Method for quickly generating high-resolution carbon emission raster data

Abstract

The invention discloses a method for quickly generating high-resolution carbon emission raster data, which comprises the following steps of: fusing national carbon emission panel data and night light remote sensing images, establishing a provincial-scale logarithmic regression model, extracting parameters of the logarithmic regression model and constructing a parameter characteristic matrix; a space-time constraint K-Means algorithm is adopted to perform clustering analysis on provincial administrative units throughout the country and divide the provincial administrative units into K carbon emission type areas, for each carbon emission area, an area regression model is established by using logarithmic carbon emission and night light data, and a clustering center parameter is used as Gaussian prior to obtain a clustering result; and performing parameter updating through Bayesian maximum posteriori estimation, estimating the carbon emission by using the model, performing correction on a pixel scale by adopting a zero error method, and finally generating a national carbon emission grid data set. The space-time constraint clustering and Bayesian maximum posteriori estimation method is introduced on the basis of carbon emission modeling, and balance of few models and high precision is achieved.

Description

Technical Field

[0001] This invention relates to a method for generating carbon emission raster data, and more particularly to a method for rapidly generating high-resolution carbon emission raster data, belonging to the field of carbon emission assessment technology. Background Technology

[0002] Human activities lead to large-scale carbon emissions, seriously threatening social production and livelihood security. Currently, existing carbon emission estimation methods generally suffer from a trade-off between efficiency and accuracy: independent regression models based on provincial administrative units can improve local accuracy, but require the establishment of 31 independent models, resulting in high computational complexity and low modeling efficiency; while nationally unified models simplify the estimation process, they neglect regional differences, leading to insufficient accuracy. Nighttime light remote sensing data is widely used in carbon emission estimation due to its high correlation with human activities. However, existing technologies struggle to simultaneously balance estimation efficiency and accuracy, resulting in spatial biases in carbon emission estimation results, severely hindering the accurate implementation and evaluation of emission reduction policies. Summary of the Invention

[0003] The technical problem to be solved by the present invention is to provide a method for rapidly generating high-resolution carbon emission raster data, achieving a balance between high accuracy and high efficiency.

[0004] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows: A method for rapidly generating high-resolution carbon emission raster data includes the following steps: S1. Integrate national carbon emission panel data with nighttime light remote sensing images to establish a provincial-level logarithmic regression model; S2. Extract the parameters of the logarithmic regression model and construct the parameter feature matrix. Use the spatiotemporal constrained K-Means algorithm to perform cluster analysis on the provincial-level administrative units across the country and divide them into K carbon emission type regions. S3. For each carbon emission region, establish a regional regression model using logarithmic carbon emissions and nighttime light data, and update the parameters using cluster center parameters as Gaussian priors and Bayesian maximum a posteriori estimation. S4. The carbon emissions are estimated using the model and corrected at the pixel scale using the zero-error method, finally generating a national carbon emission raster dataset with a spatial resolution of 500m.

[0005] Further, step S1 specifically includes: S11. Collect annual carbon emission panel data covering the whole country, and perform missing value interpolation, outlier removal and provincial administrative unit aggregation on the raw data of annual carbon emission panel data. S12. Acquire DMSP / OLS and NPP / VIIRS nighttime light remote sensing images, perform mutual correction and continuity correction on the DMSP / OLS nighttime light remote sensing image data, and remove negative values, outliers, and perform continuity correction on the NPP / VIIRS nighttime light remote sensing image data. S13. Select the overlapping years of DMSP / OLS nighttime light remote sensing image data and NPP / VIIRS nighttime light remote sensing image data to establish a brightness mapping relationship. Use linear regression or quantile matching method to achieve brightness unification of DMSP / OLS nighttime light remote sensing image data and NPP / VIIRS nighttime light remote sensing image data, and form a national nighttime light dataset. S14. Taking provincial-level administrative units as the analysis objects, the total value of nighttime light pollution and total carbon emissions are transformed using natural logarithms. The transformed total value of nighttime light pollution is the independent variable, and total carbon emissions are the dependent variable. The weighted least squares method is used to fit the optimal logarithmic regression equation to establish a preliminary carbon emission estimation model. The formula is as follows:

[0006] In the formula, CE represents the total carbon emissions, and CN represents the total nighttime light emissions. is the fitting coefficient, and b is a constant.

[0007] Further, step S2 specifically includes: S21. Extract key feature parameters from the regression models of each provincial-level administrative unit, including regression coefficients, intercepts, coefficients of determination, and extreme value range indicators, and construct provincial-level feature vectors. :

[0008] In the formula, For regression coefficients, The intercept is... As the coefficient of determination, It is the range of extreme values; Standardize the features of each dimension:

[0009] In the formula, , These are the sample mean and standard deviation, respectively. Standardized feature matrix As input for clustering; S22. Introducing both spatial and temporal constraints, a spatiotemporally constrained K-Means clustering algorithm is formed, with the following formula:

[0010] In the formula, K is the number of clusters. It is the first The point set of a cluster, It belongs to Data points, It is the first The centroid of a cluster, It is the first The centroid of a cluster, For spatial constraint weighting coefficients, For time constraint weighting coefficients, It is an adjacency matrix. The rate of change in carbon emissions, and

[0011] In the formula, For the region In the year Total carbon emissions For the region In the year Total carbon emissions; S23. The spatiotemporally constrained K-Means algorithm is used to perform cluster analysis on provincial-level administrative units across the country. The elbow method combined with the silhouette coefficient is used for comprehensive evaluation to determine the optimal number of clusters. This divides all provincial-level administrative units in the country into K carbon emission type regions.

[0012] Further, step S3 specifically includes: S31. Based on the results of cluster analysis, the natural logarithm transformed data of all provinces within each carbon emission type region are merged and refitted to construct a regional-level overall carbon emission estimation model, the formula of which is as follows:

[0013] In the formula, , These represent the total carbon emissions and total nighttime light emissions after merging regions K, respectively. These are the universal fit coefficients for region K. It is a general constant term for region K; S32. According to Bayes' theorem, the posterior probability of the parameter can be expressed as:

[0014] In the formula, This is the logarithmized nighttime light and carbon emission data, i.e., the observation dataset; Regression parameters for carbon emission type regions in region K; Let be the likelihood function of the logarithmic regression model. It is a Gaussian prior distribution with the cluster center parameters as the mean; Therefore, the Bayesian maximum a posteriori estimation function is expressed as:

[0015] For ease of calculation, we take the logarithmic form: ; S33. Obtain the optimal parameter estimates by fusing observation data and prior information using the Bayesian maximum a posteriori estimation function. , This enables the optimization and adaptive adjustment of parameters for carbon emission estimation models in various regions.

[0016] Further, step S4 specifically includes: S41. Apply the optimized carbon emission estimation model to calculate the statistical carbon emission values ​​of each provincial-level administrative region based on the statistical nighttime light data. S42. Using ArcMap software, input the preprocessed nighttime light remote sensing data into the raster calculator, generate preliminary 500m resolution carbon emission raster data through model calculation, and extract the annual carbon emission estimates for each provincial administrative unit. S43. Based on the accuracy of the carbon emission estimation model, the zero-error method is used to correct the annual pixel-level carbon emission data for each province, as shown in the following formula:

[0017]

[0018] In the formula, The adjustment factor representing the province's carbon emissions. This represents the estimated provincial carbon emissions. Carbon emissions as a representative statistic After the adjustment Estimated carbon emissions per pixel After the adjustment Statistical carbon emissions per pixel; S44. The corrected carbon emission raster data of each province are mosaicked to generate a long-term series carbon emission raster dataset of China with a spatial resolution of 500 meters. The accuracy of the carbon emission estimation model is then verified by mean relative error (MRE).

[0019] Compared with existing technologies, this invention has the following advantages and effects: It provides a method for rapidly generating high-resolution carbon emission raster data. Based on provincial carbon emission modeling, it introduces spatiotemporal constrained clustering and Bayesian maximum a posteriori estimation, innovatively achieving a balance between "few models and high accuracy." By comprehensively considering spatial proximity and temporal evolution characteristics through the ST-Kmeans algorithm, this invention effectively avoids the spatial fragmentation and temporal discontinuity problems of traditional K-Means clustering results, making regional divisions more consistent with the geographical and spatial patterns of carbon emissions. Simultaneously, by using cluster center parameters to set Gaussian priors and implementing Bayesian maximum a posteriori estimation, it improves the robustness and generalization ability of the model. This invention not only significantly reduces the number of models and improves the efficiency and speed of carbon emission estimation, but also fully preserves regional differences, achieving high-precision, high-resolution, and rapid estimation of carbon emissions nationwide, providing reliable support for carbon emission policy assessment and climate change research. Attached Figure Description

[0020] Figure 1 This is a flowchart of a method for rapidly generating high-resolution carbon emission raster data according to the present invention. Detailed Implementation

[0021] To illustrate in detail the technical solutions adopted by the present invention to achieve the intended technical objectives, the technical solutions in 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, not all embodiments. Furthermore, the technical means or technical features in the embodiments of the present invention can be replaced without creative effort. The present invention will be described in detail below with reference to the accompanying drawings and embodiments.

[0022] like Figure 1 As shown, a method for rapidly generating high-resolution carbon emission raster data according to the present invention includes the following steps: S1. Integrate national carbon emission panel data with nighttime light remote sensing images to establish a provincial-level logarithmic regression model.

[0023] S11. Collect carbon emission panel data (including sub-sectors such as energy consumption and industrial processes) covering the entire country from 2000 to 2022, and perform missing value interpolation, outlier removal, and provincial administrative unit aggregation on the raw annual carbon emission panel data.

[0024] S12. Acquire DMSP / OLS (2000-2013) and NPP / VIIRS (2013-2022) nighttime light remote sensing images. Perform mutual correction and continuity correction on the DMSP / OLS nighttime light remote sensing image data. Perform negative value removal, outlier removal, and continuity correction on the NPP / VIIRS nighttime light remote sensing image data.

[0025] S13. Given the temporal segmentation characteristics of the two types of nighttime light data, a brightness mapping relationship is established by selecting overlapping years of DMSP / OLS nighttime light remote sensing image data and NPP / VIIRS nighttime light remote sensing image data. Linear regression or quantile matching methods are used to unify the brightness of DMSP / OLS nighttime light remote sensing image data and NPP / VIIRS nighttime light remote sensing image data, forming a national nighttime light dataset with consistent caliber and continuous time.

[0026] S14. Taking provincial-level administrative units as the analysis object, the total value of nighttime light pollution and total carbon emissions are transformed using natural logarithms to eliminate differences in dimensions and heteroscedasticity. The transformed total value of nighttime light pollution is used as the independent variable, and total carbon emissions as the dependent variable. A preliminary carbon emission estimation model is established by fitting the optimal logarithmic regression equation using the weighted least squares method. The formula is as follows:

[0027] In the formula, CE represents the total carbon emissions, and CN represents the total nighttime light emissions. is the fitting coefficient, and b is a constant.

[0028] S2. Extract the parameters (slope, intercept, coefficient of determination, extreme range, etc.) of the logarithmic regression model and construct the parameter feature matrix. Introduce spatial proximity and time trend constraints in the model parameter space. Use the spatiotemporal constrained K-Means algorithm to perform cluster analysis on the provincial administrative units across the country and divide them into K carbon emission type regions.

[0029] S21. Extract key feature parameters from the regression models of each provincial-level administrative unit, including regression coefficients, intercepts, coefficients of determination, extreme value ranges, etc., and construct provincial-level feature vectors. :

[0030] In the formula, The regression coefficient (slope) is the regression coefficient. The intercept is... As the coefficient of determination, It is the range of extreme values; To eliminate dimensional differences, the features in each dimension are standardized:

[0031] In the formula, , These are the sample mean and standard deviation, respectively. Standardized feature matrix As input for clustering.

[0032] S22. Traditional K-Means clustering ignores geographical proximity and temporal evolution, easily leading to spatial fragmentation and temporal discontinuity. This invention introduces dual spatial and temporal constraints, forming a spatiotemporally constrained K-Means clustering algorithm that enables clustering results to possess characteristics of feature similarity, spatial continuity, and temporal smoothness. Its formula is as follows:

[0033] In the formula, K is the number of clusters. It is the first The point set of a cluster, It belongs to Data points, It is the first The centroid of a cluster, It is the first The centroid of a cluster, For spatial constraint weighting coefficients, For time constraint weighting coefficients, The adjacency matrix is ​​used to define the spatial constraint terms. Implementation: If the region With region Adjacent ,otherwise . For the rate of change in carbon emissions, the spatial constraint term is expressed as the rate of change in carbon emissions. Achieve, and

[0034] In the formula, For the region In the year Total carbon emissions For the region In the year Total carbon emissions.

[0035] S23. The spatiotemporally constrained K-Means algorithm is used to perform cluster analysis on provincial-level administrative units across the country. The elbow method combined with the silhouette coefficient is used for comprehensive evaluation to determine the optimal number of clusters. This divides all provincial-level administrative units in the country into K carbon emission type regions.

[0036] S3. For each carbon emission region, establish a regional regression model using logarithmic carbon emission and nighttime light data, and use the cluster center parameter as a Gaussian prior. Update the parameters through Bayesian maximum a posteriori estimation to improve the fitting accuracy and robustness of the regional model.

[0037] S31. Based on the results of cluster analysis, the natural logarithm transformed data of all provinces within each carbon emission type region are merged and refitted to construct a regional-level overall carbon emission estimation model, the formula of which is as follows:

[0038] In the formula, , These represent the total carbon emissions and total nighttime light emissions after merging regions K, respectively. It is the general fit coefficient (slope) for region K. It is the general constant term (intercept) for region K.

[0039] S32. To improve the robustness and generalization ability of various carbon emission estimation models, this invention introduces the Bayesian maximum a posteriori estimation principle based on the regression model. According to Bayes' theorem, the posterior probability of the parameters can be expressed as:

[0040] In the formula, This is the logarithmized nighttime light and carbon emission data, i.e., the observation dataset; Regression parameters for carbon emission type regions in region K; This is the likelihood function of the logarithmic regression model, reflecting the degree to which the model fits the data; It is a Gaussian prior distribution with the cluster center parameter as the mean, reflecting the overall characteristics of similar regions; Therefore, the Bayesian maximum a posteriori estimation function is expressed as:

[0041] For ease of calculation, we take the logarithmic form: .

[0042] S33. Obtain the optimal parameter estimates by fusing observation data and prior information using the Bayesian maximum a posteriori estimation function. , This enables the optimization and adaptive adjustment of parameters for carbon emission estimation models in various regions.

[0043] S4. The carbon emissions are estimated using the model and corrected at the pixel scale using the zero-error method, finally generating a national carbon emission raster dataset with a spatial resolution of 500m.

[0044] S41. Apply the optimized carbon emission estimation model to calculate the statistical carbon emission values ​​of each provincial-level administrative region based on the statistical nighttime light data.

[0045] S42. Using ArcMap software, input the preprocessed nighttime light remote sensing data into the raster calculator, generate preliminary 500m resolution carbon emission raster data through model calculation, and extract the annual carbon emission estimates for each provincial administrative unit.

[0046] S43. Based on the accuracy of the carbon emission estimation model, the zero-error method is used to correct the annual pixel-level carbon emission data for each province, as shown in the following formula:

[0047]

[0048] In the formula, The adjustment factor representing the province's carbon emissions. This represents the estimated provincial carbon emissions. Carbon emissions as a representative statistic After the adjustment Estimated carbon emissions per pixel After the adjustment Statistical carbon emissions per pixel.

[0049] S44. The corrected carbon emission raster data of each province are mosaicked to generate a long-term series carbon emission raster dataset of China with a spatial resolution of 500 meters. The accuracy of the carbon emission estimation model is then verified by the mean relative error (MRE).

[0050] This invention provides a method for rapidly generating high-resolution carbon emission raster data. Based on provincial-level carbon emission modeling, it introduces spatiotemporal constrained clustering and Bayesian maximum a posteriori estimation, innovatively achieving a balance between "few models and high accuracy." By comprehensively considering spatial proximity and temporal evolution characteristics through the ST-Kmeans algorithm, this invention effectively avoids the spatial fragmentation and temporal discontinuity problems of traditional K-Means clustering results, making regional divisions more consistent with the geographical and spatial patterns of carbon emissions. Simultaneously, by using cluster center parameters to set Gaussian priors and implementing Bayesian maximum a posteriori estimation, the robustness and generalization ability of the model are improved. This invention not only significantly reduces the number of models and improves the efficiency and speed of carbon emission estimation, but also fully preserves regional differences, achieving high-precision, high-resolution, and rapid estimation of carbon emissions nationwide, providing reliable support for carbon emission policy assessment and climate change research.

[0051] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent substitutions, and improvements made to the above embodiments without departing from the scope of the present invention, based on the technical essence of the present invention and within the spirit and principles of the present invention, shall still fall within the protection scope of the present invention.

Claims

1. A method for rapidly generating high-resolution carbon emission raster data, characterized in that... Includes the following steps: S1. Integrate national carbon emission panel data with nighttime light remote sensing images to establish a provincial-level logarithmic regression model; S2. Extract the parameters of the logarithmic regression model and construct the parameter feature matrix. Use the spatiotemporal constrained K-Means algorithm to perform cluster analysis on the provincial-level administrative units across the country and divide them into K carbon emission type regions. S3. For each carbon emission region, establish a regional regression model using logarithmic carbon emissions and nighttime light data, and update the parameters using cluster center parameters as Gaussian priors and Bayesian maximum a posteriori estimation. S4. The carbon emissions are estimated using the model and corrected at the pixel scale using the zero-error method, finally generating a national carbon emission raster dataset with a spatial resolution of 500m.

2. The method for rapidly generating high-resolution carbon emission raster data according to claim 1, characterized in that: Step S1 specifically involves: S11. Collect annual carbon emission panel data covering the whole country, and perform missing value interpolation, outlier removal and provincial administrative unit aggregation on the raw data of annual carbon emission panel data. S12. Acquire DMSP / OLS and NPP / VIIRS nighttime light remote sensing images, perform mutual correction and continuity correction on the DMSP / OLS nighttime light remote sensing image data, and remove negative values, outliers, and perform continuity correction on the NPP / VIIRS nighttime light remote sensing image data. S13. Select the overlapping years of DMSP / OLS nighttime light remote sensing image data and NPP / VIIRS nighttime light remote sensing image data to establish a brightness mapping relationship. Use linear regression or quantile matching method to achieve brightness unification of DMSP / OLS nighttime light remote sensing image data and NPP / VIIRS nighttime light remote sensing image data, and form a national nighttime light dataset. S14. Taking provincial-level administrative units as the analysis objects, the total value of nighttime light pollution and total carbon emissions are transformed using natural logarithms. The transformed total value of nighttime light pollution is the independent variable, and total carbon emissions are the dependent variable. The weighted least squares method is used to fit the optimal logarithmic regression equation to establish a preliminary carbon emission estimation model. The formula is as follows: In the formula, CE represents the total carbon emissions, and CN represents the total nighttime light emissions. is the fitting coefficient, and b is a constant.

3. The method for rapidly generating high-resolution carbon emission raster data according to claim 1, characterized in that: Step S2 specifically involves: S21. Extract key feature parameters from the regression models of each provincial-level administrative unit, including regression coefficients, intercepts, coefficients of determination, and extreme value range indicators, and construct provincial-level feature vectors. : In the formula, For regression coefficients, The intercept is... As the coefficient of determination, It is the range of extreme values; Standardize the features of each dimension: In the formula, , These are the sample mean and standard deviation, respectively. Standardized feature matrix As input for clustering; S22. Introducing both spatial and temporal constraints, a spatiotemporally constrained K-Means clustering algorithm is formed, with the following formula: In the formula, K is the number of clusters. It is the first The point set of a cluster, It belongs to Data points, It is the first The centroid of a cluster, It is the first The centroid of a cluster, These are the spatial constraint weighting coefficients. For time constraint weighting coefficients, It is an adjacency matrix. The rate of change in carbon emissions, and In the formula, For the region In the year Total carbon emissions For the region In the year Total carbon emissions; S23. The spatiotemporally constrained K-Means algorithm is used to perform cluster analysis on provincial-level administrative units across the country. The elbow method combined with the silhouette coefficient is used for comprehensive evaluation to determine the optimal number of clusters. This divides all provincial-level administrative units in the country into K carbon emission type regions.

4. The method for rapidly generating high-resolution carbon emission raster data according to claim 1, characterized in that: Step S3 specifically involves: S31. Based on the results of cluster analysis, the natural logarithm transformed data of all provinces within each carbon emission type region are merged and refitted to construct a regional-level overall carbon emission estimation model, the formula of which is as follows: In the formula, , These represent the total carbon emissions and total nighttime light emissions after merging regions K, respectively. These are the universal fit coefficients for region K. It is the general constant term for region K; S32. According to Bayes' theorem, the posterior probability of the parameter can be expressed as: In the formula, This is the logarithmized nighttime light and carbon emission data, i.e., the observation dataset; Regression parameters for carbon emission type regions in region K; For the likelihood function of the logarithmic regression model, The prior distribution is a Gaussian distribution with the cluster center parameters as the mean. Therefore, the Bayesian maximum a posteriori estimation function is expressed as: For ease of calculation, we take the logarithmic form: ； S33. Obtain the optimal parameter estimates by fusing observation data and prior information using the Bayesian maximum a posteriori estimation function. , This enables the optimization and adaptive adjustment of parameters for carbon emission estimation models in various regions.

5. The method for rapidly generating high-resolution carbon emission raster data according to claim 1, characterized in that: Step S4 specifically involves: S41. Apply the optimized carbon emission estimation model to calculate the statistical carbon emission values ​​of each provincial-level administrative region based on the statistical nighttime light data. S42. Using ArcMap software, input the preprocessed nighttime light remote sensing data into the raster calculator, generate preliminary 500m resolution carbon emission raster data through model calculation, and extract the annual carbon emission estimates for each provincial administrative unit. S43. Based on the accuracy of the carbon emission estimation model, the zero-error method is used to correct the annual pixel-level carbon emission data for each province, as shown in the following formula: In the formula, The adjustment factor representing the province's carbon emissions. This represents the estimated provincial carbon emissions. Carbon emissions as a representative statistic After the adjustment Estimated carbon emissions per pixel After the adjustment Statistical carbon emissions per pixel; S44. The corrected carbon emission raster data of each province are mosaicked to generate a long-term series carbon emission raster dataset of China with a spatial resolution of 500 meters. The accuracy of the carbon emission estimation model is then verified by the mean relative error (MRE).

Images