polarization channel, which is insensitive to atmospheric influences and has low sensitivity to changes in soil moisture content. The physical relationship between brightness temperature and surface temperature is stable (overcoming the shortcomings of traditional Ka-frequency channels). The input emissivity data is global monthly climatological data, which is assumed to have relatively stable climatological characteristics over a certain time scale and global spatial heterogeneity, making it adaptable to different regional underlying surface characteristics. The data preprocessing process has eliminated interference from non-target features and extreme weather, ensuring the reliability of the input data. Therefore, the model can meet the requirements for global-scale surface temperature retrieval.

[0094] The input data consists of two types of matching data: preprocessed Ku-frequency vertical polarization channel brightness temperature data ( Spatial resolution 25km, covering only non-precipitated land areas globally; resampled Ku-frequency vertical polarization emissivity data ( (The spatial resolution is 25km, and the range perfectly matches the brightness temperature data). Before calculation, the spatial coordinate systems of the two types of data (such as the WGS84 coordinate system) need to be unified to ensure that the pixel positions correspond accurately and avoid calculation misalignment.

[0095] Based on the constructed estimation model A pixel-by-pixel calculation tool for remote sensing imagery is used to perform a division operation on each valid pixel (non-invalid value): using the pixel's... value divided by the corresponding The value is used to obtain the estimated surface temperature of the pixel; invalid pixels (non-land areas and land precipitation areas) retain their invalid values ​​and are not included in the calculation.

[0096] After the calculation is completed, a global-scale land surface temperature distribution dataset is generated. This dataset has three main characteristics: a spatial resolution of 25km, consistent with the input brightness temperature data; a spatial range covering the global land non-precipitation area, fully covering all land areas except for oceans, water bodies, wetlands, ice and snow, and precipitation areas; and data values ​​that are the estimated land surface temperature results for each pixel, preserving the spatial heterogeneity of global land surface temperature.

[0097] In a preferred embodiment of the present invention, step 5, correcting the global-scale surface temperature distribution to finally obtain the true surface temperature, includes:

[0098] Step 51: Based on the selected Ku frequency's certain degree of penetration into the Earth's surface, analyze and confirm the global-scale surface temperature distribution data. The surface temperature distribution data reflects the soil temperature at depths below the surface. Specifically, this includes: analyzing the physical characteristics of the Ku frequency; the Ku frequency has a certain penetrating ability into soil, which is an inherent property of passive microwave bands (the lower the frequency, the stronger the penetrating ability into the medium; the Ku frequency has stronger penetrating ability than the Ka frequency, and moderate penetrating ability compared to the L frequency); the input data is the calculated global-scale surface temperature distribution data (denoted as...). This data is obtained through the model. The initial inversion result obtained from the calculation.

[0099] Based on the principles of passive microwave remote sensing, an in-depth analysis was conducted. The physical significance: During transmission, Ku-frequency microwave signals do not only interact with the surface layer (0-1mm), but also penetrate the surface soil and interact with the soil layer at a certain depth below the surface (experiments have shown that the penetration depth of Ku-frequency vertically polarized signals in dry soil is approximately 5-10cm, and in moist soil, approximately 2-5cm). Therefore, the brightness temperature and emissivity calculations yield... Essentially, it reflects the average temperature of the soil layer within that penetration depth range.

[0100] Based on the above analysis, and combined with a small amount of measured data for verification (selecting typical underlying surfaces such as bare soil, grassland, and desert, and simultaneously observing surface skin temperature and soil temperature at a depth of 5-10 cm, and comparing the results), it was found that... Correlation with soil layer temperature (R) 2 The correlation coefficient (≥0.85) was significantly higher than that with surface skin temperature (R²≤0.6), ultimately confirming that the physical attribute of the output global-scale surface temperature distribution data is the soil layer temperature at the corresponding penetration depth below the surface. This physical definition bias needs to be eliminated through subsequent correction processing to obtain the true surface temperature.

[0101] Step 52: Correct the global-scale surface temperature distribution data. Using land cover type data, surface soil moisture content data, and surface skin temperature data, establish a temperature conversion model including scaling and correction factors. Use this model to convert soil temperature to the true surface temperature. Specifically, this includes a technical solution for Ku-frequency penetration correction, i.e., eliminating bias caused by Ku-frequency penetration by constructing a correction relationship between soil temperature and surface skin temperature. Input data includes: output... (Soil temperature data); Global land cover type data (used to distinguish different underlying surface types, such as bare soil, grassland, desert, farmland, etc. Different underlying surfaces have different soil textures and vegetation cover, and there are significant differences in Ku frequency penetration depth and temperature); Global surface soil moisture content data (used to refine correction parameters. The higher the soil moisture content, the shallower the Ku frequency penetration depth, and the smaller the difference between soil temperature and surface skin temperature).

[0102] Based on a large amount of measured sample data (covering different climate zones and underlying surface types globally, simultaneously collecting Ku-frequency brightness temperature, surface skin temperature, 5-10cm soil layer temperature, and surface soil moisture content data), a calibration model was constructed for each underlying surface type. Specifically, for each type of land cover (e.g., bare soil, grassland, desert, farmland), a model for actual surface skin temperature was established. ) = Soil layer temperature ( The correction relationship is calculated as (m) × scaling factor (m) + correction factor (k), where the scaling factor (m) and correction factor (k) are fixed values ​​obtained by fitting measured data (further optimized by combining soil moisture content: for example, in bare soil areas, the scaling factor m and correction factor k are fitted respectively when soil moisture content is ≤10%, 10%-20%, 20%-30%, 30%-40%, 40%-50%, and ≥50%). This coefficient essentially reflects the average deviation between soil layer temperature and surface skin temperature under different underlying surfaces and different moisture contents.

[0103] Combine global land cover type data, global topsoil moisture content data and Spatial overlay matching is performed to ensure that each pixel matches the corresponding underlying surface type and soil moisture content level. Secondly, based on the matching results, a scaling factor *m* and a correction factor *k* are fitted to each pixel. Finally, a pixel-by-pixel correction operation is performed: using the pixel's... The value is multiplied by the corresponding scaling factor m and then added to the corresponding correction factor k to obtain the true skin temperature value of that pixel. For invalid pixels (non-land areas, areas with terrestrial precipitation), the invalid value is left unchanged and is not included in the correction calculation.

[0104] After the correction calculation is completed, global-scale true surface temperature distribution data is generated. This data completely solves the long-standing problem of physical definition deviation in passive microwave remote sensing inversion of surface temperature, and realizes accurate conversion from volume temperature (soil layer temperature) to surface temperature (surface skin temperature). The data accuracy has been verified by field measurements (the root mean square error RMSE between the corrected true surface temperature and the measured value is ≤1.8℃), which can meet the high-precision application needs of global ecological environment monitoring and assessment.

[0105] like Figure 2As shown, the satellite remote sensing estimation system for land surface temperature based on passive microwave single-channel brightness temperature includes:

[0106] The derivation module is used to derive the approximate passive microwave radiative transfer equation and establish a simplified relationship between microwave brightness temperature and surface temperature.

[0107] The analysis module is used to select the brightness temperature of the Ku-frequency vertical polarization channel as single-channel input data based on the approximate simplification conditions of the derived approximate passive microwave radiative transfer equation, as well as the analysis results of the sensitivity of microwave brightness temperature to atmospheric influence and the sensitivity of surface temperature and surface emissivity changes.

[0108] The acquisition module is used to obtain the Ku-frequency vertical polarization climatological emissivity with global spatial heterogeneity characteristics based on the global monthly microwave emissivity climatological dataset and the brightness temperature of the selected Ku-frequency vertical polarization channel.

[0109] The processing module performs quality control on the brightness temperature of the Ku-frequency vertical polarization channel and the Ku-frequency vertical polarization climatological emissivity, which has global spatial heterogeneity characteristics. It also reprojects and samples the emissivity data to the same spatial projection and resolution as the brightness temperature data to obtain preprocessed brightness temperature and emissivity data. Based on the preprocessed brightness temperature and emissivity data, a globally universal surface temperature estimation model is constructed to calculate the global-scale surface temperature distribution.

[0110] The correction module is used to correct the global-scale surface temperature distribution to obtain the true surface temperature.

[0111] A computing device, comprising:

[0112] One or more processors;

[0113] A storage device for storing one or more programs that, when executed by one or more processors, cause the one or more processors to implement the method.

[0114] A computer-readable storage medium storing a program that, when executed by a processor, implements the method.

[0115] I. Experimental Background and Objectives

[0116] This embodiment uses the Microwave Radiation Imager (MWRI) onboard the FY-3D satellite as an observation platform to verify the feasibility and accuracy of a satellite remote sensing estimation method for land surface temperature based on passive microwave single-channel brightness temperature. The core objective of the experiment is to obtain the true land surface temperature of non-precipitated areas globally by selecting a specific single-channel brightness temperature from FY-3D / MWRI, combining it with global monthly emissivity climatological data, and performing data preprocessing, model calculations, and penetration correction. This addresses the problems of poor universality and physical definition bias in traditional single-channel models.

[0117] II. Experimental Materials and Equipment

[0118] 1. Satellite sensor data: FY-3D / MWRI multi-frequency dual-polarization brightness temperature product, time range of July 2023 (typical summer month), spatial resolution of 25km, including vertical V and horizontal H polarization brightness temperature data at multiple frequencies such as 10.65GHz, 18.7GHz, 23.8GHz, 36.5GHz, and 89GHz.

[0119] 2. Emissivity Dataset: The ECMWF released the global monthly microwave emissivity climatological dataset TELSEM2, which contains vertical or horizontal polarization emissivity data at frequencies such as 19.35 GHz, 22.35 GHz, 37.0 GHz, and 85.0 GHz, covering the period from January to December, with a spatial resolution of 0.25° (approximately 25 km).

[0120] 3. Auxiliary Datasets: MODIS global land cover type data (product model MCD12Q1, spatial resolution 500m), resampled to 25km spatial resolution, using the grid mode calculation method, with the land cover type having the highest proportion within the 25km grid as the resampled type, used to identify areas such as water bodies, permanent wetlands, and snow / ice; IMERG global three-hourly precipitation data (product model IMERGFinalRun, spatial resolution 0.1°), resampled to 25km spatial resolution, using the grid mean calculation method, used to identify precipitation areas; MODIS global surface temperature data (product model MYD11C1, spatial resolution 0.05°), used to fit the scaling factor m and correction factor k of the surface temperature correction model. Actual ground surface temperature measurement data based on four-component radiation sensors, used for surface temperature accuracy verification.

[0121] 4. Data processing software: ENVI 5.6 (for brightness temperature and emissivity data preprocessing), ArcGIS 10.8 (for spatial overlay analysis), MATLAB R2022b (for model calculation and correction).

[0122] III. Experimental Procedure

[0123] Step 1: Derive the approximate passive microwave radiation transfer equation and establish a linear relationship between brightness temperature and surface temperature.

[0124] Based on the observation characteristics of the FY-3D / MWRI sensor and the principles of passive microwave remote sensing, the following assumptions are made: the atmosphere is a homogeneous, non-scattering medium; atmospheric upward and downward radiation are approximately equal; and the temperature of the vegetation canopy is approximately equal to the surface temperature, with vegetation influence temporarily neglected. Based on the Rayleigh-Jones approximation of Planck's law, an approximate formula for the passive microwave radiative transfer equation is derived: This means that the brightness temperature at each frequency of MWRI has a linear relationship with the surface temperature, laying the foundation for subsequent channel selection and model construction.

[0125] Step 2: Select a single-channel brightness temperature that meets the requirements of FY-3D / MWRI.

[0126] Based on the linear relationship established in step 1, and combined with the four channel selection criteria, all frequency polarization channels of FY-3D / MWRI were analyzed:

[0127] 1. Conditional Matching Analysis: Insensitive to atmospheric influences: Exclude 89GHz (W frequency, high atmospheric sensitivity), retain Ka and lower frequency channels such as 10.65GHz (C frequency), 18.7GHz (K frequency), 23.8GHz (K frequency), and 36.5GHz (Ka frequency); Sensitive to surface temperature: The correlation between brightness temperature of vertically polarized channels and surface temperature (R²=0.82) is significantly higher than that of horizontally polarized channels (R²=0.82). 2 =0.65), retaining the 10.65GHz (C frequency), 18.7GHz (K frequency), 23.8GHz (K frequency), and 36.5GHz (Ka frequency) vertical polarization channels; moderate penetration: excluding the 10.65GHz (C frequency, too strong penetration, easily detects deep soil); insensitive to soil moisture content: also excluding the 10.65GHz (C frequency). The remaining 18.7GHz (K frequency), 23.8GHz (K frequency), and 36.5GHz (Ka frequency) vertical polarization channels have screening value for single-channel inversion of land surface temperature. Further comparing the sensitivity of the 18.7GHz (K frequency), 23.8GHz (K frequency), and 36.5GHz (Ka frequency) vertical polarization channels to atmospheric influence, excluding the 23.8GHz (K frequency) and 36.5GHz (Ka frequency) vertical polarization channels, finally retaining the 18.7GHz (K frequency) vertical polarization channel.

[0128] 2. Final Channel Determination: Based on the above analysis, the 18.7GHz vertical polarization brightness temperature of FY-3D / MWRI (denoted as ) was selected. As a single-channel input data, this channel fully meets all the screening criteria and is the core observation channel of FY-3D / MWRI, with strong data continuity and stability.

[0129] Step 3: Obtain the climatological emissivity with global spatial heterogeneity

[0130] 1. Dataset selection: The ECMWF TELSEM2 global monthly microwave emissivity climatological dataset was adopted. This dataset covers the globe and includes multi-year monthly average emissivity with multiple frequency polarizations from 1993 to 2000 (such as emissivity with 19.35 GHz vertical and horizontal polarization, 22.3 GHz vertical polarization, 37.0 GHz vertical and horizontal polarization, and 85.0 GHz vertical and horizontal polarization). The time length (8 years) and time resolution (monthly scale) are consistent with the assumption that the climatological characteristics of emissivity are relatively stable.

[0131] 2. Emissivity Extraction: For the 18.7 GHz vertical polarization channel selected in step 2, the closest vertical polarization emissivity at 19.35 GHz was selected from the TELSEM2 dataset (the frequency difference between the two is only 0.65 GHz, and the emissivity difference is ≤0.02, meeting the accuracy requirements); combined with the experimental time in step 1 (July 2023), the 8-year average July climatological emissivity of the 19.35 GHz vertical polarization from 1993 to 2000 was extracted from TELSEM2 (denoted as ). This data can fully reflect the spatial heterogeneity of emissivity of different underlying surfaces (deserts, grasslands, forests, etc.) around the world.

[0132] Step 4: Data preprocessing, model building, and global temperature distribution calculation

[0133] Step 41: Extract land area brightness temperature data

[0134] Load MODIS land cover data MCD12Q1, filter by attribute to mark water bodies, wetlands, and snow cover areas (assigned a value of 0), and assign a value of 1 to other land areas (forests, grasslands, deserts, etc.), generating a land mask; then compare this mask with... Data overlay and processing, eliminating invalid regions, yields the global land area. data.

[0135] Step 42: Filter brightness temperature data for non-precipitation areas

[0136] Load IMERG three-hourly precipitation data, select and... The data from the time period closest to the observation time (July 15, 2023, 12:00-15:00) were used to set a precipitation threshold of 0.1 mm / h, marking precipitation areas (assigned a value of 0) and non-precipitation areas (assigned a value of 1) to generate a precipitation mask. This mask was then overlaid with the land brightness temperature data obtained in step 41, and the precipitation areas were removed to obtain the preprocessed brightness temperature data (denoted as ). ).

[0137] Step 43: Preprocess emissivity data

[0138] The land mask from step 41 and the precipitation mask from step 42 are merged to generate a composite mask. The data is masked to remove invalid areas such as water bodies, ice and snow, and precipitation, and outliers with emissivity greater than 1 are removed, resulting in preprocessed emissivity data (denoted as ). ).

[0139] Step 44: Emissivity data reprojection sampling

[0140] Through projection transformation and bilinear interpolation, Reprojection sampling to With spatial projection and spatial resolution kept consistent, the emissivity data after reprojection sampling is obtained (denoted as ). This ensures precise spatial pixel matching between the two.

[0141] Step 45: Construct a globally applicable land surface temperature estimation model based on the preprocessed brightness temperature and emissivity data, and calculate the global-scale land surface temperature distribution.

[0142] Based on the simplified linear relationship from step 1, (correspond )and (correspond Substitute the values ​​into the model and construct the estimation model: This model has global applicability and can be adapted to different non-precipitation land areas.

[0143] Will and Inputting the above model, and using MATLAB pixel-by-pixel calculations, we obtained the global land surface temperature distribution data for non-precipitated areas in July 2023 (denoted as ). With a spatial resolution of 25km, it effectively covers all land areas globally except for water bodies, ice and snow, and precipitation areas.

[0144] Step 5: Correct the temperature distribution data to obtain the true surface temperature.

[0145] Step 51: Confirm temperature data attributes

[0146] 18.7GHz belongs to the Ku frequency range, which has a certain degree of penetration into soil (actual measurements show a penetration depth of approximately 8cm in dry desert soil and approximately 3cm in moist grassland soil). (Comparison) Compared with measured data (simultaneous observation of soil temperature at 8cm / 3cm depth and surface skin temperature), it was found that... The correlation with soil temperature (R²=0.86) was significantly higher than that with surface skin temperature (R²=0.58), confirming... The temperature of the soil layer at the corresponding depth below the surface needs to be corrected for penetration.

[0147] Step 52: Perform penetration correction

[0148] Based on global land cover type and topsoil moisture content data (ERA5-Land topsoil moisture content product), a calibration model is constructed for different underlying surface types and soil moisture content isopleths: Where m is the scaling factor and k is the correction factor. Substituting the data into the model for pixel-by-pixel correction, the final result is the true surface temperature distribution data for non-precipitated land areas globally in July 2023 (denoted as ). ).

[0149] IV. Verification of Experimental Results

[0150] We selected actual ground surface temperature measurement data based on four-component radiation sensors and compared the accuracy of temperature data before and after correction: Before correction The root mean square error (RMSE) between the measured value and the actual value is 3.2℃. After correction... The RMSE was reduced to 1.7℃, meeting the high-precision requirements of global ecological environment monitoring and assessment, and verifying the effectiveness and reliability of this method.

[0151] It should be noted that this system is a system corresponding to the above method. All implementation methods in the above method embodiments are applicable to this embodiment and can achieve the same technical effect.

[0152] The above description represents the preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A satellite remote sensing estimation method for land surface temperature based on passive microwave single-channel brightness temperature, characterized in that, The method includes: Step 1: Derive the approximate passive microwave radiation transfer equation and establish a simplified relationship between microwave brightness temperature and surface temperature. Step 2: Based on the approximate simplification conditions of the derived approximate passive microwave radiative transfer equation, and the analysis results of the sensitivity of microwave brightness temperature to atmospheric influence and the sensitivity of surface temperature and surface emissivity changes, the brightness temperature of the Ku frequency vertical polarization channel is selected as the single-channel input data. Step 3: Based on the global monthly microwave emissivity climatological dataset, obtain the Ku-frequency vertical polarization climatological emissivity with global spatial heterogeneity characteristics for the selected Ku-frequency vertical polarization channel brightness temperature. Step 4: Perform quality control on the brightness temperature of the Ku-frequency vertical polarization channel and the Ku-frequency vertical polarization climatological emissivity, which has global spatial heterogeneity characteristics. Reproject and sample the emissivity data to the same spatial projection and resolution as the brightness temperature data to obtain preprocessed brightness temperature and emissivity data. Based on the preprocessed brightness temperature and emissivity data, construct a globally universal surface temperature estimation model to calculate the global-scale surface temperature distribution. Step 5: Correct the global-scale surface temperature distribution to obtain the true surface temperature.

2. The satellite remote sensing estimation method for surface temperature based on passive microwave single-channel brightness temperature according to claim 1, characterized in that, Step 1: Derive the approximate passive microwave radiative transfer equation and establish a simplified relationship between microwave brightness temperature and surface temperature, including: Step 11: Based on Rayleigh-Jones' law and Planck's radiation law, for the soil-vegetation-atmosphere surface, the vegetation is regarded as a single scattering layer on the rough soil, and the approximate passive microwave radiation transfer equation is derived. Step 12: For the derived approximate passive microwave radiation transfer equation, after ignoring atmospheric influence or correcting for atmospheric influence, it is assumed that the vegetation canopy temperature and the surface temperature are approximately equal. The physical relationship between microwave brightness temperature and surface temperature is simplified into a linear expression to establish a simplified relationship between microwave brightness temperature and surface temperature.

3. The satellite remote sensing estimation method for surface temperature based on passive microwave single-channel brightness temperature according to claim 2, characterized in that, Step 2: Based on the approximate simplification conditions of the derived approximate passive microwave radiative transfer equation, and the analysis results of the sensitivity of microwave brightness temperature to atmospheric influence and the sensitivity of changes in surface temperature and surface emissivity, the brightness temperature of the Ku-frequency vertical polarization channel is selected as the single-channel input data, including: Step 21: Based on the derived approximate passive microwave radiative transfer equation, and considering the linear relationship between passive microwave brightness temperature and surface temperature, as well as the close correlation between surface temperature estimation and surface emissivity changes, analyze the approximate simplification preconditions of the passive microwave radiative transfer equation, the sensitivity of passive microwave brightness temperature to surface temperature, and the sensitivity of surface emissivity changes to obtain the analysis results. Step 22: Based on the analysis results, select vertical polarization channels that are not sensitive to atmospheric influences but are sensitive to surface temperature from the relatively low frequency channels at or below the Ka frequency, while excluding frequency channels with excessive penetration and those sensitive to changes in soil moisture content, in order to determine the candidate channel set. Step 23: From the candidate channel set, analyze and compare the differences in the degree of atmospheric influence on each channel, and finally select the brightness temperature of the Ku frequency vertical polarization channel as the single channel input data.

4. The satellite remote sensing estimation method for surface temperature based on passive microwave single-channel brightness temperature according to claim 3, characterized in that, Step 3: Based on the global monthly microwave emissivity climatological dataset, for the selected Ku-frequency vertical polarization channel brightness temperature, obtain the Ku-frequency vertical polarization climatological emissivity with global spatial heterogeneity characteristics, including: Step 31: Select a globally covered monthly microwave emissivity climatological dataset. The monthly microwave emissivity climatological dataset contains emissivity climatological information for multiple frequencies and polarizations. Step 32: Based on the finally selected Ku frequency vertical polarization channel, extract the monthly average emissivity data corresponding to the Ku frequency vertical polarization from the global monthly microwave emissivity climatological dataset; Step 33: Based on the month corresponding to the current observation time, select the emission rate data of the corresponding month from the extracted monthly average emission rate data of Ku frequency vertical polarization to obtain the Ku frequency vertical polarization climatological emission rate with global spatial heterogeneity characteristics.

5. The satellite remote sensing estimation method for surface temperature based on passive microwave single-channel brightness temperature according to claim 4, characterized in that, Step 4: Perform quality control on the brightness temperature and climatological emissivity of the Ku-frequency vertical polarization channel, which exhibits global spatial heterogeneity. Reproject and sample the emissivity data to the same spatial projection and resolution as the brightness temperature data to obtain preprocessed brightness temperature and emissivity data. Based on this preprocessed data, construct a globally applicable land surface temperature estimation model to calculate the global-scale land surface temperature distribution, including: Step 41: Based on global land cover type information, identify and remove ocean, inland water, wetland, and snow cover areas, and extract Ku frequency vertical polarization channel brightness temperature data within the land surface range; Step 42: Based on global atmospheric precipitation information, identify and remove precipitation areas, and filter out the Ku frequency vertical polarization channel brightness temperature data of non-precipitation areas from the land surface to obtain preprocessed brightness temperature data. Step 43: For Ku frequency vertical polarization climatological emissivity data with global spatial heterogeneity, geographic masking is used to remove ocean, inland water, wetland, snow and ice covered areas and precipitation areas, and outliers with emissivity greater than 1 are removed to obtain quality-controlled emissivity data. Step 44: Reproject the quality-controlled emissivity data to the same spatial projection system and resolution as the brightness temperature data to obtain the preprocessed brightness temperature and emissivity data; Step 45: Based on the preprocessed brightness temperature and emissivity data, construct a globally applicable land surface temperature estimation model and calculate the global-scale land surface temperature distribution.

6. The satellite remote sensing estimation method for surface temperature based on passive microwave single-channel brightness temperature according to claim 5, characterized in that, Step 5, correct the global-scale surface temperature distribution to obtain the final true surface temperature, including: Step 51: Based on the selected Ku frequency, analyze and confirm the global-scale surface temperature distribution data; the surface temperature distribution data reflects the soil temperature at depths below the surface. Step 52: Correct the global-scale surface temperature distribution data. Using surface cover type data, surface soil moisture content data, and surface skin temperature data, establish a temperature conversion model that includes scaling and correction factors. Use the temperature conversion model to convert the soil layer temperature into the true surface temperature pixel by pixel.

7. A satellite remote sensing estimation system for land surface temperature based on passive microwave single-channel brightness temperature, wherein the system implements the method as described in any one of claims 1 to 6, characterized in that, include: The derivation module is used to derive the approximate passive microwave radiative transfer equation and establish a simplified relationship between microwave brightness temperature and surface temperature. The analysis module is used to select the brightness temperature of the Ku-frequency vertical polarization channel as single-channel input data based on the approximate simplification conditions of the approximate passive microwave radiative transfer equation, as well as the analysis results of the sensitivity of microwave brightness temperature to atmospheric influence and the sensitivity of surface temperature and surface emissivity changes. The acquisition module is used to obtain the Ku-frequency vertical polarization climatological emissivity with global spatial heterogeneity characteristics based on the global monthly microwave emissivity climatological dataset and the brightness temperature of the selected Ku-frequency vertical polarization channel. The processing module performs quality control on the brightness temperature of the Ku-frequency vertical polarization channel and the Ku-frequency vertical polarization climatological emissivity, which has global spatial heterogeneity characteristics. It also reprojects and samples the emissivity data to the same spatial projection and resolution as the brightness temperature data to obtain preprocessed brightness temperature and emissivity data. Based on the preprocessed brightness temperature and emissivity data, a globally universal surface temperature estimation model is constructed to calculate the global-scale surface temperature distribution. The correction module is used to correct the global-scale surface temperature distribution to obtain the true surface temperature.

8. A computing device, characterized in that, include: One or more processors; A storage device for storing one or more programs that, when executed by one or more processors, cause the one or more processors to implement the method as described in any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a program that, when executed by a processor, implements the method as described in any one of claims 1 to 6.

Images