Satellite remote sensing inversion method for port carbon dioxide emission

Abstract

The invention discloses a satellite remote sensing inversion method for the emission load of carbon dioxide in a port. The method comprises the following steps: dividing grid regions according to the geographic position of the port by utilizing the hourly NO2 and SO2 column concentrations observed by a GEMS satellite, retaining grid data with strong correlation, performing data reconstruction, taking port CEMS monitoring data as prior emission, and calculating the emission load of NOx and SO2. The emission scale factors of CO2 / NOx and CO2 / SO2 are calculated on the basis of an emission list from bottom to top, downscaling processing is carried out on the emission scale factors of the port through four space proxy indexes taking traffic density as a main part, and the specific emission scale factors of the port are obtained. And calculating to obtain a CO2 emission result under the high temporal-spatial resolution, mutually verifying the CO2 emission result to correct the CO2 emission, and popularizing the CO2 emission result to a long-time sequence result. The method solves the problems that carbon satellite data cannot directly capture port carbon emission, GEMS precision is poor, space loss exists, and long-time sequence observation of port emission cannot be achieved through foundation monitoring values.

Description

Technical Field

[0001] This invention belongs to the field of atmospheric environment remote sensing application technology, and specifically relates to a satellite remote sensing inversion method for port carbon dioxide emissions. Background Technology

[0002] As a crucial node in cargo transportation, ports contain numerous mobile and non-mobile sources of air pollutants, whose emissions have become a key factor affecting the ambient air quality of port cities. Studies show that in large port areas, port machinery and transport vessels are the largest sources of various air pollutants and greenhouse gases among both road traffic and ship-borne mobile sources. Among these, nitrogen oxides (NOx) account for a significant portion. x Carbon dioxide (CO2) is the most emitted air pollutant, while carbon dioxide is the most emitted greenhouse gas. Currently, the limited number of ground-based monitoring stations in port areas restricts a comprehensive and accurate assessment of pollution levels and CO2 emissions from port activities. Emission inventory compilation, as the primary technical means of analyzing port carbon emission sources, often faces limitations such as high uncertainty and time lag. In contrast, satellite remote sensing technology, due to its wide coverage and relatively low cost, is gradually becoming an important method for quantitatively monitoring atmospheric composition and calculating emissions. Therefore, using satellite remote sensing observation data to quantitatively analyze port CO2 emissions provides a novel approach.

[0003] However, due to the narrow swath and long transit period (4 to 6 days) of carbon satellite sensors, it is difficult to achieve hourly and full-space coverage of port CO2 emissions. Summary of the Invention

[0004] GEMS satellite observation data boasts high hourly temporal resolution and greater spatiotemporal coverage compared to carbon satellites. The Intergovernmental Panel on Climate Change (IPCC), in its latest 2019 revised report, reiterated that "top-down" approaches, i.e., atmospheric measurements, should be given high priority by the scientific community as independent validation data. This invention proposes a satellite remote sensing inversion method for port carbon dioxide emissions, aiming to combine port NOx emissions with... x CO2 emissions are indirectly calculated and cross-validated using SO2 emissions and port-specific emission ratio factors. Simultaneously, short-term eddy covariance flux tower monitoring is used to correct satellite observation estimates of CO2 emissions, which are then extended to long-term series results, thus obtaining robust port CO2 emission results for hourly and 100-meter grids.

[0005] The satellite data involved in this invention mainly includes tropospheric NO2 and SO2 column concentration data from the Geostationary Environmental Monitoring Spectrometer (GEMS), CEMS data from the Continuous Flue Gas Monitoring System, bottom-up emission inventory data, eddy covariance flux tower measured data, and various auxiliary data. The following is a brief introduction to the production technologies and parameters of the above five main data types.

[0006] The Geostationary Environment Monitoring Spectrometer (GEMS) is a high-performance geosynchronous orbit scanning device that primarily covers the ultraviolet and visible light spectral range.

[0007] A Continuous Emission Monitoring System (CEMS) is a monitoring device specifically designed for stationary pollution sources. It can monitor particulate matter concentration and gaseous pollutant concentrations (such as SO2 and NO). x ) and its total emissions are continuously and automatically monitored.

[0008] The "bottom-up" emissions inventory model covers emissions inventory data including carbon dioxide (CO2) and pollutants such as SO2 and NO. x The system constructs a unified framework for classifying and grading emission sources, as well as a detailed database of emission factors.

[0009] The eddy covariance flux tower, the IRAGSON integrated eddy covariance flux system, organically combines an open-circuit infrared CO2 / CH4 / H2O analysis sensor with a three-dimensional ultrasonic anemometer, enabling the simultaneous measurement of CO2, CH4, and H2O gas concentrations, three-dimensional wind speed, and ultrasonic temperature in the air.

[0010] Ancillary data, including various types of ancillary data, are used in the analysis of spatial proxy indicators.

[0011] To address several technical challenges in estimating port CO2 emissions, including: 1) the limited spatial coverage and low temporal frequency of existing carbon satellite data, making it impossible to directly capture port carbon emissions; 2) while GEMS satellites offer high hourly temporal resolution, their accuracy is poor and spatial gaps exist; and 3) ground-based monitoring data cannot achieve long-term series observations of port emissions. This invention aims to provide a satellite remote sensing inversion method for port carbon dioxide emissions.

[0012] The specific technical solution is as follows:

[0013] A satellite remote sensing inversion method for port carbon dioxide emissions includes the following steps:

[0014] (1) Based on the geographical distribution of ports, this invention divides port area grids using hourly long-term series column concentration data of nitrogen dioxide (NO2) and sulfur dioxide (SO2) acquired by the Geostationary Environment Monitoring Spectroradiometer (GEMS). Through a rigorous quality control process, the GEMS data is compared and verified with data from the Tropospheric Monitoring Instrument (TROPOMI) and GaoFen-5 satellite, and port grid data with significant correlations are selected and retained. Subsequently, the satellite data is optimized and reconstructed using the Data Interpolating Empirical Orthogonal Functions (DINEOF) method. This method, based on empirical orthogonal functions (EOF), reconstructs the missing port observation portions of the GEMS data after quality control and screening by constructing an empirical interpolation model.

[0015] (2) Targeting the pollutant gas nitrogen oxides (NOx) x This invention calculates the spatial divergence and temporal gradient of NO2 based on the reconstructed port GEMS NO2 tropospheric column concentration grid and meteorological data. Subsequently, these data, along with concentration data, are substituted into the mass conservation equation. Furthermore, NO2 obtained from the port area continuous emission monitoring system (CEMS) is used... x Total emissions data serve as a priori emission terms in the mass conservation equation.

[0016] (3) This invention employs a multiple linear regression method to fit and solve for the port emission characteristics (coefficients) pixel by pixel, and calculates their probability density distribution. Simultaneously, a bootstrap method is used to obtain the confidence intervals for each coefficient. Random repeated sampling is performed within the confidence intervals, and the sampling coefficient results are substituted into the equation. Combining satellite-observed column concentration data and various derived terms, the NO2 at the port is calculated. x Emissions samples. Ultimately, the mean of all samples is used as the port's hourly, grid-by-grid NO. x Emissions results.

[0017] (4) For the pollutant gas SO2, the SO2 vertical column concentration in the reconstructed port GEMS was treated in the same way as in steps (2) to (3). (The text abruptly ends here, likely due to an incomplete sentence or missing information.) x In comparison, the mass conservation equation for SO2 adds a quadratic gradient term to account for its spatial diffusion effect. By obtaining the corresponding emission characteristic coefficients, the hourly and grid-by-grid SO2 emissions of the port are calculated.

[0018] (5) Annual anthropogenic SO2 and NO2 emissions based on bottom-up emission inventories xThe carbon dioxide (CO2) emission data was interpolated to the selected port grid cell in GEMS to calculate SO2 and NO2 emissions separately. x The ratio factors of CO2 emissions to SO2 and NO2 are: x .

[0019] (6) The emission ratio factor obtained in step (5) is downscaled, and four spatial proxy indicators (traffic density, emission source density, GDP distribution, and population density) are selected, with different weights assigned to each indicator. Subsequently, the emission ratio is processed grid by grid using a normalization method to finally obtain the port-specific emission ratio factor.

[0020] (7) SO2 and NO obtained from steps (3-4) x Based on the port-specific emission ratio factor determined in step (6), the hourly high-resolution CO2 emissions of the port are estimated (based on SO2 and NO2). x By comparing the estimation results of the two methods, cross-validation was performed within the error range, and outliers were removed. Subsequently, the validated robust CO2 emissions were time-matched with short-term measured data on ground vorticity in the port area to obtain a correction coefficient, which was then applied to correct the long-term series results.

[0021] Further, in step (1), firstly, GEMS data products (including NO2 and SO2 column concentrations) are downloaded, and grids containing the port are divided according to the port's geographical location. Quality control is performed on these grid data based on three indicators: solar zenith angle, reflectivity, and cloud cover. The data is then compared and verified with TROPOMI data and Gaofen-5 data, retaining hourly results from GEMS grids with strong correlation to both (correlation coefficient greater than 0.9) to ensure accuracy. Subsequently, the DINEOF method is used to optimize and reconstruct the missing data. This method is based on spatial empirical orthogonal function (EOF) decomposition and temporal principal component (PC) decomposition to identify the spatiotemporal domains with the greatest variation. This method can make predictions under conditions of missing observation data in time and / or space. By iteratively using weighted EOFs and PCs, the missing data points can be resynthesized. The optimal number of iterations that minimizes cross-validation error is used to obtain the best reconstructed data.

[0022] In step (2), based on the "top-down" reverse characterization method and the principle of atmospheric mass conservation, an hourly NO... x Emission estimation equations were established to correlate port satellite NO2 concentrations with monitored NO2 levels. x The correlation of emissions and the quantification of their impact on port NO x The physical and chemical factors of emissions ultimately determine the NO emissions.x Emissions. Using GEMS NO2 column concentration data reconstructed using the DINEOF method, the rate of change of NO2 column concentration during two satellite passes was calculated, and the derivatives of the daily zonal and meridional fluxes were calculated as the daily NO emissions. x Spatial divergence of horizontal flux. Furthermore, NO obtained using the CEMS system. x Emission data serves as a priori emission term in the mass conservation equation. The CEMS system continuously and automatically monitors particulate matter concentrations, gaseous pollutant concentrations, and total pollutant emissions from stationary sources. NO in the CEMS system... x The measurement principle is based on chemical analysis methods, which calculate NO by measuring the concentrations of NO and NO2. x The concentration.

[0023] In step (3), based on the mass conservation model constructed in step (2), a multiple regression (multiple least squares fitting, MLR) method is used to fit the port's GEMS data to the time period containing effective CEMS monitoring data on a daily, pixel-by-pixel basis, to obtain various coefficients. These coefficients represent the port's NO x Emission characteristics (NO) x / NO2 ratio, chemical lifetime, and transport distance). Confidence intervals for the coefficients are estimated daily using the Bootstrap method; coefficients within these intervals are considered reasonable. A coefficient matrix is ​​generated daily, and the same coefficient matrix is ​​used for all hourly emission models for that day. Finally, repeated random sampling is performed on the coefficient matrix and substituted into the equations, combined with the long-term hourly GEMS NO2 column concentrations and related derivatives from step (2), to obtain a large number of emission samples. The mean of all samples is used as the final NO2 emission value. x Estimated value.

[0024] In step (4), the port data based on the sulfur dioxide (SO2) column concentration observed by GEMS is processed in the same way as steps (2) to (3). Using the SO2 emission data monitored by the port's CEMS as prior information, in NO... x Based on this, a quadratic gradient term for SO2 is introduced to obtain the corresponding emission characteristics (including chemical lifetime, transport distance, and diffusion rate). These characteristic coefficients are then substituted into the mass conservation equation to finally calculate the SO2 emissions.

[0025] In step (5), based on the bottom-up emission inventory, SO2 and NO from anthropogenic sources are obtained. x The CO2 emission data was interpolated into the port GEMS observation grid selected in step (1) and then converted into emission flux (unit: μg / m³). 2 / s). Calculations were performed to obtain the port's CO2 and proxy species (SO2 and NO). xThe emission ratio factor is used to reflect the level of fossil fuel combustion and emissions in the port area.

[0026] In step (6), since the resolution of the inventory data is usually coarse, and the port area is relatively small compared to large cities and industrial parks, in order to obtain a higher resolution port-specific emission proportion factor to achieve spatial resolution refinement, this invention adopts multiple emission proxy indicators and performs downscaling processing on the emission proportion factor obtained in step (5) based on the normalization method. In large port areas, there are a large number of mobile and non-mobile air pollutant emission sources (including various road traffic mobile sources and ship sources), among which port machinery and transport ships account for the largest proportion of emissions of various air pollutants and greenhouse gases. Therefore, traffic density is used as the most important spatial proxy indicator. The increase in exhaust emissions from road mobile sources (motor vehicles), changes in the combustion efficiency of heating boilers, and the increase in emissions from ships and railway internal combustion engines may all lead to changes in the port's emission proportion factor. Based on this, this invention combines port traffic density grid, GDP spatial distribution grid, pollution discharge sector density grid, and population density distribution grid datasets as four spatial proxy indicators. Within the resolution grid range corresponding to the inventory, the Min-Max Scaling Normalization method is used for each inventory grid to center the data according to the minimum value of the spatial proxy parameter, and then scale it according to the range (maximum-min). The order of the values ​​remains unchanged, and different weights are assigned to obtain the port-specific emission ratio factor.

[0027] In step (7), the port-specific emission ratio factor calculated in step (6) is used in conjunction with the port SO2 and NO obtained in steps (3-4). x Emissions, estimating CO2 emissions. The specific calculation process is as follows: First, through SO2 and NO... x CO2 emissions are solved indirectly and cross-validated within the error range. If the results are consistent within the error range, the robust emissions grid results are retained; if outliers are found, they are discarded.

[0028] This invention employs the IRAGSON integrated eddy covariance flux system for short-term monitoring of port areas. This system integrates an open-circuit infrared CO2 analysis sensor and a three-dimensional ultrasonic anemometer, simultaneously measuring airborne CO2 concentration, three-dimensional wind speed, and ultrasonic temperature. This avoids high-frequency flux omissions caused by cross-spatial measurements, effectively improving measurement accuracy. The system records CO2 emission flux and concentration data at 30-minute sampling intervals. To obtain more accurate indirect CO2 emission estimates, the satellite CO2 estimates are corrected using the following method: First, the difference between the satellite-indirectly estimated CO2 emissions and the eddy covariance monitoring values ​​is calculated within the monitoring period of the eddy covariance flux system in the port unit, thus obtaining a correction coefficient. Subsequently, this coefficient is used to adjust the "top-down" CO2 emission estimates for the port area, and the correction coefficient from the monitoring period is extended to long-term CO2 emission results to improve accuracy. Attached Figure Description

[0029] Figure 1 This is the overall flowchart of the present invention. Detailed Implementation

[0030] The invention will now be further described with reference to the accompanying drawings.

[0031] like Figure 1 As shown,

[0032] This embodiment includes four parts: port satellite data preprocessing, port pollutant gas emission estimation, port-specific emission ratio factor, and port CO2 emission verification and correction.

[0033] 1. Pre-processing of port satellite data

[0034] A buffer zone was established based on the port's geographical coordinates and its size range, and a GEMS satellite grid was selected based on the buffer zone range. The preprocessing of the GEMS satellite dataset included two core steps: "Step 1: Quality Control and Comparative Verification" and "Step 2: DINEOF Data Reconstruction".

[0035] Step 1: Quality Control and Comparative Verification

[0036] This step first filters GEMS satellite data using parameters such as cloud cover, and then verifies the data by comparing it with TROPOMI data and Gaofen-5 data, retaining grid data with an R value greater than 0.9 for comparison and verification.

[0037] Step 2: DINEOF Data Reconstruction

[0038] The DINEOF algorithm initializes all missing data with the same predicted value in the initial reconstruction stage. It then uses the Empirical Orthogonal Function (EOF) method to iterate and cross-validate the dataset to obtain the best reconstruction result. This method has several unique advantages, including no need for prior values, strong adaptability, and high spatiotemporal correlation, making its interpolation results particularly suitable for reconstructing large areas of missing data.

[0039] The core technologies are explained as follows:

[0040] The preprocessing step "Step 1: Quality Control and Comparative Verification" is explained as follows:

[0041] GEMS satellite data was filtered using parameters such as cloud cover to ensure the state and quality of the inversion results. The coverage of the GEMS dataset remained stable above 95% before quality control and above 30% after quality control, indicating that quality control had a significant impact on the amount of GEMS data. Cloud cover was the indicator with the greatest impact on the amount of GEMS data during the quality control process. The relevant parameters are as follows:

[0042] Table 1

[0043]

[0044] A bilinear interpolation algorithm was used to resample the GEMS, TROPOMI, and Gaofen-5 data for the port area to a uniform 0.05°×0.05° grid. In the comparative verification of TROPOMI and Gaofen-5 data, the daily averages of both were first calculated. Then, the observations from GEMS data at 12:45 and 13:45 Beijing time were extracted and their averages calculated. The integrated dataset was used for comparative analysis. After quality control, all indicators were significantly optimized, with a particularly significant improvement in the correlation coefficient R. After quality control, grids with R values ​​greater than 0.9 and their corresponding hourly data for each date were selected. These data showed high correlation between GEMS, TROPOMI, and Gaofen-5 to ensure the accuracy of the input satellite data.

[0045] Preprocessing "Step 2: DINEOF Data Reconstruction"

[0046] The DINEOF method was used to reconstruct the spatiotemporal distribution data of SO2 and NO2 to fill in missing values. The data was imported as time-series imagery and processed using DINEOF to construct a matrix consisting of the number of spatial pixels and the temporal dimension. Then, 1% of the valid data was randomly selected as the cross-validation set and set to contain missing values. After a logarithmic transformation of the dataset, anomaly reduction was performed, and the values ​​of missing points were set to 0 (considered as unbiased estimates). The processed matrix was then used to fill in the missing data through singular value decomposition, and the root mean square error (RMSE) of this reconstruction was calculated. Next, the matrix after one decomposition was used to replace the original data matrix, and the decomposition process was repeated. The iteration was considered to have converged when the RMSE results at the cross-validation points of the first and second iterations were lower than a predetermined threshold. During this process, the above steps were repeatedly executed, and the corresponding RMSE values ​​were recorded to ensure the minimization of cross-validation error. Finally, based on the optimal number of modes determined at this point, the reconstruction steps were executed again to complete the reconstruction of the missing values, achieving a significant optimization of the spatiotemporal coverage and continuity of satellite observation products.

[0047] 2. Estimation of pollutant emissions

[0048] Based on optimized and reconstructed satellite and meteorological data, a mass conservation equation was constructed, and the spatial divergence, temporal gradient, and quadratic gradient terms of the pollutants were calculated and substituted into the equation. SO2 and NO were obtained from the Continuous Emission Monitoring System (CEMS) in the port area. x Total emissions monitoring data is used as a priori emission data. A multiple linear regression method is employed to fit the port's emission characteristics (coefficients) pixel-by-pixel, and their probability density distribution is calculated. Simultaneously, a bootstrap method is used to obtain the confidence intervals of the coefficients. Substituting the obtained coefficients into the equation and combining them with column concentration data from satellite observations, the final SO2 and NO emissions at the port are calculated. x Emissions.

[0049]

[0050]

[0051] (1) The meanings of the variables in the formula are as follows: and These represent the time gradients of pollutant gases (daily rate of change of column concentration); and The column represents the concentration data of polluting gases observed by satellite; w represents the wind field data. and This represents the rate of change in column concentration per unit volume of space despite changes in wind speed and direction. The second derivative of SO2 concentration per unit volume of column space. and This represents the amount of pollutant emissions. The coefficient variables have the following meanings: α1, α2, and α3 are respectively related to NO... x NO x / NO2, chemical lifetime, and transport distance are related; represents the time gradient of pollutant gas (daily rate of change of column concentration); γ2, γ3, and γ4 are related to the chemical lifetime, transport distance, and physical diffusion of SO2, respectively.

[0052] Regarding NO x Emissions estimation using GEMS satellite data reconstructed from DINEOF Using meteorological data w, and based on a top-down inverse characterization method and the principle of atmospheric mass conservation, the spatial divergence term and temporal gradient term are calculated, and then substituted together with the concentration data into the mass conservation equation. The rate of change of NO2 column concentration during the two satellite passes is calculated. Calculate the derivatives of daily zonal and meridional fluxes As daily NO x Spatial divergence of horizontal flux. These two terms are then compared with column concentration. These three terms are themselves part of the mass conservation estimation equation. Secondly, for SO2 emission estimation, the sulfur dioxide (SO2) column concentration from GEMS is used. The observational data, in NO x Based on the existing terms, SO2 adds a quadratic gradient term.

[0053] (2) Using CEMS data as an emission prior, conduct on-site monitoring of pollutant emissions, including gaseous pollutants (SO2 and NO). x The monitoring unit mainly includes a gaseous pollutant collector, a flue gas pretreatment device, a gas controller, and a gaseous pollutant analyzer. The sample collection device should have heating, heat preservation, and backflushing purification functions. Its heating temperature is generally above 120°C and should be at least 10°C higher than the flue gas dew point temperature. The CEMS control function coordinates the timing of the entire system, and the system can automatically process the collected and recorded real-time data into 1-minute and hourly data. The formula for the hourly emission rate of flue gas pollutants is as follows: G h The emission rate (kg / h) of pollutants monitored by CEMS in hour h is represented by the corresponding unit conversion (μg / m³). 2 / s), corresponding to the hourly data time of the GEMS satellite, serves as a priori term in the mass conservation equation. and

[0054] In the formula, C QhThis represents the average dry-basis standard mass concentration of pollutant emissions measured by CEMS in the h-hour period. Hourly data should include at least 45 minutes of valid minute data within that hour. Q snh This represents the emission flow rate of dry flue gas in the h-hour period under standard conditions.

[0055] (3) After successfully constructing the mass conservation equation, the multivariate regression (multiple least squares fitting, MLR) method was used to perform hourly and pixel-by-pixel fitting. x Obtain the coefficients (α1, α2, and α3) excluding prior emissions; these coefficients represent the port's NOx emissions. x Emission characteristics (NO) x (NO2, chemical lifetime, and transport distance). SO2 also yields corresponding emission characteristic coefficients (chemical lifetime γ2, transport distance γ3, diffusion rate γ4). Confidence intervals for the coefficients are estimated daily using the Bootstrap method; coefficients within these intervals are considered reasonable. A coefficient matrix is ​​generated daily, and the same matrix is ​​used for all hourly emission models for that day. Multiple repeated random samplings are performed on the coefficient matrix, substituted into the equations, and combined with various terms to obtain a large number of emission samples. The mean of all emission samples is then used as the final emission estimate. and

[0056] 3. Estimation of Port-Specific Emission Ratio Factors

[0057] (1) Obtain SO2 and NO from the entire sector in a "bottom-up" manner based on the emission inventory. x CO2 emission data were converted into emission flux (unit: μg / m³). 2 / s). SO2 and NO emitted by pollutants x The dataset (using the highest spatiotemporal resolution and the latest dataset version) is used as the prior dataset. Interpolation was performed using the nearest neighbor interpolation method, and the data was re-gridted to a GEMS satellite port-specific grid. Additionally, a CO2 dataset provided from the carbon emissions inventory was used. Compared with the above SO2 and NO x The dataset was processed in a consistent manner, and the CO2 and proxy species (SO2 and NO) at the port were calculated. x Transformation factor of Proxies emissions The equation reflecting the level and emissions of fossil fuel combustion in the port area is shown below:

[0058]

[0059] (2) Since the resolution of inventory data is usually coarse, and the port area is relatively small compared to large cities and industrial parks, in order to obtain a higher resolution port-specific emission ratio factor for refined mapping, multiple emission proxy indicators are used, and the obtained emission ratio factors are downscaled based on a normalization method. The auxiliary data processing for the emission proxy indicators is as follows:

[0060] Spatial proxy indicator 1: Traffic density

[0061] In large port areas, there are numerous mobile and non-mobile sources of air pollutants, including various road traffic sources and ship sources. Among these, port machinery and transport vessels contribute most significantly to the emissions of various air pollutants and greenhouse gases. Therefore, traffic density is considered the most important spatial proxy indicator. For ports with a large amount of outsourced business, indirect emissions from leasing ships, trucks, cargo handling equipment, and railway locomotives may be the main source of CO2 emissions. By combining national road network hierarchical data, port pallet vector data, and ship berthing and cargo handling records for analysis, the number of traffic roads, pallet numbers, and loading / unloading frequencies within a 100-meter grid were statistically analyzed. The density distribution was then calculated by integrating the grid area, ultimately obtaining a 100-meter resolution port traffic density data set.

[0062] Spatial proxy indicator 2: Spatial distribution of GDP

[0063] A kilometer-level grid dataset of China's GDP spatial distribution was used as one of the proxy indicators for CO2 emission downscaling analysis. This dataset has significant weight in the downscaling process because the industrial gross domestic product (IGDP) of the industrial sector more accurately reflects the indirect carbon emissions generated by electricity consumption in port operations. The GDP distribution data is sourced from platforms such as the Resource and Environmental Science and Data Center. Each 1 square kilometer grid cell represents the total GDP of that region, expressed in RMB 10,000 per square kilometer.

[0064] Spatial proxy indicator 3: Distribution of industrial emission sources

[0065] Another proxy indicator is based on the location distribution of various emission sources around the port, aiming to exclude the impact of industrial sources and thermal power plant emissions (excluding the port) on port emissions within the emission inventory and satellite observation grid. For this purpose, location data of four high-energy-consuming facilities—power plants, steel mills, combined heat and power plants, and cement plants—were selected. This location data came from the Ministry of Ecology and Environment's Discharge Permit Management Information Platform (http: / / permit.mee.gov.cn), which provides information such as the name, city, and latitude and longitude of the emission sources. The number of emission sources within a 1km radius was statistically analyzed, and an emission sector density grid was generated.

[0066] Spatial proxy indicator four: population density distribution

[0067] The 1km grid dataset of China's population density spatial distribution is also used as one of the proxy indicators. It is provided by platforms such as WorldPop. This data helps to more accurately reflect the impact of population distribution on port CO2 emissions.

[0068] Increased emissions from roadside mobile sources (motor vehicles), changes in the combustion efficiency of heating boilers, and increased emissions from ship and railway diesel engines will all directly or indirectly affect port areas. Emission proportion factors. These factors cause dynamic changes in proportion factors by altering the source composition and emission characteristics of carbon emissions within a region, especially in port areas with frequent activity or complex environmental conditions. Therefore, to improve the scientific rigor and adaptability of calculations, it is necessary to introduce spatial proxy indicators that comprehensively consider multiple influencing factors.

[0069] Based on this, and combining the four spatial proxy indices D mentioned above... s This includes: 1) Port traffic density grid D Trans 1) Reflects the distribution and activities of road mobile sources (such as motor vehicles and freight fleets) in the port area; 2) GDP spatial distribution grid D GDP 3) The density grid D of the pollution discharge sector represents the contribution of the port and surrounding economic activity level to carbon emissions; Indus 4) Quantify the distribution and density of industrial emission sources; 5) Population density distribution grid D Pop As an important reference factor for emissions from residential sources.

[0070] Within the resolution grid range corresponding to the emission proportion factor, the spatial proxy index D is applied grid-by-grid using the Min-Max Scaling Normalization method. s The standardization process involves centering the values ​​according to the minimum value of the spatial proxy index, followed by scaling by the range (i.e., the maximum value minus the minimum value). The order of the values ​​within the grid remains unchanged. The formula is as follows:

[0071]

[0072] Subsequently, different weights W are assigned based on the importance of each proxy metric. s The comprehensive calculation is performed using the following formula: N Combined =W Trans ×N Trans +W GDP ×N GDP +W Indus ×N Indus +W Pop ×NPop

[0073] The calculation obtained in combination with the scheme described in (1) Obtain the port-specific emission ratio factor with final 100-meter resolution The formula is shown below:

[0074]

[0075] The above methods can more accurately reflect the dynamic changes and spatial distribution characteristics of carbon emission proportion factors from different sources in the port area, providing more reliable data support for optimizing the compilation and management of emission inventories.

[0076] 4. Port CO2 emission verification and foundation correction

[0077] Using the port-specific emission ratio factor obtained through calculation and the SO2 emissions obtained and NO x Emissions Estimated CO2 emissions As shown in the formula below:

[0078]

[0079] (1) CO2 emissions indirectly estimated from two different species and Cross-validation was performed, retaining the overlapping grids within the emission results and model uncertainty ranges of both methods, and filtering out outliers with excessively large differences, before merging to obtain robust CO2 emission results.

[0080] (2) Short-term monitoring of the port area was conducted using the IRAGSON integrated eddy covariance flux system. This system integrates an open-circuit infrared CO2 analysis sensor and a three-dimensional ultrasonic anemometer, enabling simultaneous spatial and temporal measurement of CO2 concentration, three-dimensional wind speed, and ultrasonic temperature. This avoids high-frequency flux omissions caused by cross-spatial measurements and effectively improves measurement accuracy. The system records CO2 emission flux and concentration data at 30-minute sampling intervals, calculates the hourly average emission result, and correlates it with the satellite estimate in time and space. This monitoring value is used to perform short-term correction on the satellite CO2 estimate and calculate the correction coefficient. First, the CO2 emission indirectly estimated by the satellite is calculated within the monitoring time period t of the eddy covariance flux system in the port unit. Eddy covariance monitoring values The differences between these values ​​are used to derive a correction factor, Factor. This factor is then used to adjust the "top-down" CO2 emission estimates for the port area, and the correction factor from the monitoring period is extended to the CO2 emission results for the long-term series j. To improve its accuracy, the formula is as follows:

[0081]

[0082]

[0083] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A satellite remote sensing inversion method for port carbon dioxide emissions, characterized in that: Includes the following steps: (1) The hourly long-term series of nitrogen dioxide and sulfur dioxide column concentration data acquired by GEMS were divided into port area grids; port grid data with significant correlation were selected; and satellite data were optimized and reconstructed. (2) Based on the reconstructed port GEMS NO2 tropospheric column concentration grid and meteorological data, its spatial divergence and temporal gradient were calculated, and then substituted together with the concentration data into the mass conservation equation; NO2 obtained from CEMS monitoring in the port area was used as the basis for further calculation. x Total emissions data serve as a priori emission term in the mass conservation equation; (3) Solve the port emission characteristics or coefficients pixel by pixel and calculate their probability density distribution; obtain the confidence intervals of each coefficient; perform random repeated sampling within the confidence intervals and substitute the sampling coefficient results into the equation; combine the column concentration data observed by satellite and each derived term to calculate the port's NO. x Emissions sample; the mean of all samples is used as the port's hourly, grid-by-grid NO. x Emission results; (4) The SO2 vertical column concentration of the reconstructed port GEMS is processed in the same way as in steps (2) to (3); the mass conservation equation of SO2 is increased with a quadratic gradient term, and the SO2 emissions of the port are calculated hourly and grid-by-grid by obtaining the corresponding emission characteristic coefficients. (5) Annual anthropogenic SO2 and NO2 emissions based on bottom-up emission inventories x The CO2 emission data was interpolated to the selected port grid cells in GEMS to calculate SO2 and NO2 emissions separately. x The ratio factors of CO2 emissions to SO2 and NO2 are: x ; (6) Downscale the emission ratio factor, select proxy indicators, and assign different weights to each indicator; perform grid-by-grid processing on the emission ratio to obtain port-specific emission ratio factors. (7) Based on SO2, NO x The hourly high-resolution CO2 emissions of the port are estimated using emission data and port-specific emission ratio factors. The estimation results are compared and cross-validated. The validated robust CO2 emissions are then time-matched with short-term measured data of ground eddy covariance in the port area to obtain correction coefficients, which are then applied to correct the long-term series results.

2. The method for satellite remote sensing inversion of port carbon dioxide emissions according to claim 1, characterized in that: The specific method for step (1) is as follows: First, download the GEMS data product, including NO2 and SO2 column concentrations, and divide the data into grids containing the port based on the port's geographical location. Then, perform quality control on these grid data based on three indicators: solar zenith angle, reflectance, and cloud cover. Finally, compare and verify the data with TROPOMI data and Gaofen-5 data, and retain the hourly results of the GEMS grids that are strongly correlated with both, with a correlation coefficient greater than 0.9, to ensure its accuracy. Subsequently, the DINEOF method was used to optimize and reconstruct the missing data. Based on spatial empirical orthogonal function EOF decomposition and temporal principal component PC decomposition, the spatiotemporal domains with the greatest variation are identified. By using weighted EOFs and PCs in an iterative manner, the missing data points are reconstructed by weighting each orthogonal basis function; the optimal number of iterations that minimizes the cross-validation error is used to obtain the best reconstructed data.

3. The method for satellite remote sensing inversion of port carbon dioxide emissions according to claim 1, characterized in that: Step (2) involves constructing an hourly NO model based on a top-down inverse characterization method and the principle of atmospheric mass conservation. x Emission estimation equations were established to correlate port satellite NO2 concentrations with monitored NO2 levels. x The correlation of emissions and the quantification of their impact on port NO x The physical and chemical factors of emissions ultimately determine the NO emissions. x Emissions; Using the reconstructed GEMS NO2 column concentration data obtained through the DINEOF method, the rate of change of NO2 column concentration during two satellite passes was calculated, and the derivatives of the daily zonal and meridional fluxes were calculated as the daily NO2 concentrations. x Spatial divergence of horizontal flux; In addition, NO obtained using the CEMS system x Emission data serves as a priori emission term in the mass conservation equation; the CEMS system continuously and automatically monitors particulate matter concentration, gaseous pollutant concentration, and total pollutant emissions from stationary sources; NO in the CEMS system x The measurement principle is based on chemical analysis methods, which calculate NO by measuring the concentrations of NO and NO2. x The concentration.

4. The method for satellite remote sensing inversion of port carbon dioxide emissions according to claim 1, characterized in that: In step (3), based on the mass conservation model constructed in step (2), a multiple regression method is used to fit the time period containing effective CEMS monitoring data to the port's GEMS data on a daily, pixel-by-pixel basis to obtain the coefficients; these coefficients represent the port's NO x Emission characteristics: NO x / NO2 ratio, chemical lifetime, and transport distance; confidence intervals of coefficients are estimated daily using the Bootstrap method, with coefficients within the intervals considered reasonable; a set of coefficient matrices is generated daily, and the same set of coefficient matrices is used for all hourly emission models of that day; Finally, repeated random sampling is performed in the coefficient matrix and substituted into the equation. Combined with the hourly GEMS NO2 column concentration and related derivatives from step (2) for the long-term series, emission samples are obtained. The mean of all samples is used as the final NO emission value. x Estimated value.

5. The method for satellite remote sensing inversion of port carbon dioxide emissions according to claim 1, characterized in that: In step (4), the port data on sulfur dioxide (SO2) column concentration observed by GEMS are processed in the same way as steps (2) to (3); the SO2 emission data monitored by CEMS at the port is used as prior information, and NO... x Based on this, a quadratic gradient term for SO2 is introduced to obtain the corresponding emission characteristics, including chemical lifetime, transport distance, and diffusion rate. These characteristic coefficients are then substituted into the mass conservation equation to finally calculate the SO2 emissions.

6. The method for satellite remote sensing inversion of port carbon dioxide emissions according to claim 1, characterized in that: In step (5), based on the "bottom-up" emission inventory, SO2 and NO from anthropogenic sources are obtained. x The CO2 emission data was interpolated into the port GEMS observation grid selected in step (1) and then converted into emission fluxes. The port's CO2 emissions, along with the proxy species SO2 and NO, were calculated. x The emission ratio factor is used to reflect the level of fossil fuel combustion and emissions in the port area.

7. The method for satellite remote sensing inversion of port carbon dioxide emissions according to claim 1, characterized in that: In step (6), multiple emission proxy indicators are adopted, and the emission proportion factor obtained in step (5) is downscaled based on the normalization method. The port traffic density grid, GDP spatial distribution grid, sewage discharge sector density grid and population density distribution grid dataset are combined as four spatial proxy indicators. Within the grid range corresponding to the list resolution, the maximum-minimum normalization method is used for each list grid to center according to the minimum value of the spatial proxy parameter, and then scaled according to the range. The order of the values ​​remains unchanged, and different weights are assigned to obtain the port-specific emission proportion factor.

8. The method for satellite remote sensing inversion of port carbon dioxide emissions according to claim 1, characterized in that: In step (7), the port-specific emission ratio factor calculated in step (6) is used in conjunction with the port SO2 and NO obtained in steps (3) to (4). x Emissions, estimating CO2 emissions; the specific calculation process is as follows: First, through SO2 and NO respectively x CO2 emissions are solved indirectly and cross-validated within the error range; if the results are consistent within the error range, the robust emission grid results are retained; if outliers are found, they are removed. The IRAGSON integrated eddy covariance flux system was used for short-term monitoring of the port area. The system recorded CO2 emission flux and concentration data at a sampling interval of 30 minutes. The following method was used to correct the satellite CO2 estimate: First, the difference between the satellite-indirectly estimated CO2 emission and the eddy covariance monitoring value was calculated during the monitoring period of the eddy covariance flux system in the port unit, so as to obtain the correction coefficient. The coefficient was then used to adjust the "top-down" CO2 emission estimates for the port area, and the correction coefficient during the monitoring period was extended to the CO2 emission results for the long-term series.

Images