A GIS based estimate of mercury emission from natural sources: Case study in China and USA

Semi-mechanistic and regression-based models are developed to estimate mercury emissions from natural sources in China and some parts of Asia (study domain). Most of the globally developed mercury emission inventories give detailed accounts of emissions from anthropogenic sources and neglect a large contribution from naturally occurring emissions. Thirty two categories of land cover data is incorporated to estimate mercury emissions from natural sources over a 36 km Lambert Conformal grid covering the study domain. In regression-based estimates the surface temperature and cloud corrected solar radiation are retrieved from Mesoscale meteorological models (MM5) and used for calculating diurnal variations. Emission factors used are evaluated from measured mercury flux data for selected tree species, water, ice, and desert. In the semi-mechanistic approach the net air-surface exchange of mercury for land and water was calculated. Emissions over land are modeled as a function of land cover, evapotranspiration, and temperature. Emissions over water are modeled as a function of concentration gradient and temperature. Four months' (January, April, July, and October) simulations using 2001 MM5 data provided by the University of Tennessee were performed to investigate the seasonal variation. The modeled natural source mercury emissions are mostly contributed from the northeastern and the southern part of the domain, which is mainly covered with coniferous, deciduous, and evergreen forests. In the summer season the emissions from natural sources are about 50% of the anthropogenic emissions of the domain. However, emissions decrease considerably in the winter season. From our analysis we estimate that mercury emissions from natural sources in the study domain are about 250 Mg/yr, which is almost six times higher than the emissions seen in the CONUS domain. Additional model estimates over domains with finer resolution can be performed.