A Radar‐Based Stochastic Model for Short‐Time‐Increment Rainfall

Abstract

In order to model the rainfall depths at time scales smaller than or equal to 1 day a radar‐based conceptualization of ground rainfall is developed. This conceptualization states that rainfall in time at any ground location can be described by the rainfall intensity distribution over the trajectories which are drawn by the ground location point on the rain fields that pass over this location. As such, this conceptualization uses the information on the rain fields that are detected on the weather radar plan position indicator scope. Using this conceptualization, a single‐site stochastic model is developed in order to describe the evolution of rainfall depths at all time scales smaller than or equal to 1 day. The model has the structure of a filtered Poisson cluster process. A statistical application of the model to the simultaneous radar and rain gage data is performed. The model is calibrated mainly by a nonparametric procedure using 5‐min‐increment radar microfilm and hourly rain gage data during April. The calibrated model is used first in estimating the rain‐gage‐observed hourly and daily mean rainfall depths during a period which envelops the calibration period plus an extra year's test period during April. The calibrated model is then used in the real‐time radar‐based estimation of hourly rainfall depths during the test period. The results of this application show that, in general, the radar‐based conceptualization developed in this study is a viable approach to the modeling of rainfall depths at time increments smaller than or equal to 1 day.