| Abstract | Air pollution has become an ever-growing serious issue worldwide due to the heavy burden of local emission loads from both anthropogenic and natural sources. Southeast Asia (SEA) is a region of large emissions from key anthropogenic sources including crop residue open burning, residential combustion, transportation, energy generation and industrial activities. However, the present lack of systematic emission inventory (EI) databases for the region is seen as a major constraint for effective air quality management. The fragmented EI data available for the region still contain uncertainties at magnitudes for different sources which are an obstacle to the promulgation of efficient clean air action plans in the region. This study aimed to develop a comprehensive EI dataset covering major emission sources in 11 SEA countries during 2010 – 2018 with QA/QC that is readily uploaded to share on cloud based system for management. As a way of experiment, this study also applied the inverse modeling method to generate posterior estimates to compare with the proven priori emissions produced by the EI task for a selected urban scale domain in SEA. The total annual emissions averaged over the period 2010 – 2018 from the major emission sources were 3,130 SOx; 9,486 NOx; 13,449 NMVOC; 7,657 NH3; 6,374 PM2.5; 484 BC; 2,707 OC; 27,924 CH4, and 710 N2O in Gg/yr; and 118 CO and 3134 CO2 in Tg/yr. Indonesia was the largest emitter of all pollutants, contributing from 39.1% of BC to 53.3% of OC of the SEA total, followed by the group of Thailand, Vietnam, and Philippines which collectively contributed about 20.5% - 47.7%. On-road transport was the major source of NMVOC, NOx, and BC, i.e., 66%, 59% and 23%, respectively, of the total in SEA. Forest fires contributed to 57% of OC, 47% of PM2.5, and 33% of total CO in SEA. Industrial activities and thermal power plants were the major sources of SOx emissions, collectively contributed 76% of the SEA total. Livestock farming contributed 15% of CH4 and 37% of NH3 emissions, while fertilizer application contributed about 47% of NH3 emissions. Residential combustion also contributed substantially, i.e., from 5% to 14%, to the SEA total of N2O, BC, OC, NMVOC, CO2 and CO emissions, respectively. Crop residue open burning, mainly rice straw, contributed about 22 – 27% of each PM2.5, BC, and OC emissions. The EI data for SEA region were uploaded to the Google Cloud Studio for storage which enables further updating and analysis. The data were linked from the cloud storage to the analysis platform provided by Google Cloud Studio for the visualization using tabulated and graphic tools. Lastly, the data can be shared to any third party or to the public by using the available report (dashboard) and data links. Therefore, a reliable EI database developed for SEA with necessary QA/QC can be managed on cloud for further supporting air quality management. The inverse modeling (WRF/CAMx) was done for the domain of Bangkok Metropolitan Region (BMR), Thailand for CO and BC pollutants using ground-based measurement data. WRF model produced meteorological input data for three years, 2016-2018. The model performance was evaluated based on the simulation results of 2017-2018 for temperature, relative humidity (RH), wind speed and directions collected at two international airports in BMR domain. The statistical performance analysis showed better simulation results, for temperature and RH than wind speed, whereas the wind direction was simulated with the lowest scores with only 60% of MB and 33% of MAGE values meeting the statistical criteria. Comparison of the observed synoptic and upper wind charts with the modeled surface pressure and upper wind showed consistency at the selected hours. The WRF/CAMx model was run for 3 years of 2016 - 2018 for one base case simulation and two simulations using inverse method with limited number of observational data in urban scale. The preliminary results showed that the inverse modeling simulation increase the consistency between observed and modeled CO and BC considering the time series and scatter plot analysis. Also, the difference between observed and modeled CO and BC in the mean and median values were less than 20% for most of the months. The total CO posterior emissions using observational data constraints were about 2.6-2.9 times of the total CO priori emissions in the domain in the 3 years. For BC, the total annual posterior emissions were about 2.8 and 3.2 times of the BC priori emissions in 2016 and 2017, respectively. This trial experiment showed that the application of inverse modeling approach produced posterior EI which was not comparable with the proven priori EI based on limited number of observations. The BMR domain had sparsely available monitoring data for BC with only 2 stations for weekly data, and for CO, although with more stations, the low CO values (<100 ppbv) were not properly reported. This has reduced overall accuracy of the produced posterior results. Nevertheless, the inverse method helped to refine the monthly emissions which showed clear fluctuations among the months, especially between the dry and rainy seasons which is useful input for further air quality modeling purposes as well as pointed out the uncertainties in priori emissions from open biomass burning during dry seasons in the domain. Future studies are recommended to collect more monitoring data which are adequately available in terms of both spatial and temporal distributions to test with alternative inverse methods, good spatiotemporal priori EI, and consider uncertainties in model parameters and inverse method to produce comparable posterior EI. The comprehensive EI dataset covered major emission sources in 11 SEA countries during 2010 – 2018 produced by this study can be used by authorities for policy and regulation promulgation, implementation of clean air action plan, and for proposal of different mitigation and adaption measures. In addition, the EI dataset can be used for academic research purposes and integrated with other global/continental/regional databases for international cooperation. The data should be systematically updated and managed with QA/QC which can be done using the cloud system. This study recommends that the mega data such as EI database can be stored on Google Cloud studio and other platforms to avoid data lost. The EI data can be retrieved from the data source on cloud, and finally publicized. Future studies are recommended to include data analysis and EI calculation on the cloud platform. The inverse modeling using ground-based observation data constraints and Chemical transport model (CTM) provides a promising way to produce posterior emissions and evaluate with the priori EI developed for a small domain of BMR. Besides, from the findings of this study, the approach revealed the possible bias in spatio-temporal distribution of priori EI. Uncertainties, limitations, and challenges exist in the inverse modeling, including reliable observational data with adequate spatial and temporal coverage for the domain, adequate temporal and spatial distribution priori EI, and the inverse method. The detailed knowledge of the domain characteristics and the representativeness of the measurement sites for the domain are important to produce accurate posterior emissions for a local small domain. Therefore, future studies are recommended to take into consideration these challenges and carefully analyze the uncertainties in inverse modeling approach to evaluate emission estimations. |