A Spatiotemporally Continuous Quantification Method for Surface Temperature Anomaly Variation

Abstract

This invention relates to the field of surface temperature anomaly feature extraction and spatiotemporal change monitoring of the thermal environment. Specifically, it relates to a method for quantifying spatiotemporally continuous surface temperature anomaly change features, comprising: acquiring regional remote sensing instantaneous surface temperature data and constructing a remote sensing daily average surface temperature based on ground station observation data; constructing a multi-timescale remote sensing surface temperature meteorological index based on the remote sensing daily average surface temperature; performing spatial clustering based on the remote sensing surface temperature meteorological index and combining it with spatiotemporal correlation to identify spatiotemporal clusters of surface temperature anomalies; and constructing a surface temperature hotspot evolution model based on the spatiotemporal evolution law of the spatiotemporal clusters to quantify the spatiotemporally continuous features of the surface temperature anomalies. This invention achieves the identification from discrete anomaly points to continuous anomaly regions and provides a complete and dynamic quantitative analysis of the spatiotemporal evolution features of anomaly regions, improving the accuracy and continuity of monitoring.

Description

Technical Field

[0001] This invention relates to the field of surface temperature anomaly feature extraction and thermal environment spatiotemporal change monitoring technology, specifically to a spatiotemporally continuous surface temperature anomaly change feature quantification method. Background Technology

[0002] Against the backdrop of increasingly severe global climate change, extreme weather events such as heat waves and cold waves are occurring more frequently. As a key physical quantity directly reflecting the energy balance of the Earth's surface, abnormal changes in surface temperature have a profound impact on the stability of regional ecosystems, agricultural production safety, and even public health. Therefore, rapid and accurate monitoring and quantitative analysis of the spatiotemporal distribution and evolution patterns of surface temperature are of great scientific value and practical significance for a deeper understanding of regional climate response mechanisms, assessment of environmental dynamics, and guidance for the formulation of disaster prevention and mitigation strategies. This is an important research direction in the current field of Earth science and environmental monitoring. Currently, there are significant limitations in quantifying the characteristics of abnormal surface temperature changes. On the one hand, at the data source level, traditional methods for estimating the emissivity of complex and heterogeneous surfaces often contain biases, resulting in insufficient accuracy of basic remote sensing surface temperature data and introducing uncertainty into subsequent analysis from the outset. On the other hand, in terms of analytical methods, most studies adopt statistical discrete point hotspot analysis. This approach not only severs the continuity and topological correlation that anomalous events should have in the spatiotemporal dimensions, but also makes it difficult to effectively characterize the complete dynamic process of the generation, movement and dissipation of anomalous regions, resulting in a one-sided and static understanding of the evolution of climate events. Summary of the Invention

[0003] The purpose of this invention is to provide a method for quantifying the spatiotemporally continuous characteristics of surface temperature anomalies. This method enables the identification of discrete anomaly points into continuous anomaly regions and performs a complete and dynamic quantitative analysis of the spatiotemporal evolution characteristics of the anomaly regions, thereby improving the accuracy and continuity of monitoring.

[0004] This invention is achieved through the following technical solution:

[0005] A method for quantifying spatiotemporally continuous surface temperature anomaly characteristics, comprising the following steps:

[0006] Acquire instantaneous surface temperature data of the region using remote sensing, and construct the daily average surface temperature using remote sensing based on ground station observation data;

[0007] Based on the aforementioned remotely sensed daily average land surface temperature, a multi-timescale remotely sensed land surface temperature meteorological index is constructed.

[0008] Spatial clustering is performed based on the remotely sensed surface temperature meteorological indicators, and spatiotemporal correlation is combined to identify spatiotemporal clusters of abnormal surface temperature changes.

[0009] Based on the spatiotemporal evolution patterns of the aforementioned spatiotemporal clusters, a model for the evolution of hot and cold spots in surface temperature is constructed to quantify the spatiotemporal continuity characteristics of the anomalous changes in surface temperature.

[0010] Optionally, the construction of the remote sensing daily average land surface temperature specifically involves:

[0011] Based on the observations of longwave radiation from the Earth's surface, a statistical relationship was established between the daily average and instantaneous values ​​of surface temperature observed on the ground.

[0012] The statistical relationship is mapped to instantaneous remote sensing surface temperature data to calculate the daily average remote sensing surface temperature.

[0013] Optionally, before establishing the statistical relationship, the calculation of the broadband emissivity of the earth's surface is also included, specifically as follows:

[0014] Based on sub-pixel scale land use classification data, the types and proportions of building components within observed pixels are obtained;

[0015] Based on a linear mixing model, the broadband emissivity of pixels is calculated by utilizing the proportions of local landform components and their corresponding bulk emissivity.

[0016] Optionally, the establishment of a statistical relationship between the daily average and instantaneous values ​​of surface temperature observed from ground observations specifically includes:

[0017] Based on the Stefan-Boltzmann law, using surface upward longwave radiation, surface downward longwave radiation, and broadband emissivity, the following calculations are performed. The instantaneous value of the ground surface temperature at a given time is calculated using the following formula:

[0018]

[0019]

[0020] in, It is the Stefan-Boltzmann constant. It is longwave radiation rising above the Earth's surface. It is long-wave radiation flowing down to the Earth's surface. This refers to the instantaneous value of the Earth's surface temperature as observed from the ground. For the broadband emissivity of the mixed pixels, subscript Representing each component, The proportions of each component satisfy the following conditions: and , The bulk emissivity of each component;

[0021] For all times within a single day The arithmetic mean was calculated to obtain the daily average surface temperature observed on the ground. ;

[0022] Establish a linear relationship model between the daily average surface temperature observed from ground observations and the instantaneous surface temperature at two times of day and night:

[0023]

[0024] Among them, subscript , These are the daytime and nighttime satellite transit times corresponding to this observation station. and These are the instantaneous ground surface temperature values ​​observed during the day and night, respectively. and The number of valid observations during the day and night, respectively. and These are the day and night coefficients for ground stations, respectively. This is a constant term.

[0025] Optionally, the construction of multi-timescale remote sensing surface temperature meteorological indicators specifically includes:

[0026] Calculate the remote sensing monthly average land surface temperature and the remote sensing annual average land surface temperature based on the aforementioned remote sensing daily average land surface temperature;

[0027] The surface temperature anomalies at the daily, monthly, and annual scales are obtained by subtracting the arithmetic mean of the corresponding time series from the surface temperature at the daily, monthly, and annual scales, and are used as the remote sensing surface temperature meteorological indicators.

[0028] Optionally, the identification of spatiotemporal clusters of abnormal surface temperature changes specifically includes:

[0029] The Getis-Ord Gi* hotspot analysis model was used to spatially cluster remote sensing surface temperature meteorological indicators at specific time points to obtain discrete surface temperature hotspots and cold spots.

[0030] By overlaying surface temperature hot and cold spot markers at different time points, a spatiotemporal three-dimensional data cube is constructed.

[0031] The spatiotemporal connectivity component search method is used to search for a set of pixels with spatiotemporal connectivity in the spatiotemporal three-dimensional data cube to determine the spatiotemporal clustering region.

[0032] Optionally, the Getis-Ord Gi* hotspot analysis model has the following specific calculation formula:

[0033]

[0034] Among them, subscript and Representing the center pixel and neighboring pixels respectively. This is the weight matrix. , Center pixel and neighboring pixels The distance between them Center pixel The surrounding area The surface temperature value of each neighboring pixel. Center pixel Surrounding area Local weighted sum of surface temperature values ​​of neighboring pixels, Represents the number of neighboring pixels. The global mean. , For neighboring pixels The surface temperature value, The global standard deviation, , for The score, with positive and negative values ​​representing high-value clusters and low-value clusters respectively, corresponds to hot and cold spots of surface temperature.

[0035] Optionally, the quantification of the spatiotemporal continuity characteristics of the surface temperature anomaly specifically includes:

[0036] Targeting the identified spatiotemporal clusters, we extract their average intensity, total area, total magnitude, or frequency of occurrence as time series indicators.

[0037] The time series index is set to change over time. The changes are linear. The slope of the linear change of the long-term series index for n samples is calculated using the least squares fitting method. The calculation formula is as follows:

[0038]

[0039] in, Indicators representing average intensity, total area, total magnitude, and frequency of occurrence. The slope represents the linear change, characterizing the quantified evolution trend. From 1 to A set of integers, with index Representative set Each sample in the dataset.

[0040] Optionally, the constraint condition for quantifying the spatiotemporal continuity characteristics of the surface temperature anomaly changes is:

[0041] Quantification is only performed on spatiotemporal clusters whose cumulative area exceeds the set value of the total area of ​​the study area and whose frequency of occurrence is greater than the set number of times.

[0042] Optionally, quantifying the spatiotemporal continuity characteristics of the surface temperature anomaly further includes:

[0043] Based on the linear change slope The size and significance level of the evolution characteristics of the spatiotemporal cluster are used to classify the evolution characteristics of the spatiotemporal cluster into a preset evolution category. The preset evolution category includes: first rate of heating, second rate of heating, second rate of cooling, first rate of cooling and temperature stabilization, wherein the first rate is greater than the second rate.

[0044] The technical solution of the present invention has at least the following advantages and beneficial effects:

[0045] This invention significantly improves the accuracy of basic remote sensing daily average land surface temperature data by introducing mixed pixel broadband emissivity and constructing statistical relationships based on ground observation data. Furthermore, it innovatively utilizes spatiotemporal data cubes and connected component search techniques to intelligently identify spatiotemporally continuous anomalous clusters from discrete anomalies, reconstructing the dynamic process of events. Moreover, this invention constructs a multi-dimensional index system integrating intensity, area, and frequency, and calculates its linear variation trend, thereby achieving a complete, dynamic, and quantitative description of the evolution of land surface temperature anomalies, improving the accuracy and continuity of monitoring. Attached Figure Description

[0046] Figure 1 A flowchart illustrating the method for quantifying spatiotemporally continuous surface temperature anomaly variations provided by this invention.

[0047] Figure 2 This is a logical diagram illustrating the identification of spatiotemporal clusters of abnormal surface temperature changes provided by the present invention. Detailed Implementation

[0048] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0049] This invention provides a method for quantifying spatiotemporally continuous surface temperature anomaly variations, which can be executed on a computer device. This computer device can be a personal computer, server, workstation, or a cloud computing platform composed of multiple servers. The device possesses data storage and data processing capabilities to execute the various steps of the method described in this invention.

[0050] Reference Figure 1 As shown, Figure 1 A flowchart illustrating the method for quantifying spatiotemporally continuous surface temperature anomaly changes provided by this invention.

[0051] In one embodiment, this embodiment provides an implementation process for a complete spatiotemporally continuous method for quantifying surface temperature anomaly variation characteristics.

[0052] First, perform step S101: acquire regional remote sensing instantaneous surface temperature data, and construct the remote sensing daily average surface temperature based on ground station observation data.

[0053] In this embodiment, step S101 is the data foundation of the entire method. Its purpose is to construct a high-quality, high spatiotemporal resolution daily average land surface temperature dataset to provide accurate input for subsequent anomaly analysis. This step can be specifically decomposed into two sub-processes: constructing a remote sensing instantaneous land surface temperature database and calculating the remote sensing daily average land surface temperature.

[0054] When constructing the remote sensing instantaneous land surface temperature database, the processing unit first obtains land surface temperature (LST) and sea surface temperature (SST) products for the study area from publicly available or commercial remote sensing data sources. These products typically contain instantaneous temperature observations for both daytime and nighttime. Since images acquired by different satellites or sensors may differ in spatial coverage and imaging time, spatial image stitching and temporal matching are required. In this embodiment, the nearest neighbor resampling method is used. This method unifies images from different sources and at different times into the same geographic coordinate system and spatial resolution, forming a complete image covering the entire study area through spatial stitching, and creating a continuous, long-term time-series remote sensing instantaneous land surface temperature database through temporal matching. This database serves as the direct data source for subsequent calculations of daily average temperature.

[0055] The core idea in calculating the daily average land surface temperature by remote sensing is to use more precise ground station observation data to correct and extrapolate the daily average temperature over a large area. Since satellite transit times are fixed, the instantaneous temperatures they acquire cannot fully represent the average conditions for the entire day, whereas ground radiation stations can conduct continuous observations. Therefore, this embodiment establishes a mapping relationship from ground observation data to remote sensing data.

[0056] The process first requires accurate calculation of the broadband emissivity of the land surface, as it is a key parameter in land surface temperature retrieval and directly affects the accuracy of temperature calculations. In complex surface areas, a single remote sensing pixel may contain multiple land cover types, such as vegetation, bare soil, impermeable surfaces, and water bodies, forming a mixed pixel. To accurately estimate the overall emissivity of such mixed pixels, this embodiment first acquires sub-pixel-scale remote sensing land use classification images, which provide the composition and area proportions of various land cover types within the land surface temperature observation pixel. Assuming that the surfaces of all land cover components within the pixel are Lambertian bodies and that the temperature is uniformly distributed within each component, the broadband emissivity of the mixed pixel can be calculated based on a linear mixing model. Specifically, the broadband emissivity is obtained by considering the proportions of each component. i The emissivity is calculated by weighting the corresponding bulk emissivity. The proportions of each component satisfy the condition that the sum is 1 and each proportion is greater than or equal to 0. Based on prior knowledge, the bulk emissivity of each component can be set as empirical values, for example, 0.985 for vegetation, 0.968 for bare soil, 0.942 for impermeable surfaces, and 0.989 for water bodies. The broadband emissivity calculated in this way can more realistically reflect the radiation characteristics of heterogeneous surfaces.

[0057] The surface temperature is calculated using observational data from ground-based radiation stations. The processing unit acquires observed values ​​of downward and upward longwave radiation from the surface and, based on the Stefan-Boltzmann law, calculates the surface temperature... The instantaneous value of the ground surface temperature observed at a given time. Its calculation formula is: In this formula, For ground observation The instantaneous value of the Earth's surface temperature at that moment. This is the Stefan-Boltzmann constant, with a value of 5.67 × 10⁻⁶. -8 W / m²K 4 , It is longwave radiation rising above the Earth's surface. It is long-wave radiation flowing down to the Earth's surface. For the broadband emissivity of mixed pixels, Subscript Representing each component, The proportions of each component satisfy the following conditions: and , The emissivity of each component is represented by its bulk emissivity. By arithmetically averaging the instantaneous temperatures at all observation times throughout the day, a high-precision daily average of the land surface temperature based on ground observations can be obtained.

[0058] Next, a statistical relationship is established between the daily average and instantaneous values ​​of surface temperature observed from ground-based observations. This step is a crucial bridge connecting ground-based point observations and remote sensing surface observations. This embodiment establishes a linear relationship model that expresses the relationship between the daily average surface temperature and the instantaneous surface temperature at specific times, namely, during the day and night when the satellite passes overhead. This linear relationship model can be expressed as: Among them, subscript , These are the daytime and nighttime satellite transit times corresponding to this observation station. and These are the instantaneous ground surface temperature values ​​observed during the day and night, respectively. and The number of valid observations during the day and night, respectively. and These are the day and night coefficients for ground stations, respectively. This is a constant term.

[0059] Finally, this statistical relationship established at points is extended to the entire study area. The processing unit uses spatial interpolation methods, such as inverse distance weighted interpolation and Kriging interpolation, to process the data calculated at each ground station. , and The values ​​are interpolated to generate coefficient raster maps and constant term raster maps covering the entire study area. Finally, the instantaneous temperature values ​​corresponding to daytime and nighttime in the remote sensing instantaneous surface temperature database are substituted into this spatialized linear relationship model, and the remote sensing daily average surface temperature of the entire study area can be calculated pixel by pixel. The resulting daily average temperature data has both the advantage of wide coverage of remote sensing data and the high precision of ground observation data.

[0060] Step S102: Based on the remotely sensed daily average surface temperature, construct a multi-time-scale remotely sensed surface temperature meteorological index.

[0061] In this embodiment, the purpose of this step is to transform the raw temperature data into a standardized indicator that more effectively reflects anomalous changes. Raw surface temperature values ​​are strongly influenced by background climate factors such as seasonal cycles and interannual fluctuations, making it difficult to determine whether a temperature is "abnormal" by directly comparing temperature values ​​at different times and locations. Therefore, it is necessary to construct a meteorological indicator that can eliminate the influence of background climate and highlight anomalous signals.

[0062] In this embodiment, the processing unit first calculates the remote sensing monthly average land surface temperature and the remote sensing annual average land surface temperature based on the remote sensing daily average land surface temperature dataset generated in step S101 by arithmetically averaging the effective daily average temperature values ​​over a month or a year. This forms three land surface temperature sequences at different time scales: daily, monthly, and annual. Subsequently, to eliminate the influence of climate background and enhance spatiotemporal comparability, this embodiment calculates the anomalies of the land surface temperature. Specifically, the total average temperature over the entire study period is calculated at the daily, monthly, and annual time scales. Then, the temperature value at each moment is subtracted from the total average of the corresponding time series to obtain the remote sensing land surface temperature anomalies at the daily, monthly, and annual scales, respectively. These anomalies represent the degree of deviation of the land surface temperature from its long-term average state; positive values ​​indicate higher temperatures, and negative values ​​indicate lower temperatures. The absolute value reflects the intensity of the deviation. These anomalies constitute the multi-time-scale meteorological indicators of land surface temperature used for subsequent analysis.

[0063] like Figure 2 As shown, step S103 is executed: spatial clustering is performed based on the remote sensing surface temperature meteorological index, and spatiotemporal correlation is combined to identify the spatiotemporal clustering areas of abnormal surface temperature changes.

[0064] In this embodiment, this step aims to identify continuous, contiguous temperature anomaly regions from a spatiotemporal perspective, rather than isolated anomaly points, thus solving the problem of traditional methods fragmenting spatiotemporal continuity. This step can also be decomposed into two sub-processes: spatial clustering to obtain discrete hot and cold spots, and spatiotemporal correlation to identify continuous clustered areas.

[0065] First, spatial clustering is performed to identify discrete hot and cold spots. The processing unit employs the Getis-Ord Gi hotspot analysis model to perform spatial clustering analysis on surface temperature meteorological indicators, i.e., temperature anomalies, at specific time points. The purpose of this model is to determine whether high or low values ​​of a pixel have statistically significant spatial clustering. For each pixel in the study area... Calculate its local value. The formula for calculating the value is: Among them, subscript and Representing the center pixel and neighboring pixels respectively. This is the weight matrix. This is the weight matrix. , Center pixel and neighboring pixels The distance between them Center pixel and surrounding areas Local weighted sum of surface temperature values ​​of neighboring pixels, Represents the number of neighboring pixels. The global mean. , For neighboring pixels The surface temperature value, The global standard deviation, , for The score, with positive and negative values ​​representing high-value clustering and low-value clustering respectively, corresponds to hot and cold spots in surface temperature. Spatial weights. The setting of the weights is crucial to the analysis results, as it quantifies the influence of neighboring pixels on the central pixel. In this embodiment, the inverse distance method is used to quantify the weights, i.e. , Center pixel and neighboring pixels The distance between them means that the closer the pixels are, the greater their weight and the greater their influence. The calculated... The value is a The score, its sign and magnitude, has clear statistical significance. A significantly positive score... A value of indicates that the cell and its neighborhood form a high-value cluster, i.e., a surface temperature hotspot; a significantly negative value indicates a high-value cluster. The value indicates that the pixel and its neighborhood are a low-value cluster, i.e., a cold spot in the surface temperature. The higher the absolute value, the tighter the clustering effect, and the stronger its significance as a cold or hot spot. By calculating for each pixel in the entire study area, a discrete map of cold and hot spots at that point in time can be obtained.

[0066] Subsequently, spatiotemporal correlation is performed to identify continuous clusters. The hotspot analysis described above yields a series of temporally discontinuous "snapshots." To capture the complete lifecycle of anomalous events, these discrete hotspots and cold spots need to be connected in time and space. This embodiment first constructs a spatiotemporal three-dimensional data cube. This cube uses geographic coordinates X and Y as two horizontal dimensions and time T as the vertical dimension, forming a three-dimensional grid (X×Y×T). The processing unit maps the hotspot and cold spot analysis results obtained in the previous step into this cube. If a cell is identified as a hotspot or a cold spot at a certain moment, the corresponding position (x,y,t) in the cube is marked as 1; otherwise, it is marked as 0.

[0067] After constructing the data cube, a spatiotemporal connectivity component search method is used to identify continuous clusters. The core of this method is defining the "connectivity" between pixels. A pixel is associated not only with its neighbors in the same spatial time but also with its "neighbors" at the same spatial location in adjacent time intervals. This embodiment uses a 26-neighborhood connectivity criterion, meaning that for a central pixel in the 3D cube, its surrounding 26 pixels (up, down, left, right, four corners, and corresponding positions in the previous and next time layers) are considered its neighborhood. The search algorithm starts with any pixel marked as 1 in the cube, using it as a seed point. Then, the algorithm searches all 26 neighboring pixels. If a neighboring pixel is also marked as 1, it is assigned to the same connectivity component and marked as visited. Next, using this newly added pixel as the new center, the search for its neighborhood is recursively continued until all pixels spatiotemporally connected to the initial seed point are found and marked. This collection, composed of all spatiotemporally connected pixels, forms a complete and spatiotemporally continuous region of abnormal surface temperature changes. .in, The values ​​are used to label connected components, with 1 and 0 representing connected and disconnected states, respectively, and the subscripts are used for the labels. Represents spatial location ,time The pixels. Spatial topology and temporal correlation are determined for the above data cube, pixels. The neighborhood includes spatial neighbors at the same time. Time neighbors of adjacent times The connectivity of the 26-neighborhood can be determined by the following mathematical expression:

[0068]

[0069] Based on the spatiotemporal connectivity component search method, neighboring pixels that are connected to the central pixel are marked as 1, and this new pixel is used as the central pixel for searching. Finally, spatiotemporally continuous areas of abnormal surface temperature changes can be obtained.

[0070] Execution step S104: Based on the spatiotemporal evolution law of the spatiotemporal cluster area, construct a surface temperature hotspot evolution model to quantify the spatiotemporal continuity characteristics of the surface temperature anomaly changes.

[0071] In this embodiment, this step provides a quantitative description of the dynamic evolution process of the identified spatiotemporal clusters.

[0072] First, to focus on significant climate events with practical implications, this embodiment filters all identified spatiotemporal clusters. The filtering criteria are: the cumulative area of ​​the cluster over its entire lifecycle must exceed a preset threshold, such as 5%, of the total area of ​​the study area; and the frequency of its occurrence, i.e., the duration, must be greater than a preset frequency threshold, such as three time units (e.g., three days or three months). This filtering step effectively filters out small-scale, short-lived random or localized temperature fluctuations, allowing the analysis to focus on persistent anomalous events that have a significant impact on regional climate.

[0073] For the selected spatiotemporal clusters, the processing unit extracts a series of time-series indicators that describe their evolutionary characteristics. These indicators include, but are not limited to: average intensity (the average temperature anomaly of all pixels within the cluster); total area (the geographical area covered by the cluster at each time point); total magnitude (the product of average intensity and total area, comprehensively reflecting the intensity and extent of the anomaly); and frequency of occurrence (the duration of the event). Any one or more of these indicators can form a long-term time-varying indicator, which can be represented as follows: .

[0074] The evolution trend of time series indicators is quantified. This embodiment assumes that these indicators change linearly with time t, i.e. Where k is the linear slope and b is the intercept. To find the slope k that best represents the data trend, the least squares fitting method is used. For a long-term series indicator with n samples, the formula for calculating its linear slope k is: .in, It is the index value corresponding to the m-th time point. From 1 to A set of integers, with index Representative set For each sample, the calculated slope k value visually quantifies the rate and direction of change of the indicator. A positive k value indicates that the indicator is increasing over time, such as the intensity of hotspots increasing or the area expanding; a negative k value indicates a weakening trend; and a k value close to 0 indicates that the indicator is in a relatively stable state.

[0075] Finally, based on the rate of change and significance level of multiple indicators, the evolution characteristics of spatiotemporally continuous surface temperature anomaly regions are comprehensively described and classified. The processing unit can comprehensively consider the slope k values ​​and statistical significance (e.g., determined by p-value) of multiple indicators such as average intensity and total area, classifying the evolution characteristics of anomaly regions into several pre-defined and easily understood categories. For example, when the slope k of both intensity and area are significantly positive and large, it can be described as "rapid warming"; when the slope k is significantly positive but small, it is described as "temperature warming"; when the slope k is significantly negative, it can be described as "stable cooling" or "rapid cooling" depending on its magnitude; when the slope k is not statistically significant, it is described as "temperature stability". In this way, the present invention ultimately outputs intuitive and quantitative classification results of the spatiotemporal evolution process of surface temperature anomaly events, providing a scientific basis for climate change research and decision support.

[0076] In summary, this invention, by constructing high-precision daily average temperature data, introducing a spatiotemporal cube to identify continuous anomalous regions, and establishing a multi-index evolution model, achieves a complete, dynamic, and quantitative description of the spatiotemporal continuity of abnormal surface temperature changes, significantly improving the continuity, accuracy, and mechanistic explanation capabilities of climate anomaly monitoring.

[0077] The above are merely preferred embodiments of the present invention and are not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for quantifying spatiotemporal continuous surface temperature anomaly change features, characterized in that, The steps of the method comprise: Obtaining regional remote sensing instantaneous surface temperature data, and constructing remote sensing daily average surface temperature based on ground station observation data; Based on the remote sensing daily average surface temperature, constructing a multi-time scale remote sensing surface temperature meteorological index; Based on the remote sensing surface temperature meteorological index, spatial clustering is performed, and the spatio-temporal aggregation area of abnormal change of surface temperature is identified in combination with the spatio-temporal correlation; Based on the spatio-temporal evolution law of the spatio-temporal aggregation area, a surface temperature hot spot evolution model is constructed to quantify the spatio-temporal continuous characteristics of the abnormal change of surface temperature; The identified spatio-temporal aggregation area of abnormal change of surface temperature is specifically: A Getis-Ord Gi* hotspot analysis model is used to perform spatial clustering on the remote sensing surface temperature meteorological index at a specific time point, and discrete surface temperature hot spots are obtained; The surface temperature hot spot marker information at different time points is superimposed to construct a spatio-temporal three-dimensional data cube; Through a spatio-temporal connection component search method, a pixel set with spatio-temporal connectivity is searched in the spatio-temporal three-dimensional data cube to determine the spatio-temporal aggregation area; The quantification of the spatio-temporal continuous characteristics of the abnormal change of surface temperature is specifically: Taking the identified spatio-temporal aggregation area as the target, the average intensity, total area, total magnitude or occurrence frequency thereof are extracted as time series indexes; Setting the time series indicator over time in a linear change, calculated using least squares fitting the linear change slope of the long time series indicator of the individual sample, calculated as wherein, represents the average intensity, total area, total magnitude or frequency of occurrence, is the linear change slope, representing the evolution trend characteristics after quantization, is an integer set from 1 to , subscript represents each sample in the set , is the index value corresponding to the th time point.

2. The spatio-temporally continuous surface temperature anomaly change feature quantification method according to claim 1, characterized in that, The construction of the remote sensing daily average surface temperature is specifically: Based on the ground surface long-wave radiation observation value, a statistical relationship between the daily average value and the instantaneous value of the ground observation surface temperature is established; The statistical relationship is mapped to the remote sensing instantaneous surface temperature data to calculate the remote sensing daily average surface temperature.

3. The spatio-temporally continuous surface temperature anomaly change feature quantification method according to claim 2, characterized in that, Before establishing the statistical relationship, it also includes calculating the surface wide band emissivity, which is specifically: Based on the sub-pixel scale land use classification data, the types and proportions of ground object components within the observation pixel are obtained; Based on the linear mixing model, the wide band emissivity of the pixel is calculated by using the proportions of each ground object component and the corresponding intrinsic emissivity.

4. The spatio-temporally continuous land surface temperature anomaly change feature quantification method according to claim 3, characterized in that, The establishment of the statistical relationship between the daily average value and the instantaneous value of the ground observation surface temperature is specifically: Based on the Stefan-Boltzmann law, the surface up long-wave radiation, the surface down long-wave radiation and the wide-band emissivity are used to calculate The instantaneous value of the ground observation surface temperature at the moment, and its calculation formula is: where, is the Stefan-Boltzmann constant, is the upwelling longwave radiation at the surface, is the downwelling longwave radiation at the surface, is the instantaneous value of the surface temperature observed at the ground, is the broadband emissivity of the mixed pixel, subscript denotes the components, is the proportion of each component, satisfying and , is the bulk emissivity of each component; The arithmetic average of all times in a single day is taken to obtain the daily average of the ground observation of the surface temperature ;​ A linear relationship model between the daily average value of the surface temperature and the instantaneous surface temperature at the white and night time is established: where subscript , are the satellite overpass time of day and night corresponding to the observation site, and are the ground observed surface temperature instantaneous value corresponding to the day and night, and are the effective observation number of day and night, and are the day and night coefficients of the ground site, is the constant term.

5. The spatio-temporally continuous surface temperature anomaly change feature quantification method according to claim 4, characterized in that, The construction of the multi-time scale remote sensing surface temperature meteorological index is specifically: Based on the remote sensing daily average surface temperature, remote sensing monthly average surface temperature and remote sensing annual average surface temperature are calculated; The daily, monthly and annual scale surface temperature anomalies are obtained by subtracting the corresponding time series arithmetic mean value from the daily, monthly and annual scale surface temperature, which are used as the remote sensing surface temperature meteorological index.

6. The spatio-temporally continuous surface temperature anomaly change feature quantification method according to claim 5, characterized in that, The specific calculation formula of the Getis-Ord Gi* hotspot analysis model is: wherein the subscripts and represent the central pixel, the neighborhood pixel, is the weight matrix, , is the distance between the central pixel and the neighborhood pixel , is the land surface temperature value of the th neighborhood pixel around the central pixel , is the local weighted sum of the land surface temperature values of the central pixel and the neighborhood pixels around the central pixel, represents the number of neighborhood pixels, is the global mean, , is the land surface temperature value of the neighborhood pixel , is the global standard deviation, , is the score, whose positive and negative values represent the high-value cluster and the low-value cluster, respectively, corresponding to the hotspot and the cold spot of the land surface temperature.

7. The spatio-temporally continuous surface temperature anomaly change feature quantification method according to claim 6, characterized in that, The constraint condition for quantifying the spatio-temporal continuous characteristics of the abnormal change of surface temperature is: Only for the spatio-temporal aggregation area with cumulative area exceeding a set value of the total area of the study area and occurrence frequency greater than a set number of times, the quantification is performed.

8. The spatio-temporally continuous surface temperature anomaly change feature quantification method according to claim 7, characterized in that, The quantification of the spatio-temporal continuous characteristics of the abnormal change of surface temperature also includes: based on the magnitude and significance level of the linear change slope classify the evolution feature of the spatiotemporal cluster into a preset evolution category, the preset evolution category including: a first rate warming, a second rate warming, a second rate cooling, a first rate cooling, and temperature stability, wherein the first rate is greater than the second rate.

Images