A Precipitation Model and Its Use in Real-time River Flow Forecasting

Abstract

A one-dimensional, physically based, station precipitation model is proposed and tested. The model state variable is the liquid water equivalent mass in a unit area cloud column. Model inputs are the air temperature, dew-point temperature, and pressure at the ground surface. The precipitation rate at the ground surface is the model output. Simplified cloud microphysics give expressions for the moisture input and output rates in and from the unit area column. Parameterization of the model physical quantities: updraft velocity, cloud top pressure, and average layer cloud-particle diameter is proposed, so that parameters, will remain reasonably constant for different storms. Conceptual soil and channel routing models were used together with the proposed precipitation model in formulating a general Rainfall-Runoff model. Hourly data from eleven storms of different types and from two different locations in the US, served as the data-base for the station precipitation model tests. Performance in predicting the hourly precipitation rate was good, particularly when a sequential state estimator was used with the model. The general Rainfall-Runoff model formulated, complemented by a sequential state estimator, was used with six-hourly hydrological data from the Bird Creek basin, Oklahoma, and with six-hourly meteorological data from the somewhat distant Tulsa, Oklahoma, site. Forecasts of both the mean areal precipitation rate and the basin outflow discharge were obtained. Performance indicated the value of the precipitation model in the real-time river flow forecasting.