Evaluation of Precipitation Regional Characteristics in NARCCAP Simulations over a Coastal Region

The potential use of high-resolution climate simulations for coastal ecosystem assessment and impact studies has been highlighted in several recent studies by the scientific and operational communities. An important question is to assess whether such simulations preserve the spatio-temporal structures of rainfall fields on a regional scale, and whether they can provide scientifically sound and physically consistent representations of future rainfall projections that account for expected climate change. In the current study, simulations of current-climate precipitation simulations produced by the NARCCAP program are assessed over the Gulf of Mexico central coastal zone in the US. We performed an analysis of the daily, monthly, seasonal and annual variability of precipitation produced by the different NARCCAP models. The assessment is performed using various statistics such as correlation, bias, mean absolute error, mean absolute percentage error, and root mean square error. A measure of variability simulated by each model is computed as the ratio of its standard deviation, in both space and time, to the corresponding standard deviation of observation. The statistics are stratified based on different climate divisions in the region. The analysis also examines some of the important underlying precipitation characteristics in the region such as diurnal and seasonal cycles, and spatial consistency and correlation fields.