7th webinar
Who?  Dr. Pablo Lichtig (Comisión Nacional de Energía Atómica)
When? 30 April 2026, 13hs UTC (10:00 Argentina, 15:00 France)
Where? Online at Visio (link)

Abstract

Air quality remains a critical environmental issue in South America, driven by rapid urbanization, anthropogenic activities, and biomass burning. However, limited computational capabilities currently make a high-resolution forecasting system unfeasible at the continental scale. This study presents an intercomparison and evaluation of five atmospheric chemistry-transport models (MUSICAv0, two implementations of WRF-Chem, EMEP/MSC-W, and CHIMERE) run at a medium resolution (~27 km) for South America for January, July, and September 2019, as part of a second stage of the Prediction of Air Pollution in Latin America and the Caribbean (PAPILA) project.

For all models, anthropogenic emissions were adopted from the global emissions inventory CAMS v5.3., while other inputs vary between models. We evaluate simulated concentrations against in situ observations of CO, NO, NO₂, O₃, PM10, and PM2.5 in four major megacities (Bogotá, São Paulo, Santiago, and Buenos Aires) and assess regional performance using satellite retrievals from MOPITT (CO), TROPOMI (NO₂), and MODIS (AOD).

Comparisons with satellite data reveal large uncertainties in biomass burning emissions, resulting in strong regional biases during the Amazon fire season (September). These biases, combined with the varying vertical sensitivities and specific overpass times of different satellites, mean that physically representing relevant processes (such as fire plume rise and diurnal emission cycles) can paradoxically appear to degrade model performance.

We find that models at an intermediate resolution can be effective at a regional and sometimes local scale. Model performance varies significantly depending on the statistical metric used, the region, the pollutant, and the time period, with no single «best» model emerging across all categories. Consequently, it is essential to reach a consensus on recommended model evaluation methods tailored to specific use cases, which are briefly discussed.

Bio
Pablo Lichtig is a research scientist in the Atmospheric Chemistry Division of the Argentine National Commission for Atomic Energy (CNEA). He has worked with filter observations and chemical characterizations of fine particulate matter (PM2.5) and modeling, both regional (WRF-Chem) and global with regional refinement (MUSICAv0), with special focus on carbon monoxide as a way to evaluate the origins of pollution in South America. He is a core developer of MELODIES-MONET, a model evaluation tool using observations, to which he added modules for evaluation with satellite products such as TEMPO and TROPOMI, and the ground-based remote sensing PANDORA network.