Abstract
The objective of this project was to bring light on the problems of using potential evapotranspiration (PE) as input to rainfall-runoff models and to assess the relevance of this variable in this modelling context. The tests carried out in this project are based on several rainfall-runoff models that were run on large databases with either general or regional climatic characteristics. The tests have shown that using a simple PE formulation only based on temperature as measured variable can give as good results as using a more complex combinatory formulation like the Penman one. Besides, results show that rainfall-runoff models are not able to make a better use of more detailed PE information, either on a temporal or a spatial point of view. One reason is the incapacity of their production function to exploit this extra information. In a Mediterranean context, it is even possible to use a constant limit PE as an optimised parameter instead of using actual PE data.