Considering diversity in precipitation variability when updating seasonal flow forecasts

The Israel Hydrological Service forecasts the flow to Lake Kinneret by use of a hydrometeorological model which relates seasonal flow volumes to predi cted annual depths of precipitation at representative stations. Issued fore casts include an array of 10 values assigned with the exceedance probabilit ies of 5-95%, as well as their average, which is an estimate of the expecte d value. A naive forecast is issued at the beginning of the precipitation s eason, and subsequent updating forecasts are issued during the season. Pred icted annual depths of precipitation in an updating forecast are composed o f the observed depth until the updating date and an array of predicted dept hs for the subsequent sub-season. No explicit relation was found between co rresponding prior and subsequent depths concerning any updating date. But, the distribution of subsequent depths was found related to the category (i. e. low, medium, or high) of the corresponding prior depth. Empirical, param etric and nonparametric tests revealed a number of significant diversities in the distributions of later depths with respect to the corresponding cate gories of earlier depths. This diversity could result in considerable diffe rences in the forecasted flow volumes, and is particularly important in cas es of extreme situations, when management decisions are tied to revisions b efore the end of the precipitation season. (C) 2001 Elsevier Science B.V. A ll rights reserved.