Catchment and land surface hydrology

To understand and predict the behaviour of a river, we need to look at their catchments: the land from which their water originates. The processes at the land surface determine whether precipitation (rain and snow) ends up in the river (and how fast), or that it evaporates underway. The exchange of water between the atmosphere, the ground and surface water bodies depends on many factors, such as vegetation, soil, weather and the surface water system. We aim to understand the hydrological processes and simulate them, in order to predict how systems react to changes in land use, climate or water management. We pay special attention to understanding and forecasting hydrologic extremes such as floods, droughts and heatwaves, and to quantifying the uncertainty belonging to these simulations. In the following sections we describe a few (but not all) directions of our research.

Land-atmosphere interactions

Evaporation is controlled by atmospheric conditions (solar radiation, temperature, wind speed humidity), availability of water in the ground, and properties of the land surface (type and state of the vegetation, soil). These factors are not independent, but influence each other via feedback mechanisms, and these feedbacks are different under normal conditions than under extremely dry or hot conditions. Understanding these feedbacks helps to prepare for future changes in climate and land use and find solutions to reduce the impact.

Related research topics:
- Understanding drought and heatwave processes and impact
- Predicting the impact of land cover and forest change on the water cycle
- Using big data, machine learning and remote sensing to analyse land-surface
processes

Illustration of the feedbacks at the land surface. Figure from A.J. Teuling (2018): A hot future for European droughts. Nature Clim Change 8, 364–365

Rainfall-runoff processes in lowland catchments

The Netherlands is a typical example of a lowland, with limited topography, abundant surface water and many structures to control the water levels. The characteristics of this landscape affect the pathways rain water follows to the river: water does not only flow downwards into the ground (percolation) but can also move upward (capillary rise), the activation of fast flow routes (for example through drainpipes) depends mostly on the wetness of the catchment, and water does not only flow from the ground to the surface water but also the other way around. Accounting for these processes in hydrological models is necessary to simulate the response of rivers to rainfall events, and to forecast floods and droughts in lowlands.

Related research topics:
- Understanding the importance of initial conditions in flood generation
- Making projections of discharge and groundwater levels based on climate
scenarios
- Simulating the effect of water management on hydrological processes
- Implementing lowland-specific processes in hydrological models

The identification of important processes in lowland catchments (left) and how they are implemented in a model (right). Figures related to C.C. Brauer, A.J. Teuling, P.J.J.F. Torfs, R. Uijlenhoet (2014): The Wageningen Lowland Runoff Simulator (WALRUS): a lumped rainfall-runoff model for catchments with shallow groundwater, Geosci. Model Dev., 7, 2313-2332

Hydrologic forecasting

Many people around the world rely on timely and accurate forecasts of floods and
droughts because early warning can save lives and reduce damage. The performance of hydrological forecasts depends on the quality of the input data (typically precipitation and evapotranspiration), the model used for the simulations and the estimate of initial conditions in the model (the wetness of the catchment prior to the event). We study the techniques to improve these factors, such as rainfall nowcasting to improve the input or data assimilation to improve the initial conditions, and their effectiveness in increasing the forecast skill.

Related research topics:
- Improving the skill of flood and drought forecasts and warnings
- Improving flood early warning in urban areas
- Nowcasting (short-term forecasting) of rainfall as input for flood early warning
systems
- Developing hydrological models (including machine learning and artificial
intelligence) and parameter estimation techniques
- Using observations to improve initial conditions in real-time simulations (data-
assimilation)
- Assessing the effect of real-time control and other water management actions
- Simulating related processes, such as salt intrusion in estuaries or reservoir
operation

Example of a discharge peak forecast issued at four times (colors), using radar rainfall nowcasting (panel b shows whether different nowcasting algorithms lead to threshold exceedance). From R.O. Imhoff, C.C. Brauer, K.J. van Heeringen, R. Uijlenhoet, A.H. Weerts (2022): Large-sample evaluation of radar rainfall nowcasting for flood early warning, Water Resour. Res., 58, e2021WR031591

Hydrologic model uncertainty

Hydrologic simulations, both offline simulations and real-time forecasts are never
perfectly accurate. There are measurement errors in the input data, assumptions in the model set-up (which are necessary to simplify the complex reality to a workable set of mathematical equations) and uncertainty in the model parameters (which are related to catchment characteristics but not directly measurable in the field). When setting up a model or designing a study, each modeller will take different decisions. We explore what the effect of those decisions is on the model output.

Related research topics:
- Estimating uncertainty in hydrological modelling and global change assessment
- Understanding the role of context in modelling