A Hydrologic Drying Bias in Water-Resource Impact Analyses of Anthropogenic Climate Change

For water-resource planning, sensitivity of freshwater availability to anthropogenic climate change (ACC) often is analyzed with offline hydrologic models that use precipitation and potential evapotranspiration (E-p) as inputs. Because E-p is not a climate-model output, an intermediary model of E-p must be introduced to connect the climate model to the hydrologic model. Several E-p methods are used. The suitability of each can be assessed by noting a credible E-p method for offline analyses should be able to reproduce climate models' ACC-driven changes in actual evapotranspiration in regions and seasons of negligible water stress (E-w). We quantified this ability for seven commonly used E-p methods and for a simple proportionality with available energy (energy-only method). With the exception of the energy-only method, all methods tend to overestimate substantially the increase in E-p associated with ACC. In an offline hydrologic model, the E-p-change biases produce excessive increases in actual evapotranspiration (E), whether the system experiences water stress or not, and thence strong negative biases in runoff change, as compared to hydrologic fluxes in the driving climate models. The runoff biases are comparable in magnitude to the ACC-induced runoff changes themselves. These results suggest future hydrologic drying (wetting) trends likely are being systematically and substantially overestimated (underestimated) in many water-resource impact analyses.