Towards an estimate of on-road emission inventory in a low-and-middle income setting: data, methods, and challenges

Abstract

<p><strong>Background</strong></p> <p>Low-and-middle income countries (LMICs) often have the highest levels of air pollution. At the same time, they suffer from a lack of data to inform emission inventories that can be employed for developing cost-effective policies. Estimates provided by international research groups (e.g. EDGAR) may be useful in high-income countries or at the regional scale, but their spatial resolution is too coarse to represent within-city variation. There is an urgent need to develop methods and identify data sources that can be readily applied across LMICs to generate emission inventories.</p> <p><strong>Objective</strong></p> <p>We aim to estimate traffic characteristics for Delhi, India (2021 pop: ~21 million) to parameterise a bottom-up traffic emissions inventory model. While many inventories have been published for this setting in the past, there are limitations that remain to be addressed and emerging datasets that may further improve estimates. Our focus is to identify new data sources, highlight their strengths and limitations, and develop methods that can be replicated across many cities in India, and potentially in other LMICs.</p> <p><strong>Data and method</strong></p> <p>We used the following datasets— records of the emission testing program for Delhi (six years) and all-India (one year) with measured emissions along with vehicle specifications (n=5 million tests/year), 15-year individual-level vehicle registration data (n=5.5 million), and model specification data for cars (n=8,000) and motorcycles (n=300) scraped from online sources. We conducted on-road observational surveys to record license plate numbers of randomly selected vehicles and looked up their specifications in a government online portal. Through data fusion, we enriched these datasets and corrected them for biases. For the total fleet, stratified by vehicle types, we estimated longitudinal trends in age distribution, fuel-use distribution, and idling emissions from vehicle testing. With repeated measurements of emissions for the same cohort of vehicle population (e.g., 2010 models, diesel cars), we used a panel data approach to estimate age-dependent emission deterioration rates.</p> <p><strong>Results</strong></p> <p>We found self-selection bias in the vehicle testing database (older vehicles less likely to be tested) through a comparison with the on-road observational surveys. We generated weights to correct this bias for the final analysis. Using the vehicle testing database and on-road survey, we found that a significant share of cars (16-40%) operating in Delhi are registered outside the city. Country-wide vehicle testing data indicates that car retirement policy in Delhi (diesel: 10 years, and petrol: 15 years) may be leading to the migration of older car fleet to other states. The yearly trend of average emission measurements clearly reflects the changes in emission standards implemented over the past decade.</p> <p><strong>Conclusion</strong></p> <p>Our results show that the secondary data sources, though highly useful because of their size and coverage, have biases and limitations. However, these problems can be overcome through integration with other datasets and low-cost, easily replicable primary surveys. The talk will discuss the policy implications of the results, and the next steps toward developing an emissions inventory.</p>