DEVEX-disdrometer evaluation experiment: Basic results and implications for hydrologic studies

Abstract

Increasing our understanding of the small scale variability of drop size distributions (DSD), and therefore of several bulk characteristics of rainfall processes, has major implications for our interpretation of the remote sensing based estimates of precipitation and its uncertainty. During the spring and summer of 2002 the authors conducted the DEVEX experiment (disdrometer evaluation experiment) to compare measurements of natural rain made with three different types of disdrometers collocated at the Iowa City Municipal Airport in Iowa City, Iowa in the Midwestern United States. This paper focuses on the evaluation of the instruments rather than analysis of the hydrometeorological aspects of the observed events. The comparison demonstrates discrepancies between instruments. The authors discuss the systematic and random effects in terms of rainfall quantities, drop size distribution properties, and the observed drop size vs. velocity relationships. Since the instruments were collocated, the effects of the natural variability of rain are reduced some with time integration, isolating the instrumental differences. The authors discuss the status of DSD measurement technologies and the implications for a range of hydrologic applications from remote sensing of rainfall to atmospheric deposition to soil erosion and sediment transport in the environment. The data set collected during the DEVEX experiment is made available to the research community.

Introduction

Despite the technological advances, measurement of rain drop size distribution (DSD) remains a difficult challenge. While the general understanding of the physical mechanisms leading to drop formation and their effect on size, shape, and velocity range is in place, details of space and time variability require observational means that are still elusive. The space/time information on DSD, which determines other variables such as radar-reflectivity, optical extinction, kinetic energy, and of course rainfall-rate, is needed for interpretation of radar and satellite remote sensing, prediction of hydrologic processes, and agricultural considerations.

Until recently, there have been only in situ means of measuring DSD using mechanical and optical devices called disdrometers. Advancements in radar polarimetry provide opportunities for estimating DSD over large areas, e.g. [3], [4]. These estimates are based on some assumptions regarding drop shape and thus need to be independently assessed using disdrometer observations. Discussion of radar polarimetric methods is outside of the scope of this paper.

There is no widely accepted standard for in situ measurement of DSD. A mechanical device (JW) developed by Joss and Waldvogel [7], [14] seems to be the most popular instrument but its technology is dated and it has well-recognized limitations [5], [23], [27]. Most of the new instruments use optical principles to measure DSD, e.g., [1], [17], [18], [22], but there are also other technologies [24].

Our objective for this paper is to report on a field experiment evaluation of four (see Appendix for additional explanation) disdrometers we conducted in Iowa in the summer of 2002. The experiment, which we call DEVEX, was an outgrowth of our earlier effort reported by Miriovsky et al. [19], in which we attempted to learn about the spatial variability of DSD at the scale relevant to radar-rainfall estimation, i.e., about 1 km². In that effort we used a collection of several different instruments, but at the end we concluded that we could not separate the natural from instrumental variabilities. We decided that the first step in our quest for learning how the DSD varies in space has to be an intercomparison of collocated instruments to improve our understanding of the instrumental effects. We gathered several new instruments either by purchasing them or by inviting their developers to participate in the effort.

In this paper we describe our experimental setup, discuss the deployed instruments, present initial intercomparison results, justify adjustments to our algorithms, and show some basic results of data analysis. The main question we asked was: Will the data agree well given that the instruments are essentially collocated? And if they do not: will we be able to explain the observed differences? Our hypothesis was that the results should intercompare very well as we took great care of preparing the instruments for this experiment. We have used the two-dimensional video disdrometer (2DVD) instrument in several field experiments and are very familiar with its workings [17]; the PARSIVEL [18] was brand new, out of the box and we carefully followed the instructions for use by its manufacturer PMTech AG of Germany; and developers of the dual beam spectropluviometer (DBS) from the Centre d’Etude des Environnements Terrestre et Planetaire (CETP), Paris, France, set it up in DEVEX. We also used a JW disdrometer, which returned from the factory repair only few months earlier. However, a post-experiment analysis of the JW data cast question on its operation and thus we excluded its results from the intercomparison. We briefly elaborate in the Appendix.

We organized this paper as follows. First, in Section 2, we describe the experimental site and briefly discuss the instruments we evaluate. Then, in Section 3, we present the results taking the data from each instrument without any adjustments. This identifies several discrepancies that we try to explain. In Section 4 we describe corrective procedures for the suspected causes of the discrepancies. We correct the data and present updated intercomparison results. We close with a discussion of implications of our results for future investigations of the DSD observations.