The procedure to determine evaporation in hydrological models is considered to be unsatisfactory by some researchers; ‘too’ accurate by others. In this procedure catchment scale evaporation is related to some form of potential evaporation, determined with point scale meteorological data. The main criticism is that the potential evaporation is not representative for the catchment and that spatio-temporal dynamics in vegetation cannot appropriately be expressed with the time-invariant, spatially lumped model parameters in the above mentioned procedure. Using remotely sensed observations, catchment scale estimates on evaporation and vegetation dynamics can be derived. It is hypothesized that by integrating remotely sensed evaporation estimates and additional information on vegetation dynamics in conceptual rainfall-runoff models, we can get more insight into the realism of the modelled evaporation flux and the role of vegetation dynamics. With a more realistic representation of evaporation, the water partitioning can be modelled more accurately, eventually improving our understanding of the catchment behaviour. The way the evaporation estimates can be used depends on the spatial and temporal resolution and the reliability of the products. The hypothesis is tested in the well studied Ourthe catchment, located in Belgium. The climate is Atlantic temperate, the streamflow is characterized by a quick response. Vegetation dynamics in both space and time are investigated in a principal component analysis on multi year MODIS Normalized Difference Vegetation Index (NDVI) data. Areas with similar temporal dynamics are distinguished. The most important temporal dynamics are related to phenology and agricultural growing reasons. Areas with an increasing trend in NDVI are identified as well, but the spatial extent is too small to be relevant for hydrological applications. In a validation study three remotely sensed evaporation products are examined in terms of their reliability and applicability in conceptual models, namely EARS (daily, 4km x 9km), WACMOS (daily, clear days, 1km x 1km) and MOD16 (8-daily, 1km x 1km). EARS and WACMOS are surface energy balance models, based on land surface temperature observations. MOD16 uses the Penman-Monteith equation, with distributed surface characteristics and a Jarvis-like approach to calculate the surface resistance. The remotely sensed products are validated with ground measurements of evaporation from five eddy covariance towers and - if applicable - with a multi year water balance analysis. Mainly based on the water balance analysis, we concluded that the EARS product gives relatively accurate evaporation estimates and can be used in the last step of the research. WACMOS (SEBS) was shown to have an extremely poor correlation between the remotely sensed evaporative fraction and the evaporative fraction determined from eddy covariance measurements and was not further considered. MOD16, often criticized for not taking into account soil moisture constraints on evaporation, did deviate from EC measurements in the growing season, but this was shown not to be related to limited soil moisture. The main drawback of MOD16 in relatively moist climates such as the Ourthe catchment, seems to be the application of constant canopy conductance and the large scale of the meteorological input data. Catchment scale evaporation appeared to be slightly underestimated, which may be attributed to the fact that interception is neglected in this version of the product. In the last step of the research, i.e. confronting a water balance model with the ancillary data on vegetation dynamics and remotely sensed evaporation, the hypothesis is further specified. Lead by the spatio-temporal resolution of the EARS product, EARS evaporation (ERS,EARS) is used as forcing of a daily lumped conceptual model. Disadvantage of the coarse spatial resolution and the lumped modelling approach, is that the link between vegetation dynamics and evaporation cannot be made. This is left for further research. In a comparative modelling approach i) the realism of the conventional procedure to determine evaporation is examined, and ii) we investigate whether by imposing ERS,EARS on the model, a more realistic representation of the water partitioning can be simulated. Three models are compared: FLEXEp with the conventional procedure to determine evaporation, FLEXEp,RS also forced with potential evaporation, but determined from the catchment average net radiation, and FLEXE, forced with RS,EARS. The hypothesis is that especially the seasonal dynamics in streamflow generation can be better simulated by FLEXE. This was not the case. It was shown that the evaporation modelled with the conventional conceptualization of the evaporation flux (FLEXEp), is fundamentally different from the EARS remotely sensed evaporation estimates (ERS,EARS). Yet in terms of streamflow simulation, FLEXEp outperforms (FLEXE), especially in spring and autumn. Furthermore, parameter identifiability was shown not to be better for FLEXE than for FLEXEp. These results indicate that either the EARS product is not as accurate as we expected, or that the lumped conceptual model does not represent the dominant processes occurring in the catchment. Assuming the latter, this can be explained by i) a too complex model with too many degrees of freedom as the limited parameter identifiability indicates, ii) an erroneous representation of the dominant processes or a wrong hypothesis on the occurring dominant processes and iii) the level of aggregation of evaporation and other hydrological processes is too high and the model too simple to be representative for the heterogeneous catchment. To look further into the latter, we suggest the combined use of topography driven semi-distributed models and the patterns in vegetation as derived by the PCA analysis. Although we assumed the EARS product to reliably estimate the catchment scale evaporation, there are some unexplained issues. Forest evaporation seems te be overestimated by the EARS product, whereas in the water balance analysis the catchment scale evaporation appears to be in the right order of magnitude. If the overprediction indeed is the case, somewhere in the catchment or at some time evaporation is underestimated. Furthermore on clear days in winter, evaporation tends to be zero according to the parametrization of the EARS product, which is not observed by the eddy covariance measurements. It is recommended to further look into the parameterization of the EARS algorithm especially during the changing seasons.