Decision‐Support System for Long‐Range Stream Flow Forecasting

Abstract

A sensitivity analysis of model forecasts provided possible insight into development of a hybrid technique to improve forecast quality using engineering knowledge and experience. A rule‐based decision support system (DSS) is developed to demonstrate this technique for selecting long‐range water‐supply forecast models. The knowledge‐based system is also used to document experience gained in developing time series models for the purpose of tutoring inexperienced time series modelers. The DSS allows automation of data manipulation, modeling tasks, and as a facility for coding descriptions of expert knowledge. It also acts as an aid for learning time‐series concepts through modules which assist users in producing representative time series models. Use of Unix workstations for decision support is explored through the integration of various software tools with the expert system development tool Nexpert Object. Nexpert Object provides the platform for managing coded experience, as well as controlling external tools for analysis and presentation of results.