Premium
Stochastic Model of Seasonal Runoff Forecasts
Author(s) -
Krzysztofowicz Roman,
Watada Leslie M.
Publication year - 1986
Publication title -
water resources research
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.863
H-Index - 217
eISSN - 1944-7973
pISSN - 0043-1397
DOI - 10.1029/wr022i003p00296
Subject(s) - surface runoff , snowmelt , bayesian probability , categorical variable , environmental science , meteorology , econometrics , statistics , computer science , climatology , mathematics , geography , geology , ecology , snow , biology
Each year the National Weather Service and the Soil Conservation Service issue a monthly sequence of five (or six) categorical forecasts of the seasonal snowmelt runoff volume. To describe uncertainties in these forecasts for the purposes of optimal decision making, a stochastic model is formulated. It is a discrete‐time, finite, continuous‐space, nonstationary Markov process. Posterior densities of the actual runoff conditional upon a forecast, and transition densities of forecasts are obtained from a Bayesian information processor. Parametric densities are derived for the process with a normal prior density of the runoff and a linear model of the forecast error. The structure of the model and the estimation procedure are motivated by analyses of forecast records from five stations in the Snake River basin, from the period 1971–1983. The advantages of supplementing the current forecasting scheme with a Bayesian analysis are discussed.