Precipitation research, 1975–1978

Abstract

Precipitation research in the United States during the past quadrennium, 1975–1978, has been directed at all three steps of a scientific discipline: description, explanation, and prediction. Most of the work, however, has been in describing, in various ways, the occurrence of precipitation: instruments and measurements, stochastic and analytic models, and accounts of extremes of precipitation or lack thereof: deep snows, cloudbursts and hurricane rains causing floods, unusually wet periods and droughts. These observed precipitation patterns were explained, somewhat, in a few papers as arising from variations in the inputs to general circulation models (changes in atmospheric dust or in sea surface temperatures) as well as from local causes: changes due to urbanization, irrigation, or industry, and to the apparent effects of weather modification experiments and operations. Prediction has made considerable progress by using multivariate regressions (“Model Output Statistics”) to relate the probability of precipitation (“PoP"), and even its amount in quantitative precipitation forecasts ("QPF”), to the various predictions, 12, 24, and 36 hours in advance, of operational circulation models (PE, LFM, etc.). Each of these aspects of precipitation research will be discussed in the order given.

Quantitative Precipitation Forecast Techniques for Use in Hydrologic Forecasting

“…This reveals another important difficulty associated with the currently available QPFs, namely, the fact that the operational QPF procedures cannot be directly coupled to the hydrologic forecast procedures, in a conceptually appealing and efficient hydrometeorological system operating as a whole. In a review of the current status of precipitation research in hydrology, Court (1979) comments that the need is evident for a generalized rainfall-runoff system.…”

A Point Rainfall Generator With Internal Storm Structure

“…A point rainfall generator is a probability model for the time series of rainfall as observed in one geographical point. Several rainfall models have been formulated in the literature (see, for example, Todorovic and Woolhiser [1974], Kavvas and Delleur [1975], and Court [1979] for a review]. In building the present model, a method used by many others (see, for example, Grace and Eagleson [1967], Eli and Croley [1975], or Arnorocho and Wu [1977]) was followed; the observed storm characteristics, such as dry time before the storm, total rainfall, and duration of the storm, were considered as realizations of a multidimensional random variable, to which a probability distribution was fitted.…”