Distributed Hydrological Modelling and Application of Remote Sensing Data

The use of physically-based distributed hydrological models demands comprehensive amounts of data both to parameterise and force the model as well as to validate it. Conventionally measured data are mainly restricted to point measurements which are less optimal for use in a spatial distributed context. However, remote sensing can offer such area integrated measures albeit through a more indirect measure than in conventional measures. In the present thesis a modified version of the physically-based distributed MIKE SHE model code has been applied to the Senegal River Basin using conventional and remotely sensed data, respectively. The first study (reported in appendix A) investigates the model performances to be obtained using only conventional data. A rigorous procedure is applied in the parameterisation, calibration and validation of the model in order to maintain control of the data use and enable examination of the effects of calibration and internal model validation. Calibration against one station and internal validation against eight additional stations revealed significant shortcomings for some of the tributaries, in particular in the semi-arid zone of the river basin. Further calibration against additional discharge stations improved the performance levels of the validation for the different subcatchments. Due to lack of validation data below the subcatchment scale the model could only be validated at this scale. Remotely sensed estimates of soil moisture seem to be a possible solution to such a validation below subcatchment scale and can also be used to update the soil moisture state variable in the model. The second study (reported in appendix B) investigates the perspectives in using a remotely sensed dryness index in the model. The index is derived from observations of surface temperature and vegetation index as measured by the NOAA-AVHRR sensor. The index is examined for its relation to model simulated soil moisture and evaporation. The correlation results between the index and the simulation results are of mixed quality. A sensitivity analysis, conducted on both estimates, reveals significant noise on both. The study suggests that the remotely sensed dryness index with its current use of NOAA-AVHRR data does not offer information that leads to a better calibration or validation of a simulation model in a spatial sense.