Quantification of the spatial variability of rainfall based on a dense network of rain gauges

The spatial variability of rainfall within a single Local Area Weather Radar (LAWR) pixel of 500 × 500 m is quantified based on data from two locations. The work was motivated by the need to quantify the variability on this scale in order to provide an estimate of the uncertainty of using a single rain gauge for calibrating the LAWR. A total of nine rain gauges were used, each representing one-ninth of the 500 × 500 m area. The analysis was carried out based on a dataset obtained using tipping bucket gauges during the summer and fall of 2007 and 2008, and the results were compared with results from an earlier campaign in 2003. The fact that the 2007–2008 dataset was almost four times larger than the original dataset from 2003 motivated this extended study. Two methods were used to describe the variability: the coefficient of variation and the spatial correlation structure of the rainfall field. Despite the small area of 0.25 km2, accumulated rainfall was found to vary significantly within individual events with durations ranging from 5 min to 13 h. The coefficient of variation was found to range from 1–26% in the 2007–2008 dataset and in some special cases even higher. The 95% prediction interval for a given rainfall depth is estimated and can be used to address the uncertainty of using a single rain gauge to represent the rainfall within a 500 × 500 m area.