The role of local conditions in explaining high and low flow trends in Norway

Knowledge on streamflow variability is important for natural ecosystems, water supply and hydropower among others.

In Norway, we have observed significant changes in natural low and high flow the last decades, but the spatial and seasonal pattern in the changes, and whether they are positive or negative, varies across the country. We know that streamflow is a combined response of the meteorology and local conditions in the catchment.

However, we still need to improve our understanding of how important the different factors are in explaining the various trends, and in particular the role of different catchment characteristics. This is important to improve our understanding of low and high flow trends in ungauged catchments and what to expect where in the future.

The project formulation is flexible and will be shaped in collaboration with the student.

  • Potential research questions are:
  • Which local characteristics and meteorological factors are most important in explaining the pattern in high and low flow trends in Norway?
  • Are the meteorology sufficient to understand the trends we see, or are some (if so; which) local characteristics important as well?
  • What is the role of seasonality, or changes in seasonality?
  • How well can we predict whether a Norwegian catchment has a decreasing trend, increasing trend or no trend in low and high flow?

Methods include using statistical/machine learning methods, such as lasso regression and random forest, to investigate which factors are most important for predicting which stations experience significant trend in a flow index, and if so, in which direction. Flow indices could be winter low flow, summer low flow, spring high flow, autumn high flow, seasonal average flow, etc. Do the most important predictors vary with the flow indices and/or season?

All data is available from The Norwegian Water Resources and Energy Directorate (NVE), The Norwegian Meteorological Institute (MET) or online. We will need natural streamflow data in Norway, and corresponding catchment characteristics and meteorological data.

The project can be expanded to include other countries as well if we are able to get sufficient data.

Tags: Hydrology, climate change, data analysis, machine learning, drought, floods
Published Oct. 20, 2020 11:42 AM - Last modified Oct. 20, 2020 11:49 AM

Scope (credits)