Sampling Errors in the Measurement of Rainfall

Abstract

Two prototype impact disdrometers have been developed and deployed to measure the raindrop size distribution at several different locations in New Zealand. These disdrometers potentially offer significant improvements over existing designs: they are much cheaper, more durable, and more portable. The data collected has been used to determine the sampling errors in the measurement of the rainfall rate and to investigate the micro-physical properties of several different types of rainfall. A new theoretical framework has been developed for the sampling errors in rainfall measurement. Based on moment theory, it allows explicitly for non-normal, skewed distributions of rainfall sampling errors. The disdrometer data revealed such distributions for short duration rainrates. These data have been used to verify the theoretical calculations over a large range of rainrates and sampling times. The lognormal distribution has been found to be a good approximation to the actual sampling error distribution in many situations. The sampling error theory has been used to estimate the number of raindrops that need to be sampled to allow detection of a step change in raindrop size distribution parameters, and to define error limits for analyses. Differences in the micro-physical properties of rainfall for stratiform and convective rainfall have been examined in terms of the raindrop size distribution. The proportion of the variance in the rainfall rate attributable to changes in the raindrop arrival rate or mean raindrop size was found to be different for stratiform or convective rainfall. Similar proportions were found at three different geographic locations. Clear differences were also observed in the parameters of the raindrop size distribution at some of these sites, though these may have been caused by the structure of the convective rainfall rather than by differences in the raindrop formation processes.