Stochastic generation of weather for estimating hydrometeorological extremes. Potential of a Random Pulse Model for Swiss River Catchments

4 to 6 months. As soon as possible

Estimating the risk for rare to very rare floods at any location of a given river is still a key challenge for hydrologists worldwide. For a given location, this typically requires long time series of streamflow records. Most locations are however ungauged and have no such records. For locations where streamflow records are available, the length of the record is typically much too short and in addition, available streamflow time series are often unusable because, in many cases, the hydrological behaviour of the river catchment upstream has been significantly modified over time by numerous waterworks and activities (e.g. multi-usage water reservoirs).

One powerful alternative approach developed for decades by hydrologists worldwide for estimating flood risk is Continuous Hydrometeorological simulation (CS). Long time series of high resolution synthetic streamflow are simulated thanks to an appropriate hydrological model of the river catchment of interest. The weather scenarios required as input for the simulation are obtained in a preliminary step thanks to a stochastic weather generator able to generate long time series of high-resolution weather for the considered catchment, especially precipitation and temperature.

IGE and INRAE have a worldwide recognised expertise in the development of weather generators. Our current developments focus on the generation of weather scenarios for small to mesoscale alpine river catchments in Switzerland. They are mainly carried out within 2 PhD thesis funded by the EXAR project, a collaborative project funded by the Swiss Confederation.

The objective of the present master work is to evaluate the potential of the BLRPM model, a weather generator based on the so called “Random Pulse” generation approach where rainfall sequences are assumed to be the results of the successive random occurrences of different rainfall pulses of different lengths and intensities. The content of the work will be :
• Estimation of the BLRPM parameters for a large ensemble of small (gauged) Swiss catchments.
• Generation of long time series of precipitation scenarios for each catchment and evaluation of the simulation skill (ability to reproduce a number of key statistical properties of observed precipitation).
• Development and evaluation of a spatial estimation model for the BLRPM parameters (that will allow to apply the model for any ungauged catchment anywhere in Switzerland).
• Comparison of the simulation skill of the regional BLRPM with the skill of a reference weather generator, currently developed within the EXAR PhD thesis.
Depending on the results, the master report is expected to be submitted for a scientific publication in a peer-review journal.