Quantification of uncertainty in reference evapotranspiration climate change signals in Belgium

Projections of evapotranspiration form the basis of future runoff and water availability assessment in a climate change context. The scarcity of data or insufficiency of time/funds compels the application of simple reference evapotranspiration (ET_o) methods requiring less meteorological inputs for ET_o projections which adds uncertainty to the projected changes. This study investigates the bias in ET_o climate change signals derived from seven simple temperature- and radiation-based methods (Blaney–Criddle, Hargreaves–Samani, Schendel, Makkink, Turc, Jensen–Haise, Tabari) compared with that from the standard Penman–Monteith FAO 56 method on the basis of 12 general circulation model (GCM) outputs from the Coupled Model Intercomparison Project Phase 5 for central Belgium for four future greenhouse gas scenarios (RCP2.6, RCP4.5, RCP6.0, RCP8.5). The results show the lack of conformity on the amount of ET_o changes between the simple and standard methods, with biases of over 100% for some simple methods. The uncertainty affiliated with ET_o methods for monthly ET_o changes is smaller but of comparable magnitude to GCM uncertainty, which is usually the major source of uncertainty, and larger for daily extreme ET_o changes. This emphasizes the imperative of addressing the uncertainty associated with ET_o methods for quantifying the hydrological response to climate change.