Seamless global forecasting from days to seasons

Global weather forecasts are available for up to around 2 weeks but lose skill beyond this. Beyond this time scale, statistical or ensemble forecasts based on historical conditions are more common. This project will develop methods to merge the two approaches and so produce seamless probabilistic forecasts from a global biophysical model predicting soil moisture, streamflow and crop production, among others. The influence of model drift and bias in different forcing data sources will be investigated and techniques developed to address them.

This project is in collaboration with Prof. Eric Wood and Dr Justin Sheffield (Princeton University).

Candidates should have some experience with scripting languages and programming. A scholarship may be available.

For more information about this potential research topic or activity, or to discuss any related research area, please contact the supervisor.