Optimizing model performance: variable size resolution in cloud chemistry modeling

Under many conditions size-resolved aqueous-phase chemistry models predict higher sulfate production rates than comparable bulk aqueous-phase models. However, there are special circumstances under which bulk and size-resolved models offer similar predictions. These special conditions include alkaline conditions (when there is a high ammonia to nitric acid ratio or a large amount of alkaline dust) or conditions under which the initial H2O2 concentration exceeds that of SO2. Given that bulk models are less computationally-intensive than corresponding size-resolved models, a model equipped to combine the accuracy of the size-resolved code with the efficiency of the bulk method is proposed in this work. Bulk and two-section size-resolved approaches are combined into a single variable size-resolution model (VSRM) in an effort to combine both accuracy and computational speed. Depending on initial system conditions, bulk or size-resolved calculations are executed based on a set of semi-empirical rules. These rules were generated based on our understanding of the system and from the results of many model simulations for a range of input conditions. For the conditions examined here, on average, the VSRM sulfate predictions are within 3% of a six-section size-resolved model, but the VSRM is fifteen times faster.