Doctorate research: new techniques to improve discharge forecasts for the Rhine at Lobith

Published: 13 May 2020

Water managers use discharge forecasts all the time as the basis for their decisions. A good forecast depends on estimating how 'full' a river basin is first before looking to the future using weather forecasts. A series of steps are needed to estimate how full a river basin is, including estimating the spatial distribution of precipitation, temperature and potential evapotranspiration, and high-resolution spatial hydrological modelling.


Researcher Bart van Osnabrugge studied all these steps and selected the discharge of the Rhine at Lobith as a case study. His doctorate award ceremony was yesterday at Wageningen University. Bart conducted his study at Deltares, under the banner of the H2020 IMPREX project. One of the components he studied was interpolation for precipitation. He specifically investigated how a reliable spatial distribution of precipitation can be determined on the basis of a limited dataset that is available in near-real time. This is based in part on monthly precipitation distributions derived from a more accurate and longer climatological dataset.


He also investigated the effect of various potential evapotranspiration (PET) products on hydrological discharge forecasts. He investigated the differences between using a PET climatology based on a long time series of historical PET, and PET in near-real time based on current satellite data and weather forecasts. It emerged that, in the case of a ten-day discharge forecast, there is hardly any difference between the two approaches when it comes to estimating PET.

Lake levels in Switzerland and improved use of discharge data

Bart also looked at adjusting hydrological model conditions on the basis of water levels in Swiss lakes and measured discharges in German tributaries. It emerged that using the water levels in the Swiss lakes improved the discharge forecast at Basel for a window of about ten days. In addition, using discharge data for German tributaries such as the Moselle, Main and Neckar to adjust hydrological model conditions had a positive effect on the quality of the discharge forecasts for these tributaries in the slightly longer term.


In addition to the scientific relevance in terms of operational discharge forecasting, the results of this study also led to specific recommendations for improving hydrological forecasting for the Rhine basin. The quality of the Rhine discharge forecasts at Lobith in summer can be improved with the assimilation of the water levels of the Swiss lakes. Further improvements could be made in this respect by using the discharge data for the tributaries in Germany differently.

The operational approach developed here for spatial precipitation interpolation has already been implemented in the operational instrument used by Rijkswaterstaat Water, Traffic and Environment. The use of daily estimates of spatial potential evapotranspiration based on satellite measurements is recommended for estimating the current status of the Rhine basin. Particularly in combination with a high-resolution spatially distributed hydrological model like the one that has also been developed in this H2020 IMPREX project.