Assessing the uncertainty in future runoff estimates due to variability in outputs of a global climate model

Global climate models (GCMs) are an important tool and widely used in climate change impact studies on water resource assessments. Generally, climate projections from the GCMs are downscaled first and then adopted into a hydrological model to estimate climate change impacts on future runoff. However, the projections from the GCMs are considered to be associated with a large degree of uncertainty in quantifying the climate change impacts on hydrology. One of the sources of uncertainties in GCM projections comes from model initial conditions. These conditions normally occur when a GCM is run multiple times with a scenario but produces slightly different output in each run with equal probability. However, studies on quantifying uncertainties due to variability in GCM outputs in future runoff estimates are still limited. This study investigated the impacts of variability of a GCM (CSIRO Mk3) outputs on predicted runoff in the Blue Mountains catchments, New South Wales, Australia. A total of 50 realisations of a GCM outputs under A2 emission scenario were adopted into a hydrological model (Australian Water Balance Model) to identify the variation in estimated runoffs and quantify the uncertainty by calculating coefficient of variation of the simulated runoffs. Preliminary results indicated a notable uncertainty (around 33-37%) in runoff estimates under the projected future climate due to variability in GCM outputs. Since, this large uncertainty may affect decision making in future water resources management, the uncertainty in GCM outputs needs to be investigated along with other potential uncertainties when using GCMs in runoff estimation or climate change impact assessment studies.