The sensitivities of ozone and PM2.5 concentrations to the satellite-derived leaf area index over East Asia and its neighboring seas in the WRF-CMAQ modeling system

Abstract

Vegetation plays an important role as both a sink of air pollutants via dry deposition and a source of biogenic VOC (BVOC) emissions which often provide the precursors of air pollutants. To identify the vegetation-driven offset between the deposition and formation of air pollutants, this study examines the responses of ozone and PM 2.5 concentrations to changes in the leaf area index (LAI) over East Asia and its neighboring seas, using up-to-date satellite-derived LAI and green vegetation fraction (GVF) products. Two LAI scenarios that examine (1) table-prescribed LAI and GVF from 1992 to 1993 AVHRR and 2001 MODIS products and (2) reprocessed 2019 MODIS LAI and 2019 VIIRS GVF products were used in WRF-CMAQ modeling to simulate ozone and PM 2.5 concentrations for June 2019. The use of up-to-date LAI and GVF products resulted in monthly mean LAI differences ranging from −56.20% to 96.81% over the study domain. The increase in LAI resulted in the differences in hourly mean ozone and PM 2.5 concentrations over inland areas ranging from 0.27 ppbV to −7.17 ppbV and 0.89 μg/m 3 to −2.65 μg/m 3 , and the differences of those over the adjacent sea surface ranging from 0.69 ppbV to −2.86 ppbV and 3.41 μg/m 3 to −7.47 μg/m 3 . The decreases in inland ozone and PM 2.5 concentrations were mainly the results of dry deposition accelerated by increases in LAI, which outweighed the ozone and PM 2.5 formations via BVOC-driven chemistry. Some inland regions showed further decreases in PM 2.5 concentrations due to reduced reactions of PM 2.5 precursors with hydroxyl radicals depleted by BVOCs. The reductions in sea surface ozone and PM 2.5 concentrations were accompanied by the reductions in those in upwind inland regions, which led to less ozone and PM 2.5 inflows. The results suggest the importance of the selective use of vegetation parameters for air quality modeling.