2007 Annual Meeting and Arctic Forum | Abstracts

Identification of Dominant Patterns in Pan Arctic River Discharge Trends

Asa K. Rennermalm¹, Eric F. Wood^2
1Department of Civil and Environmental Engineering, Princeton University, Engineering Quad, Princeton University, Princeton, NJ, 08544, USA, Phone 609-356-2659, arennerm@princeton.edu
²Department of Civil and Environmental Engineering, Princeton University, Engineering Quad, Princeton University, Princeton, NJ, 08544, USA, Phone 609-258-4675, Fax 609-258-2799, efwood@princeton.edu

In the 21st century, climate change due to anthropogenic release of greenhouse gases is expected to be most pronounced in the arctic region. That arctic climate change already might be underway is indicated by 20th century increase of Eurasian river discharge into the Arctic Ocean. However, there is a spatial variability in arctic discharge trends exemplified by the late 20th century decreased river discharge into Hudson Bay. The causes for the spatially varying arctic discharge trends and are largely unknown. Finding explanations for 20th century discharge trends have been difficult because of the sparse temporal and spatial coverage of arctic environmental and hydrological data.

We propose that trend attribution can be aided by the identification of regions within the arctic drainage basin that have similar trend characteristics. Here we present a framework to seek for dominant patterns in arctic river discharge trends. The framework combines a simple description of trends in station discharge measurements with a clustering and classification technique to reduce the complexity of arctic discharge data. Indeed, with this method we are able to identify dominant trend patterns of pan arctic discharge trends that also show some spatial patterns. Thus, the presented framework has the potential to significantly simplify the trend attribution process by reducing the problem from finding the cause of the trends in every single river basin to only finding the cause to a few dominant trend patterns. The limitations to trend attribution set by data gaps in time and space may be solved for by using environmental and hydrological data from all regions where a dominant trend pattern appears.