A COMPUTATIONALLY EFFICIENT METHOD FOR THE CHARACTERIZATION OF SUB-GRID-SCALE PRECIPITATION VARIABILITY FOR SULFUR WET REMOVAL ESTIMATES

The Regional Lagrangian Model of Air Pollution (RELMAP) is a mass-conserving, Lagrangian puff model that simulates the concentration and deposition of sulfur dioxide (SO2) and sulfate (SO4(2-)) over the eastern United States of America and southeastern Canada. In 1986, a model evaluation showed that the RELMAP overestimated total sulfur wet deposition for the warm seasons by 25-35%, partly due to sub-grid-scale variability in the observed precipitation fields. Refinement of the model grid to fully resolve the precipitation observation network would increase the computational requirements of the model to unacceptable levels. Instead, a new parameterization for the wet removal fraction was implemented in hopes of reducing the model's sensitivity to precipitation variability. The effects of observed sub-grid-scale precipitation variability on SO2 and SO4(2-) wet removal in the original and updated RELMAP have been isolated and analysed. Spatial averaging of precipitation to the length scales of the RELMAP grid (approximately 100 km) is shown to increase SO4(2-) wet removal fractions by as much as 400% over those obtained from individual observations. A method of Categorized Event Distribution (CED) analysis has been developed to characterize sub-grid-scale variability so that more accurate estimates of wet deposition may be made while preserving the ability of the RELMAP to be intensively applied on modest computing facilities. The use of CED analysis of precipitation in the current version of RELMAP is shown to systematically reduce monthly domain total sulfur wet deposition estimates by 6-12%, demonstrating that sub-grid-scale precipitation variability must still be addressed. CED analysis offers the numerical modeler the ability to characterize sub-grid-scale variabilities that might otherwise preclude the use of scientifically valid relationships between observed phenomena.