Hydrological ensemble forecasts provide a set of possible realizations of future streamflow. They are used to estimate uncertainties in flood forecast and thereby provide users with more information on the likelihood of flooding.
By using meteorological ensemble forecasts to create hydrological ensemble forecasts, we cover the uncertainty in meteorological data. In addition, uncertainty in the initial state of the model can be taken into account.
Two important initial conditions are the amount of snow that is in the catchment, as well as soil- and groundwater deficit. We can explore different algorithms for statistical calibration of ensemble forecasts and approaches to incorporate uncertainty in initial states. This will contribute to more accurate hydrological forecasts, and thus potentially better flood alerts.
The thesis will require scripting in R and some background in statistics.