Assessment of the Statistical Performance of Chemical Transport Model Studies in India

India’s public health is seriously endangered by air pollution and linked to ∼0.8–0.98 million and 0.17 million premature deaths annually from chronic exposure to particulate matter (PM2.5) and ozone, respectively. Air quality forecasting tools are needed to assess the spatiotemporal variability of air pollutants and evaluate the impact of potential mitigation measures. Gridded concentrations of PM2.5 and ozone are typically derived from first principle models called Chemical Transport Models (CTMs). Widespread application of CTMs, particularly for regulatory purposes, requires an understanding of model performance in comparison with observation data. A landscape approach that systematically reviews CTM model predictions and proposes benchmarks for assessing model performance can offer a way forward to build confidence in CTM-backed forecasting tools. Following similar efforts in the United States and China, we systematically shortlisted and examined model performance outcomes for 46 CTM studies reporting PM2.5 and ozone predictions in India between 2008 and 2023. The reported statistical metrics for each pollutant (normalized mean bias (NMB), normalized mean error (NME), coefficient of correlation (r), and index of agreement (IOA) were rank ordered to identify two types of benchmarks: “goals” (highest achievable model accuracy based on top one-third performing studies) and “criteria” (typical model accuracy based on the top two-third performing studies). The identified goal performance for PM2.5 was 17% (NMB), 34% (NME), and 0.67 (r), and ozone was 14% (NMB), 43% (NME), and 0.89 (r). These benchmarks are less restrictive than those reported for the United States and China. This highlights the need for the CTM community to coalesce around a common set of practices in evaluating CTMs. Additionally, complementary efforts in developing representative emissions inventories and including air pollution data from the expanding observational networks are required to update such benchmarks periodically. This study seeks to advance the capabilities of CTMs as a cornerstone of air quality forecasting tools and augment air quality management frameworks in India.