A stochastic model relating rainfall intensity to raindrop processes

James A. Smith, Richard D. De Veaux

Research output: Contribution to journalArticlepeer-review

17 Scopus citations

Abstract

The temporal variability of rainfall and raindrop processes is examined at time scales ranging from less than 1 min to 1 hour. Raindrop processes are represented in terms of drop arrival rate, mean diameter, and coefficient of variation of drop diameter and modeled as time‐varying stochastic processes. It is shown that rainfall rate and accumulated rainfall have simple and accurate representations in terms of raindrop processes. Using these results the temporal variability of rainfall rate is examined in terms of temporal variability of raindrop processes. It is shown that the temporal variability of rainfall rate varies systematically across a range of climatic settings and, more importantly, that these climatic contrasts in rainfall rate can be related to contrasting properties of raindrop processes. Two statistical models of rainfall rate and raindrop processes are examined in detail: a lognormal model with fixed parameters and a lognormal model with parameters that vary from storm to storm. Temporal correlation structure of rainfall rate exhibits qualitatively different behaviour under the two models. Lognormal models of rainfall rate are extended to models in which dependence on the averaging time interval is explicitly represented. Scale properties of rainfall rate are examined empirically for averaging time intervals ranging from 1 min to 30 min. Empirical analyses are based on drop‐size data from North Carolina, New Jersey, Oregon, Alaska, and the Marshall Islands.

Original languageEnglish (US)
Pages (from-to)651-664
Number of pages14
JournalWater Resources Research
Volume30
Issue number3
DOIs
StatePublished - Mar 1994

All Science Journal Classification (ASJC) codes

  • Water Science and Technology

Fingerprint Dive into the research topics of 'A stochastic model relating rainfall intensity to raindrop processes'. Together they form a unique fingerprint.

Cite this