Downscaling of GCM forecasts to streamflow over Scandinavia

Research output: Contribution to journalArticle

Abstract

A seasonal forecasting technique to produce probabilistic and deterministic streamflow forecasts for 23 basins in Norway and northern Sweden is developed in this work. Large scale circulation and moisture fields, forecasted by the ECHAM4.5 model 4 months in advance, are used to forecast spring flows. The technique includes model output statistics (MOS) based on a non-linear Neural Network (NN) approach. Results show that streamflow forecasts from Global Circulation Model (GCM) predictions, for the Scandinavia region are viable and highest skill values were found for basins located in south-western Norway. The physical interpretation of the forecasting skill is that stations close to the Norwegian coast are directly exposed to prevailing winds from the Atlantic ocean, which constitute the principal source of predictive information from the atmosphere on the seasonal timescale.

Details

Authors
Organisations
Research areas and keywords

Subject classification (UKÄ) – MANDATORY

  • Water Engineering

Keywords

  • neural networks, statistic, model output, general circulation model, downscaling, forecasting
Original languageEnglish
Pages (from-to)17-26
JournalNordic Hydrology
Volume39
Issue number1
Publication statusPublished - 2008
Publication categoryResearch
Peer-reviewedYes