A method, system, device and medium for inverting black carbon emissions

Abstract

The application discloses a kind of black carbon emission inversion method, system, equipment and medium, it is related to carbon emission inventory inversion technical field, comprising: using empirical orthogonal function model to black carbon emission inventory is carried out spatial and temporal decomposition dimension reduction, obtain the multiple emission source regions of global scale indicating black carbon emission source;In each emission source region, consider the linear and nonlinear interaction between multiple influence factors, establish the nonlinear regression model between black carbon emission under the interaction of multiple influence factors;Satellite remote sensing observation true value and reanalysis meteorological field data of multiple influence factors are input nonlinear regression model to obtain the black carbon emission inventory in emission source region.The application can obtain accurate black carbon emission, and then obtain accurate black carbon emission inventory data.

Description

Technical Field

[0001] This invention relates to the field of carbon emission technology, and in particular to a method, system, device and medium for retrieving black carbon emissions. Background Technology

[0002] Black carbon is a type of fine particulate matter (PM2.5) in the atmosphere. 2.5 Black carbon is a significant component of the Earth's atmosphere, originating from the incomplete combustion of fossil fuels and biomass. Its surface can adsorb large amounts of volatile organic compounds (VOCs) and other carcinogens, which can enter the human respiratory system and harm human health. Black carbon aerosols can act as cloud condensation nuclei, affecting cloud droplet size distribution, liquid water content in clouds, and cloud cover, thus influencing rainfall and leading to more frequent droughts or torrential rains in different regions. Furthermore, black carbon aerosols have a strong absorption effect on visible light, generating positive radiative forcing and altering the Earth's surface radiative energy balance, thereby affecting climate change. Therefore, statistically analyzing black carbon pollution and developing targeted measures to control it based on local emission characteristics are crucial for estimating the climate effects of black carbon, assessing public health exposure, and developing emission inventories. A black carbon emission inventory is a tool for recording and reporting black carbon emissions from a specific region or industry over a given period.

[0003] In existing technologies, black carbon emission inventories are mainly used to count the amount of black carbon pollution emitted into the atmosphere from various black carbon pollution sources. Currently, black carbon emission inventories are mostly calculated based on the classified emission sources and emission factors.

[0004] However, there are too many types of emission sources, and it is impossible to comprehensively count emission sources in all regions. Furthermore, the black carbon emitted from emission sources is affected by a variety of factors. The methods mentioned above do not take into account the various factors affecting emission sources when obtaining black carbon emission inventories, which leads to errors in the black carbon emission amounts in the black carbon emission inventories. Summary of the Invention

[0005] This invention provides a method, system, device, and medium for inverting black carbon emissions, which can solve the problem that the black carbon emissions in the black carbon emission inventory obtained in the prior art are inaccurate.

[0006] This invention provides a method for inverting black carbon emissions, comprising the following steps: obtaining a black carbon emission inventory including black carbon emissions, and obtaining multiple influencing factors affecting black carbon emissions, including nitrogen dioxide, ultraviolet aerosol index, and wind; performing spatiotemporal decomposition and dimensionality reduction on the black carbon emission inventory using an empirical orthogonal function model to obtain multiple emission source regions including black carbon emission sources; within each emission source region, considering the linear and nonlinear interactions between multiple influencing factors, establishing a nonlinear regression model between the interaction of multiple influencing factors and black carbon emissions; inputting the multiple influencing factors into the nonlinear regression model to obtain black carbon emissions, thus completing the inversion of black carbon emissions.

[0007] Further, the specific steps for obtaining multiple emission source regions including black carbon emission sources include: using an empirical orthogonal function model to obtain multiple feature contribution rates from the black carbon emission inventory; selecting the top three largest feature contribution rates from all feature contribution rates; labeling the selected top three largest feature contribution rates from largest to smallest as EOF1, EOF2, and EOF3 respectively; filtering values ​​greater than 0.5 and less than -0.5 in the feature vector corresponding to feature contribution rate EOF1; filtering values ​​greater than 0.8 and less than -0.8 in the feature vectors corresponding to feature contribution rates EOF2 and EOF3; obtaining the geographical locations of preliminary emission source regions from the black carbon emission inventory based on the numerical differences of the filtered results; excluding regions with an area less than 100 pixels in the preliminary emission source region, and using an edge detection algorithm to extract the regional boundaries of the excluded preliminary emission source regions, thus dividing them into six emission source regions.

[0008] Furthermore, the establishment of a nonlinear regression model between the interaction of multiple influencing factors and black carbon emissions includes the following steps: multiple influencing factors reflecting black carbon emissions include nitrogen dioxide. NO 2 UV aerosol index UVAI With the wind Wind The linear interaction between influencing factors and black carbon emissions needs to be considered, as well as the nonlinear interaction between influencing factors, including the ultraviolet aerosol index. UVAI With the wind Wind The interaction between nitrogen dioxide and nitrogen dioxide NO 2 With the wind Wind The interaction;

[0009] The formula for the established nonlinear regression model is as follows:

[0010]

[0011] in, , , , and Nitrogen dioxide NO 2 UV aerosol index UVAI ,wind Wind ,wind Wind With UV aerosol index UVAI Interaction and wind Wind With nitrogen dioxide NO 2 The coefficient of interaction, It is a constant.

[0012] Furthermore, the specific steps for obtaining black carbon emissions by inputting multiple influencing factors into a nonlinear regression model include: obtaining nitrogen dioxide... NO 2 UV aerosol index UVAI Japanese style Wind The monitoring data from multiple months were input into a linear regression model to obtain the black carbon emissions for each month. Based on the obtained black carbon emissions, the annual average, quarterly average, and monthly average black carbon emissions were analyzed, and the differences were compared with the black carbon emission inventory to determine the accuracy of the inversion results.

[0013] Furthermore, the specific steps for obtaining the black carbon emission inventory include: obtaining an optimized estimate of black carbon emissions using the Kalman filter method based on historical black carbon emissions at a global scale; obtaining a scaling factor to make the black carbon emissions conform to the actual situation based on the difference between the optimized estimate and historical black carbon emissions; and scaling the monthly black carbon emission data from the Global Atmospheric Research Emissions Database (EDGAR) to obtain the black carbon emission inventory.

[0014] This invention provides a system for retrieving black carbon emissions, comprising:

[0015] The data acquisition module is used to acquire a black carbon emission inventory including black carbon emissions and to identify multiple influencing factors affecting black carbon emissions, including nitrogen dioxide, ultraviolet aerosol index, and wind. The model building module is used to perform spatiotemporal decomposition and dimensionality reduction on the black carbon emission inventory using an empirical orthogonal function model to acquire multiple emission source regions including black carbon emission sources. Within each emission source region, considering the linear and nonlinear interactions between multiple influencing factors, a nonlinear regression model is established between the interaction of multiple influencing factors and black carbon emissions. The black carbon emission inversion module is used to input multiple influencing factors into the nonlinear regression model to obtain black carbon emissions, thus completing the inversion of black carbon emissions.

[0016] This invention provides a computer device, including a memory and a processor; the memory stores a computer program, and the processor executes the computer program to implement the aforementioned method for inverting black carbon emissions.

[0017] This invention provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the aforementioned method for inverting black carbon emissions.

[0018] This invention provides a method, system, device, and medium for retrieving black carbon emissions. Compared with existing technologies, its advantages are as follows:

[0019] Factors influencing black carbon emissions include nitrogen dioxide, ultraviolet aerosol index, and wind. Furthermore, considering the interactions among these factors, a nonlinear regression model was established to examine the relationship between black carbon emissions and the interaction of these factors. This nonlinear regression model not only considers the individual effects of nitrogen dioxide, ultraviolet aerosol index, and wind on black carbon emissions, but also the impact of interactions among these factors. Ultimately, by inputting multiple influencing factors into the nonlinear regression model, an accurate black carbon emission level can be obtained through inversion. Attached Figure Description

[0020] Picture 1 A technical flowchart provided for embodiments of the present invention;

[0021] Figure 2 is a PDF analysis diagram of each factor in the linear regression model for region A provided in this embodiment of the invention, where (a) represents the coefficients. In the analysis results for each emission source region, (b) represents the coefficient. In the analysis results for each emission source region, (c) represents the coefficient. Analysis results for each emission source region;

[0022] Figure 3 is a PDF analysis diagram of each factor in the nonlinear regression model for region A provided in this embodiment of the invention, where (a) represents the coefficients. In the analysis results for each emission source region, (b) represents the coefficient. In the analysis results for each emission source region, (c) represents the coefficient. In the analysis results for each emission source region, (d) represents the coefficient. In the analysis results for each emission source region, (e) represents the coefficient. Analysis results for each emission source region;

[0023] Picture 4 The inverted global monthly average time series of black carbon is provided for embodiments of the present invention. Detailed Implementation

[0024] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Many specific details are set forth in the following description to provide a thorough understanding of the present invention. However, the present invention can be practiced in many other ways different from those described herein, and those skilled in the art can make similar modifications without departing from the spirit of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0025] See Picture 1 This invention provides a method for retrieving black carbon emissions, comprising the following steps:

[0026] Step 1: Obtain a black carbon emission inventory that includes black carbon emissions, and identify multiple factors that affect black carbon emissions, including nitrogen dioxide, ultraviolet aerosol index, and wind.

[0027] The data used in this invention specifically include:

[0028] 1. Ozone monitor data

[0029] This invention utilizes the Level-3 daily global gridded (0.25 × 0.25 degrees) nitrogen dioxide product OMNO2d (OMI / TOMS NitrogenDioxide (NO2) Total and Tropospheric Column L3 Daily Best Pixel Data) from the Goddard Center for Geoscience Data and Information (GDS) website, spanning from 2005 to 2018. The OMNO2d data product is a three-level gridded product where high-quality pixel-level data is averaged across a 0.25 × 0.25 degree grid. This product includes a total NO2 column and a total tropospheric NO2 column, applicable to all atmospheric conditions and sky conditions with cloud cover less than 30%. Based on this, this paper retains the daily tropospheric NO2 data from the product and converts it into monthly data, storing it yearly for future reference.

[0030] The ultraviolet aerosol index can be used to indicate ultraviolet absorbing aerosols. Black carbon (BC), as the main light-absorbing substance in aerosols, should have a positive correlation between black carbon emissions and changes in absorption intensity monitored by satellite. The higher the concentration of absorbing aerosols, the larger the ultraviolet aerosol index (UVAI) value. This paper uses Level-2 global grid data from 2005-2018 (0.25) from the NASA Goddard Earth Science Data and Information Service website. Near-ultraviolet aerosol optical thickness and single-scatter albedo V3 product OMAERUVG data (0.25°). This Level-2 G daily global grid product OMAERUVG is based on the pixel-level OMI Level-2 Aerosol product OMAERUV. The OMAERUVG data product is a special Level-2 gridded product where pixel-level products are divided into a 0.25×0.25° grid. It contains scene data for all observation times between 00:00:00 and 23:59:59.999 UTC. The OMAERUVG files are stored in the EOS hierarchical data format version 5 (HDF-OSE5). Each file contains daily data for approximately 15 orbitals mapped on a global 0.25×0.25° grid. Based on this, UVAI data preprocessing was performed, converting the daily irregular latitude and longitude grid into a regular latitude and longitude grid similar to that of the nitrogen dioxide data, and calculating the monthly average data for later use.

[0031] 2. Black carbon data from the Global Atmospheric Research Emissions Database (EDGAR BC data)

[0032] This invention uses monthly black carbon emission data from EDGAR 2002-2018 HTAP_v3 (https: / / edgar.jrc.ec.europa.eu / dataset_htap_v3), with a resolution of 0.1×0.1. The data includes emissions from international shipping, industrial emissions, and land transportation, measured in tons per month. HTAP_v3 was developed within the framework of the ECE Air Convention to improve scientific knowledge regarding the intercontinental transport of air pollution over the Northern Hemisphere, based on global monthly air pollutant data (SO2, NO). x , CO, NMVOCs, NH3, PM10, PM 2.5 The emissions grid consists of data from the European EMEP (CAMS-REG-v5.1) and the US Environmental Protection Agency (US EPA), covering time series from 2000 to 2018 and emissions from all anthropogenic sectors excluding land use, land use change, and forestry. The emissions grid is compiled from data collected from official reports such as the European EMEP (CAMS-REG-v5.1) and the US EPA. The EDGARv6.1 air pollutant emissions grid already covers all other countries in the world. Furthermore, compared to previous versions like HTAPv1 and HTAPv2.2, HTAP_v3 aims to expand the temporal coverage, industry classification, and geographical coverage of official data for air pollutant emissions, enabling analysis of trends, research on transboundary transport of air pollutants, and assistance for policymakers in addressing and reducing emissions from relevant sectors. Based on this, the sum of emissions from each sector is calculated, and the resolution is converted to 0.25 × 0.25 using mathematical transformation principles. .

[0033] 3. Reanalysis of data (NCEP-NCAR Reanalysis 1)

[0034] The data used is the monthly average wind value data from the NOAA Physical Sciences Laboratory (PSL). The NOAA Physical Sciences Laboratory (PSL) is a laboratory under the National Oceanic and Atmospheric Administration (NOAA) dedicated to researching and providing data and information related to climate, atmospheric science, and geophysics. The u-wind mean monthly 10m data refers to east-west wind speed data measured at a height of 10 meters. This data records the average wind speed and wind direction for each month. By analyzing this data, we can understand the changes in wind speed in different regions and time periods, thereby enabling climate and weather forecasting and research.

[0035] Step 2: Use an empirical orthogonal function model to perform spatiotemporal decomposition and dimensionality reduction on the black carbon emission inventory to obtain multiple emission source regions including black carbon emission sources; within each emission source region, consider the linear and nonlinear interactions between multiple influencing factors and establish a nonlinear regression model between the interaction of multiple influencing factors and black carbon emissions.

[0036] Step 3: Input multiple influencing factors into a nonlinear regression model to obtain black carbon emissions, thus completing the inversion of black carbon emissions. Furthermore, the inverted black carbon emissions have the characteristics of long-term series and gridded structure.

[0037] The specific analysis process for the above steps is as follows:

[0038] 1. Construct a priori black carbon emission inventory

[0039] Based on the location information of monthly black carbon emissions data from EDGAR 2002-2018 HTAP_v3, and combined with the optimized estimate of global black carbon emissions of 17.8±5.6 Tg / year obtained by Cohen (2014) using the Kalman filter method for the first time, the scaling factor was calculated to vary from 2.21 to 12.06. A new prior black carbon emission inventory was constructed and used as the benchmark for subsequent calculations.

[0040] 2. Identification and Analysis of Black Carbon Emission Areas

[0041] Empirical Orthogonal Function (EOF) analysis is also known as Principal Component Analysis (PCA) in mathematical statistics.

[0042] EOF analysis (using empirical orthogonal functions) was performed on BC data from 2005 to 2010 to identify and delineate emission source regions. According to the EOF analysis results, the contribution rates of the first five eigenvalues ​​were 0.5296, 0.1653, 0.1166, 0.0456, and 0.0232, respectively. The latter two had relatively small contribution rates, so the first three were retained for analysis. These are denoted as: EOF1 = 0.5296, EOF2 = 0.1653, and EOF3 = 0.1166. Outlier values ​​with absolute EOF values ​​less than 0.1 were uniformly removed from the images.

[0043] To make the regional differences between high and low black carbon values ​​more intuitive and discernible, the values ​​of EOF1, EOF2, and EOF3 were further filtered. Values ​​of EOF1 greater than 0.5 and less than -0.5 were filtered, and values ​​of EOF2 and EOF3 greater than 0.8 and less than -0.8 were filtered.

[0044] Six emission source regions were designated: Source Region 1 is Region A, mainly consisting of parts of Country A; Source Region 2 is Region B, mainly consisting of Country B; Source Region 3 is Region C; Source Region 4 is Region D, mainly consisting of Countries D, E, and F; Source Region 5 is Region E, mainly consisting of Country G; and Source Region 6 is Region F, mainly consisting of Country H.

[0045] Assume there is There are 10 related variables, and each variable has 100 related variables. The samples constitute a matrix. Through the Performing a linear transformation can... A linear combination of variables forms a new variable:

[0046]

[0047] in, The principal components of the original variable, This is a linear transformation matrix. This process concentrates most of the information from the original multiple variables into the principal components of a few independent variables.

[0048] Principal component analysis (PCA) in a climate variable field involves... spatial points Variables constituted by the observations Seen as A linear combination of spatial feature vectors and their corresponding time weight coefficients:

[0049]

[0050] Here, V is the spatial function matrix, and T is the time function matrix. In this process, the main information of the variable field is concentrated in several typical eigenvectors. This means that we can approximate the original data using fewer spatial and time function modes, achieving a dimensionality reduction effect.

[0051] 3. Correlation Analysis between Black Carbon Emissions and Different Factors

[0052] Combustion typically produces both black carbon and nitrogen dioxide. Black carbon mainly originates from particulate matter generated by the incomplete combustion of organic matter, while nitrogen dioxide is a gaseous pollutant produced during high-temperature combustion reactions. Black carbon particles can act as carriers of nitrogen dioxide. Nitrogen dioxide adsorbed on the surface of black carbon particles can be converted into nitrates in the atmosphere and then fall to the ground, negatively impacting the environment and human health. Furthermore, black carbon (BC) strongly absorbs solar radiation across multiple wavelengths. The Ultraviolet Aerosol Index (UVAI) indicates the absorption of light-absorbing aerosols in the ultraviolet band and is a remote sensing indicator for identifying light-absorbing aerosols in the atmosphere. A higher UVAI value indicates a higher concentration of light-absorbing aerosols in the atmosphere. Therefore, by monitoring the UVAI value, the content of absorbing aerosols in the atmosphere, including black carbon particles, can be indirectly determined. Simultaneously, meteorological factors such as wind also significantly influence the transport and diffusion of black carbon in the atmosphere. Therefore, constructing multiple linear regression relationships between black carbon and nitrogen dioxide, ultraviolet aerosol index, and wind can help us better understand air pollution sources and pollutant emissions. By analyzing these relationships, we can identify major pollution sources, emission levels, and pollution source control strategies; we can also assess the contribution and impact of different factors on pollutant concentrations, providing a scientific basis for formulating pollutant control strategies. Furthermore, by adding nonlinear terms to the interactions between nitrogen dioxide and wind, and between ultraviolet aerosol index and wind, we can further verify the accuracy of the estimates and the impact of each factor on black carbon emissions.

[0053] During the analysis, linear regression models and nonlinear regression models are established and the results are compared to evaluate which model obtains more accurate results.

[0054] First, a linear regression model and a nonlinear regression model are established to explore the correlation between nitrogen dioxide, ultraviolet aerosol index, wind and black carbon in different regions, and the degree of their impact on black carbon. The linear regression model 1 is as follows:

[0055] In the linear regression model, region A and All with Positive correlation, and Negative correlation; in region B and It shows a positive correlation, and in region B right The impact is significant. and right The impact was relatively weak, but all were positively correlated; in region C In the central and eastern regions and Negative correlation In the whole and Positive correlation, The impact is mainly concentrated in the eastern region of country C, and... A positive correlation was observed in region D; Except for a negative correlation in some regions of country F, the overall correlation is positive. , There are both positively and negatively correlated components, and it can be seen that... For the area The impact of emissions is significant; in the G country region and All with Positive correlation, For the entire G country region The correlation between emissions and emissions is significant. The correlation is predominantly negative; in region J of country H, There is a clear positive correlation. , and Both positive and negative correlations exist, among which Primarily positive correlation The correlation is predominantly negative.

[0056] The physical meaning of the three influencing factors used in the linear regression model is explained as follows: nitrogen dioxide and black carbon have the same emission source; the ultraviolet aerosol index represents the absorptivity of aerosols in the ultraviolet band, and black carbon has a strong absorptive effect in the ultraviolet band; meteorological field data is an influencing factor due to the influence of surrounding black carbon transport.

[0057] The following is a nonlinear regression model 2:

[0058]

[0059] Based on linear regression 1, the following was added and Two nonlinear terms are introduced to explain the interaction between the independent variables. By introducing these nonlinear terms, we can explore... , , The mutual influence and nonlinear relationship between them A more comprehensive understanding of the impact of emissions. , , right The mechanism of influence.

[0060] In the nonlinear regression model, in addition to the three influencing factors mentioned above, there are two other nonlinear influencing factors, whose relevant physical meanings can be interpreted as follows: express The impact of emission transmission on black carbon emissions mainly refers to the aging effect of black carbon in the air; This indicates the impact of surrounding black carbon emission transmission on local black carbon emissions, mainly representing the dynamic transmission mechanism of black carbon emissions.

[0061] Based on the nonlinear regression model, it can be analyzed that in region A... In the northern region and It is mainly positively correlated, and in the southern region with The correlation is mainly negative. and Positive correlation, In the northeastern region, excluding H City and B City, and Overall, there is a positive correlation. and Overall, there is a positive correlation. Besides parts of HL City and JL City, Overall, there is a positive correlation. In region B, and It shows a positive correlation, and in this region right The impact is significant and the strongest. and Positive correlation, and Positive and negative correlations alternate, and and In this area The impact is relatively weak. In region C, In its western region and A positive correlation was observed, while a negative correlation was predominant in other regions. In its northern and A negative correlation was observed in other regions, while a positive correlation was observed in others. and Positive and negative correlations intertwine; In its northern and A negative correlation was observed in other regions, while a positive correlation was observed in others. In the central and western regions and The correlation was negative, while other correlations were predominantly positive; in region D, At both the north and south ends The middle part shows a negative correlation, while the middle part shows a positive correlation. Predominantly negatively correlated; The correlation is negative in the north and positive in the south. , The correlation is predominantly positive. In country G, except for the southwestern corner of country G... Overall and Positive correlation, The correlation is mainly negative in the east and mainly positive in the west. Then, except for a local area in the southeast corner, the whole and Positive correlation, The correlation is negative in the middle and positive around the edges. and A positive correlation is observed in the east, while a predominantly negative correlation is observed in the west. In region F of country H, , and They are all predominantly positively correlated. Predominantly negative correlation, Positive and negative correlations are mixed.

[0062] 4. Impact of source region data on black carbon emissions

[0063] This study investigates the impact of various data points in the source region on black carbon emissions by examining the magnitude and range of coefficients corresponding to each factor, as well as the effect of coefficient changes on black carbon emission results. Therefore, probability density function (PDF) analysis is used to understand the distribution characteristics of each coefficient in the two linear regression models, such as coefficient range, peak value, variance, and median. Sensitivity analysis is then used to assess the correlation between factors and black carbon, determining which factors have a significant impact on black carbon emission results and which have a relatively small impact.

[0064] The relative coefficients obtained from equation 1 of the linear regression model , , The results of the probability density function PDF analysis in each source region are shown in Figure 2.

[0065] In region A, Corresponding coefficients The data exhibits a positively skewed distribution, with the 10th and 90th quantiles at -0.0816 and 0.3687, respectively. The coefficients... The average value is 0.1164, the median value is 0.0874, and the value in the range of -0.2 to +0.2 accounts for 56.11% of the overall coefficient distribution. The peak value is located in the range of 0.04 to 0.08, accounting for 8.82%.

[0066] Corresponding coefficients The coefficients are relatively concentrated, with the 10th and 90th percentiles being -0.0053 and 0.2197, respectively. The average coefficient is 0.1046 and the median is 0.0958. The coefficients in the range of -0.2 to +0.2 account for 68.56% of the overall coefficient distribution, and the peak value is located in the range of 0.04 to 0.08, accounting for 15.86%.

[0067] Corresponding coefficients Comparison , The coefficients are relatively discrete, with the 10th and 90th quantiles being -0.1252 and 0.5175, respectively. The average coefficient is 0.1255, and the median is 0.0473. The coefficients in the range of -0.2 to +0.2 account for 51.77% of the overall coefficient distribution, while the peak value is located in the range of -0.04 to 0, accounting for 10.75%.

[0068] As can be seen from Figure 2, , Corresponding coefficient , Its proportion in the 0.4-0.8 range is much higher than Corresponding coefficient , , The percentages were 6.23% and 10.79% respectively, while The proportion is only 0.08%. , The proportion of coefficients in the high-value region is higher than coefficient, therefore in region A , right The impact of emissions is higher than .

[0069] In region B, Corresponding coefficients The data follows a normal distribution, with the 10th and 90th quantiles being -0.0685 and 0.0834, respectively. The coefficients... The average value is 0.1175, the median value is 0.1063, and the value in the range of -0.2 to +0.2 accounts for 71.07% of the overall coefficient distribution. The peak value is located in the range of 0.08-0.12, accounting for 13.15%.

[0070] Corresponding coefficients The coefficients are relatively concentrated, with the 10th and 90th percentiles being -0.0342 and 0.2847, respectively. The average coefficient is 0.0105 and the median is 0.0153. The coefficients in the range of -0.2 to +0.2 account for 92.87% of the overall coefficient distribution, and the peak value is located in the range of 0-0.04, accounting for 27.04%.

[0071] Corresponding coefficients The distribution is also very concentrated, with the 10th and 90th percentiles being -0.0780 and 0.0621, respectively. The mean and median are very small, at 0.0021 and 0.0072, respectively. The range of -0.2 to +0.2 accounts for 93.28% of the overall coefficient distribution, and its peak value is located in the range of 0-0.04, accounting for as high as 30.48%.

[0072] in, Corresponding coefficients The data is widely distributed, and the proportion of high values ​​is also higher than that of other data points. , Corresponding coefficient, therefore in region B right The impact of emissions is the highest.

[0073] In region C, Corresponding coefficients The data exhibits a negatively skewed distribution, with the 10th and 90th quantiles at -0.3051 and 0.2437, respectively. The coefficients... The average value is -0.0205, the median value is -0.0155, and the value in the range of -0.2 to +0.2 accounts for 45.24% of the overall coefficient distribution. The peak value is located in the range of 0 to 0.04, accounting for 5.89%.

[0074] Corresponding coefficients The data exhibits a positively skewed distribution, with the 10th and 90th quantiles at -0.0529 and 0.3158, respectively. The coefficients... The average value is 0.1072, the median value is 0.0828, and the value in the range of -0.2 to +0.2 accounts for 49.28% of the overall coefficient distribution. The peak value is located in the range of 0 to 0.04, accounting for 9.26%.

[0075] Corresponding coefficients The data exhibits a bimodal distribution. The first peak occurs in the 0-0.04 range, accounting for 11.93%, and the second peak occurs in the 0.2-0.24 range, accounting for 5.51%. The 10th and 90th quantiles are -0.0395 and 0.2660, respectively. The average value is 0.0948, the median value is 0.0709, and 51.78% of the coefficients fall within the range of -0.2 to +0.2.

[0076] in, The coefficient has the widest distribution range and a high proportion of values ​​less than 0, indicating a negative relationship with BC emissions. , Both are correct. The emissions have a positive impact.

[0077] The coefficients obtained from the nonlinear regression equation 2 , , , and The PDF results in each source region are shown in Figure 3.

[0078] In region A, Corresponding coefficients The data exhibits a positively skewed distribution, with the 10th and 90th quantiles at -0.3420 and 0.5767, respectively. The data range is relatively large, and the coefficients... The average value is 0.0933, the median value is 0.0693, and the value in the range of -0.2 to +0.2 accounts for 41.81% of the overall coefficient distribution. The peak value is located in the range of 0-0.04, accounting for 5.91%.

[0079] Corresponding coefficients The data exhibits a positively skewed distribution, with the 10th and 90th quantiles being 0.0242 and 0.3721, respectively. The average coefficient is 0.1760, and the median is 0.1477. The coefficients within the range of -0.2 to +0.2 account for 51.57% of the overall distribution, while the peak value is located in the range of 0.08 to 0.12, accounting for 11.49%.

[0080] Corresponding coefficients The data exhibits a positively skewed distribution, with the 10th and 90th quantiles being -0.3398 and 0.9568, respectively. The average coefficient is 0.1731, and the median is 0.0228. The coefficients within the range of -0.2 to +0.2 account for 41.37% of the overall coefficient distribution, while the peak value is located in the range of 0 to 0.04, accounting for 9.72%.

[0081] Corresponding coefficients The data exhibits a negatively skewed distribution, with the 10th and 90th quantiles being -0.2592 and 0.0508, respectively. The average coefficient is -0.0716, and the median is -0.0314. 65.70% of the overall coefficient distribution falls within the range of -0.2 to +0.2, while the peak value is located in the interval between -0.04 and 0, accounting for 17.94%.

[0082] Corresponding coefficients The data exhibits a positively skewed distribution, with the 10th and 90th quantiles being -0.4142 and 0.4574, respectively. The average coefficient is 0.0253, and the median is 0.0083. The coefficients within the range of -0.2 to +0.2 account for 46.77% of the overall coefficient distribution, while the peak value is located in the range of -0.04 to 0, accounting for 10.32%.

[0083] , , Corresponding coefficient , , The data range is quite broad, and Corresponding coefficients Mostly greater than 0, Corresponding coefficients Most of them are less than 0, among which , In region A The impact of emissions is the greatest.

[0084] In region B, Corresponding coefficients The data approximates a normal distribution, with the 10th and 90th quantiles at -0.0749 and 0.3384 respectively. The data distribution range is relatively large, and the coefficients... The average value is 0.1229, the median value is 0.1145, and the value in the range of -0.2 to +0.2 accounts for 63.34% of the overall coefficient distribution. The peak value is located in the range of 0.12 to 0.16, accounting for 10.04%.

[0085] Corresponding coefficients The data exhibits a negatively skewed distribution, with the 10th and 90th quantiles at -0.0833 and 0.0837, respectively. The average coefficient is 0.0043, and the median is 0.0115. 91.70% of the coefficients fall within the range of -0.2 to +0.2, indicating a relatively concentrated distribution. The peak value is located in the 0-0.04 range, accounting for 26.05%.

[0086] Corresponding coefficients The data exhibits a positively skewed distribution, with the 10th and 90th quantiles being -0.1642 and 0.1548, respectively. The average coefficient is -0.0046, and the median is -0.0025. 80.93% of the overall coefficient distribution falls within the range of -0.2 to +0.2, while the peak value is located in the interval between -0.04 and 0, accounting for 18.08%.

[0087] Corresponding coefficients The data is highly concentrated, with the 10th and 90th percentiles being -0.0174 and 0.0408, respectively. The average coefficient is 0.0079, and the median is 0.0026. 93.52% of the overall coefficient distribution is in the range of -0.2 to +0.2, and the peak value is in the range of 0-0.04, accounting for 45.63%.

[0088] Corresponding coefficients The data exhibits a negatively skewed distribution, with the 10th and 90th quantiles being -0.1232 and 0.1106, respectively. The average coefficient is -0.0020, and the median is 0.0021. 86.52% of the overall coefficient distribution falls within the range of -0.2 to +0.2, while the peak value is located in the range of 0 to 0.04, accounting for 24.09%.

[0089] , Corresponding coefficients , The data are highly concentrated in the range of -0.2 to +0.2. Corresponding coefficients The distribution is also relatively concentrated. Combined with the magnitude of the coefficients, it can be seen that Wind... UVAI and In region B The impact of emissions is minimal. right The impact of emissions is the greatest.

[0090] In region C, Corresponding coefficients The data distribution ranges widely, with values ​​in each interval between -0.8 and +0.8. The 10th and 90th quantiles are -0.7858 and 0.5706, respectively. (The coefficient is missing from the original text.) The average value is -0.0903, the median value is -0.0731, and the value in the range of -0.2 to +0.2 accounts for 22.47% of the overall coefficient distribution. The peak value is located in the range of 0 to 0.04, accounting for 2.54%.

[0091] Corresponding coefficients The data exhibits an irregular distribution, with the 10th and 90th quantiles being -0.2630 and 0.4143, respectively. The average coefficient is 0.0791, and the median is 0.0727. The coefficients within the range of -0.2 to +0.2 account for 35.73% of the overall coefficient distribution, while the peak value is located in the range of -0.04 to 0, accounting for 4.93%.

[0092] Corresponding coefficients The data exhibits a flat-top distribution, with the 10th and 90th quantiles at -0.5582 and 0.5414, respectively. The average coefficient is 0.0034, and the median is 0.0145. The coefficients within the range of -0.2 to +0.2 account for 27.41% of the overall coefficient distribution, while the peak value is located in the range of 0 to 0.04, accounting for 3.35%.

[0093] Corresponding coefficients The data are approximately normally distributed, with the 10th and 90th quantiles being -0.1678 and 0.2521, respectively. The average coefficient is 0.0276, and the median is 0.0101. The coefficients in the range of -0.2 to +0.2 account for 53.56% of the total coefficient distribution, while the peak value is located in the range of 0 to 0.04, accounting for 9.32%.

[0094] Corresponding coefficients Data distribution trend and coefficients Similarly, the 10th and 90th percentiles are -0.4589 and 0.6392, respectively, with an average coefficient of 0.0681 and a median of 0.0382. The coefficients in the range of -0.2 to +0.2 account for 28.44% of the overall coefficient distribution, with the peak value located in the range of 0 to 0.04, accounting for 3.73%. The proportion in the range of -0.04 to 0 is almost the same as that in the range of 0 to 0.04, at 3.72%.

[0095] In region C, besides Corresponding coefficients Besides exhibiting an approximately normal distribution, other coefficients, while showing significantly high values, are scattered, with each interval having a large proportion. This aligns with the situation depicted in the coefficient graphs of linear regression model 2. The coefficient graphs show that in region C, the coefficients are alternating between positive and negative values, indicating a complex situation without a unified high-value area. The PDF analysis results conform to this pattern. Corresponding coefficients Most of them are negative, therefore In region C Emissions have a negative effect, while positive effects are... and right The impact of emissions is significant.

[0096] This invention provides a system for retrieving black carbon emissions, comprising:

[0097] The data acquisition module is used to obtain a black carbon emission inventory including black carbon emissions, and to obtain multiple influencing factors affecting black carbon emissions, including nitrogen dioxide, ultraviolet aerosol index and wind.

[0098] The model building module is used to perform spatiotemporal decomposition and dimensionality reduction on the black carbon emission inventory using an empirical orthogonal function model to obtain multiple emission source regions including black carbon emission sources. Within each emission source region, considering the linear and nonlinear interactions between multiple influencing factors, a nonlinear regression model between the interaction of multiple influencing factors and black carbon emissions is established.

[0099] The black carbon emission inversion module is used to input multiple influencing factors into a nonlinear regression model to obtain black carbon emissions and complete the inversion of black carbon emissions.

[0100] This invention provides a computer device, including: a memory and a processor; the memory stores a computer program, and the processor executes the computer program to implement the steps of a method for retrieving black carbon emissions.

[0101] This invention provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of a method for retrieving black carbon emissions.

[0102] A specific example is as follows:

[0103] Analysis of the spatiotemporal distribution and transmission characteristics of black carbon emission inventory.

[0104] A black carbon emission inventory from 2011 to 2018 was derived using a linear regression model. The spatiotemporal characteristics of the derived black carbon emission inventory were analyzed. Two high-value emission source areas, region A and region B, were selected. Global and these two high-value emission source areas' black carbon emission inventories were calculated based on annual, quarterly, and monthly average values, and their spatiotemporal differences were analyzed and compared. Utilizing the similarity and common origin of black carbon and carbon monoxide (CO), overlapping areas were calculated using MOPITT CO data and black carbon data to compare the spatiotemporal transport of black carbon and CO, verifying the regional accuracy of the black carbon emission inventory. Simultaneously, CO transport in different atmospheric layers was used to indicate the transport of black carbon emissions. The main conclusions are as follows:

[0105] After calculating the black carbon emission inversion inventory for the global market and regions A and B, it was found that:

[0106] 1. Global Situation. Global black carbon emissions generally experienced a process of first increasing, then decreasing, and then increasing again between 2011 and 2018. Specifically, there was a slight increase between 2011 and 2013, a decrease between 2014 and 2015, and a slow increase again between 2016 and 2018. High emission areas are mainly concentrated in region A in eastern country A, region B in country B, region C in countries Z and X, as well as countries V, N, G, H, and F. Although the months in which black carbon emissions reach their minimum vary from year to year, they are generally concentrated between June and August. This may be because most major emission source regions in the Northern Hemisphere experience a decrease in emissions during the summer rainy season or the season of changing wind direction, especially a reduction in winter heating emissions. Conversely, global black carbon emissions reach their maximum from November to February, which is precisely the Northern Hemisphere winter, when increased combustion for heating leads to a rise in black carbon emissions. In addition, this period is also a time of frequent celebrations in many countries, resulting in short-term concentrations of black carbon emissions.

[0107] 2. Northern, Eastern, and Central Regions of Country A. From 2011 to 2013, black carbon emissions in the northern, eastern, and central regions of Country A gradually increased, followed by a downward trend from 2013 to 2015. Emissions in 2016 were basically the same as in 2015, while from 2016 to 2018, black carbon emissions in this region slowly increased. Overall, the annual average time series trend of this region is similar to the global annual average time series, and the trend of black carbon emission changes is also relatively consistent. Black carbon emissions in the northern, eastern, and central regions of Country A show obvious seasonal variations, with smaller high-emission areas in summer and the largest in winter. The seasonal average emission time series is highly consistent with the global seasonal average emission time series trend. Monthly average emissions show obvious changes. From February to August, the area of ​​high-emission areas gradually shrinks, especially in many eastern provinces and the eastern region where the changes are more obvious. In addition, black carbon emissions in region A continued to decline from January to June each year, reaching the lowest value in June, then fluctuated slightly upward between June and September, and continued to increase from September to December.

[0108] 3. Region B. The high-emission areas in Country B within Region B are mainly concentrated near the northern plains. This region is one of the most densely populated areas in the world and a major agricultural, political, economic, and cultural center of Country B. Black carbon emissions in this region increased significantly between 2015 and 2017. Based on the annual average black carbon emissions data and corresponding time series for Region B, it was found that black carbon emissions increased slightly from 2011 to 2012, declined slowly from 2012 to 2015, and then increased sharply before declining again between 2015 and 2017. The quarterly average time series of black carbon emissions in Region B is also highly consistent with the global quarterly average time series trend, with emissions decreasing from spring to summer and then increasing again after summer. The monthly average data for black carbon emissions in Region B is more complex, with the highest emissions in December and March. The high-emission area in Region B shrinks significantly in May, emissions increase somewhat in June and July, and decrease again in August and September.

[0109] This invention primarily focuses on the correlation analysis between black carbon and nitrogen dioxide, ultraviolet aerosol index, and wind; the analysis of the impact of each factor on black carbon; and the analysis of the spatiotemporal distribution patterns of black carbon emission inventories. This research contributes to establishing an accurate black carbon emission inventory, better understanding the sources, distribution, and impacts of black carbon, providing a scientific basis for reducing black carbon emissions, and facilitating subsequent global black carbon emission reduction efforts. The main conclusions are as follows:

[0110] 1. The scaling factor between the constructed prior emission inventory and the EDGAR emission inventory ranges from 2.21 to 12.06. High-value emission areas include Region A, Region B, and Region C, with high-value years from 2010 to 2013. Black carbon emissions reached an inflection point in 2010, increased year by year from 2005 to 2010, and then decreased year by year from 2011 to 2018.

[0111] 2. The correlations, coefficient distributions, and dominant influencing factors of various elements with black carbon differ across regions. In regions A and B, nitrogen dioxide has the greatest impact on black carbon emissions, with both nitrogen dioxide and ultraviolet aerosol index showing a positive correlation with black carbon. In region C, while nitrogen dioxide remains the primary influencing factor, the correlation is more complex; nitrogen dioxide shows a negative correlation with black carbon in some areas, while ultraviolet aerosol index is generally positively correlated with black carbon. In region C, and in region D, nitrogen dioxide has a greater impact on black carbon emissions than wind and ultraviolet aerosol index. In region F of country H, wind has the greatest impact on black carbon emissions. Globally, nitrogen dioxide has the greatest impact on black carbon emissions, followed by ultraviolet aerosol index, with wind having the least impact. The impacts of meteorological factors and pollutants on black carbon emissions vary significantly across different regions, exhibiting clear regional characteristics.

[0112] 3. The black carbon emission inventory exhibits significant interannual and seasonal variations. Globally, from 2011 to 2018, black carbon emissions generally showed a trend of first increasing, then decreasing, and then increasing again. Black carbon emissions reached their minimum from June to August and their maximum from November to February, which is related to the summer rainy season or increased combustion for heating in the Northern Hemisphere winter. In the northern, eastern, and central regions of country A, black carbon emissions gradually increased from 2011 to 2013, then decreased from 2013 to 2015, and slowly increased from 2016 to 2018. Black carbon emissions show obvious seasonal variations, with lower emissions in summer and maximum emissions in winter. In region B, the seasonal variation of black carbon emissions is consistent with the global trend, decreasing from spring to summer and increasing after summer. Furthermore, black carbon emissions in region B are mainly concentrated near the plains in northern country B, with a significant increase between 2015 and 2017. Picture 4 The results shown are the findings of this invention.

[0113] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these all fall within the protection scope of the present invention. Therefore, the protection scope of this invention patent should be determined by the appended claims.

Claims

1. A method for retrieving black carbon emissions, characterized in that, Includes the following steps: Obtain a black carbon emission inventory including black carbon emissions, and identify multiple factors affecting black carbon emissions, including nitrogen dioxide, ultraviolet aerosol index, and wind. The black carbon emission inventory is decomposed and dimension-reduced using an empirical orthogonal function model to obtain multiple emission source regions including black carbon emission sources. Within each emission source region, considering the linear and nonlinear interactions between multiple influencing factors, a nonlinear regression model between the interaction of multiple influencing factors and black carbon emissions is established. By inputting multiple influencing factors into a nonlinear regression model, the amount of black carbon emissions is obtained, thus completing the inversion of black carbon emissions. The specific steps for establishing a nonlinear regression model between black carbon emissions and the interaction of multiple influencing factors include: Several factors reflecting black carbon emissions include nitrogen dioxide. NO 2 UV aerosol index UVAI With the wind Wind The linear interaction between influencing factors and black carbon emissions needs to be considered, as well as the nonlinear interaction between influencing factors, including the ultraviolet aerosol index. UVAI With the wind Wind The interaction between nitrogen dioxide and nitrogen dioxide NO 2 With the wind Wind The interaction; The formula for the established nonlinear regression model is as follows: in, , , , and Nitrogen dioxide NO 2 UV aerosol index UVAI ,wind Wind ,wind Wind With UV aerosol index UVAI Interaction and wind Wind With nitrogen dioxide NO 2 The coefficient of interaction, It is a constant.

2. The method for retrieving black carbon emissions as described in claim 1, characterized in that, The specific steps for obtaining multiple emission source regions, including black carbon emission sources, include: The empirical orthogonal function model was used to obtain multiple eigenvalue contribution rates for the black carbon emission inventory. The top three largest eigenvalue contribution rates were selected from all eigenvalue contribution rates and labeled as EOF1, EOF2 and EOF3 respectively from largest to smallest. Values ​​greater than 0.5 and less than -0.5 in the eigenvector corresponding to eigenvalue contribution rate EOF1 are filtered; values ​​greater than 0.8 and less than -0.8 in the eigenvector corresponding to eigenvalue contribution rates EOF2 and EOF3 are filtered. Based on the numerical differences in the filtered results, the geographical locations of the preliminary emission source regions are obtained from the black carbon emission inventory; Regions with an area of ​​less than 100 pixels in the initial emission source area are excluded, and the boundary of the excluded initial emission source area is extracted using an edge detection algorithm, thus dividing the area into six emission source regions.

3. The method for retrieving black carbon emissions as described in claim 1, characterized in that, The specific steps for obtaining black carbon emissions by inputting multiple influencing factors into a nonlinear regression model include: Obtaining nitrogen dioxide NO 2 UV aerosol index UVAI Japanese style Wind The monitoring data from multiple months is input into a linear regression model to obtain the black carbon emissions for each month. The annual, quarterly, and monthly average black carbon emissions were analyzed based on the obtained black carbon emissions, and the differences were compared with the black carbon emission inventory to determine the accuracy of the inversion results.

4. The method for retrieving black carbon emissions as described in claim 1, characterized in that, The specific steps for obtaining the black carbon emission inventory include: Based on historical black carbon emissions at a global scale, an optimized estimate of black carbon emissions is obtained using the Kalman filter method. Based on the difference between the optimized estimate of black carbon emissions and historical black carbon emissions, a scaling factor is obtained to make the black carbon emissions conform to the actual situation. By scaling the monthly black carbon emission data from the Global Atmospheric Research Emissions Database (EDGAR), a black carbon emission inventory can be obtained.

5. A black carbon emission inversion system applied to the black carbon emission inversion method according to any one of claims 1 to 4, characterized in that, include: The data acquisition module is used to acquire a black carbon emission inventory including black carbon emissions, and to acquire multiple influencing factors affecting black carbon emissions, including nitrogen dioxide, ultraviolet aerosol index and wind. The model building module is used to perform spatiotemporal decomposition and dimensionality reduction on the black carbon emission inventory using an empirical orthogonal function model to obtain multiple emission source regions including black carbon emission sources. Within each emission source region, considering the linear and nonlinear interactions among multiple influencing factors, a nonlinear regression model is established between the interaction of multiple influencing factors and black carbon emissions. The black carbon emission inversion module is used to input multiple influencing factors into a nonlinear regression model to obtain black carbon emissions and complete the inversion of black carbon emissions.

6. A computer device, comprising: Memory and processor; The memory stores a computer program, characterized in that when the processor executes the computer program, it implements a method for inverting black carbon emissions according to any one of claims 1 to 4.

7. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements a method for inverting black carbon emissions as described in any one of claims 1 to 4.

Images