An Evaluation of the Statistics of Rainfall Extremes in Rain Gauge Observations, and Satellite-Based and Reanalysis Products Using Universal Multifractals

Abstract

Abstract Confidence in the estimation of variations in the frequency of extreme events, and specifically extreme precipitation, in response to climate variability and change is key to the development of adaptation strategies. One challenge to establishing a statistical baseline of rainfall extremes is the disparity among the types of datasets (observations versus model simulations) and their specific spatial and temporal resolutions. In this context, a multifractal framework was applied to three distinct types of rainfall data to assess the statistical differences among time series corresponding to individual rain gauge measurements alone—National Climatic Data Center (NCDC), model-based reanalysis [North America Regional Reanalysis (NARR) grid points], and satellite-based precipitation products [Global Precipitation Climatology Project (GPCP) pixels]—for the western United States (west of 105°W). Multifractal analysis provides general objective metrics that are especially adept at describing the statistics of extremes of time series. This study shows that, as expected, multifractal parameters estimated from the NCDC rain gauge dataset map the geography of known hydrometeorological phenomena in the major climatic regions, including the strong orographic gradients from west to east; whereas the NARR parameters reproduce the spatial patterns of NCDC parameters, but the frequency of large rainfall events, the magnitude of maximum rainfall, and the mean intermittency are underestimated. That is, the statistics of the NARR climatology suggest milder extremes than those derived from rain gauge measurements. The spatial distributions of GPCP parameters closely match the NCDC parameters over arid and semiarid regions (i.e., the Southwest), but there are large discrepancies in all parameters in the midlatitudes above 40°N because of reduced sampling. This study provides an alternative independent backdrop to benchmark the use of reanalysis products and satellite datasets to assess the effect of climate change on extreme rainfall.