Fuzzy Rule-Based Forecasting of Extreme Rainfall Probability Conditioned on Sea Surface Temperature

See related content

by Andras Bardossy,
Lucien Duckstein,
Istvan Bogardi,

Document Type: Proceeding Paper

Part of: WRPMD'99: Preparing for the 21st Century

Abstract: (No paper) The recent El Nino event demonstrated the importance of sea-surface temperature on rainfall anomalies. The purpose of this paper is to develop a methodology to forecast extreme rainfall probability based on sea surface temperature anomalies. As a first step in an exploratory analysis the regions of influence are identified. The second step is to describe the relationship between maximal daily precipitation and sea-surface anomalies. The complex and non-linear relationship between the variables can be described flexibly with fuzzy rules. The annual cycle of this inter-variable relationship is also taken into account. Separate rules were established for each season. The rules and the corresponding forecasting uncertainty are assessed using a split sampling approach. The methodology is applied in the Ruhr catchment (4,000 km2) in Germany. A relationship between monthly sea-surface anomalies of the North Atlantic and the areal precipitation are established using data from the time period 1961-1990. The probability distribution of rainfall amounts is forecasted with a time lag of 2 to 3 months. The probabilities are compared to observed maxima using the likelihood ratio approach.

Subject Headings: Forecasting | Probability | Rainfall | Temperature effects | Seas and oceans | Fuzzy sets | Climate change | Water resources | Germany | Europe

Services: Buy this book/Buy this article


Return to search