Computer simulation models are an important tool for hydrologists. With these models they can predict how rainfall is distributed over the different hydrological variables (discharge, soil moisture and groundwater). This way they can retrieve spatially distributed information that is important for water management, like when to start or stop irrigation gifts or how much water potentially can be stored in a catchment.
The spatial resolution of these models has increased considerably the last 30 years as a result of increased computer capacity and development of GIS. At the same time, information based on remote sensing techniques has become more easily available and its quality improved. These remote sensing data (for example rainfall radar and evapotranspiration images based on satellite data) can be used as input and validation sources for the hydrological models. Besides, meteorologists developed numerical weather prediction models, which outcomes (e.g. rainfall forecasts) can be used by hydrologist in order to make hydrological forecasts of for example groundwater level or soil moisture availability. However, in practice these data are not commonly used due to outstanding questions which formed the research questions of this thesis:
1. Does the accuracy of the hydrological models improve when using rainfall radar data and satellite based evapotranspiration fields?
2. Is it feasible to accurately predict the spatial distribution of soil moisture by using rainfall forecasts of numerical weather prediction model as input for a hydrological model?
To answer these questions we set up a coherent framework to integrate hydrometeorological variables into spatially-distributed models: the Hydrological Now and Forecasting System (HNFS).
The main conclusions of this research are that it is important to take into account the spatial distribution of rainfall in order to get insight in the day-to-day variability of the hydrological system. Using rainfall radar together with rain gauges generates better rainfall fields than using rain gauges only. Besides, information about spatial patterns of satellite based evapotranspiration helps to detect potential model errors. Finally, we found that the accumulated rainfall in our study period (March-Nov 2006) was forecasted very well. However, the spatial variation shown by measured rainfall is not taken into account by rainfall forecasts (due to the lower spatial resolution of numerical weather prediction models). This leads to a spatial bias of forecasted hydrological variables that resembles the spatial pattern in total rainfall within the study area.
This study has shown how remotely sensed and forecasted hydrometeorological variables can be integrated into distributed hydrological models. As this study is based on real data, it has shown the potentials and limitations of applying a system like the HNFS in practice. Finally, considerations about future implementation of this system are given.