P3.16
Bias in dynamically downscaled precipitation data and its impact on simulated streamflow in climate change impact studies
Susan C. Steele-Dunne, TU Delft, Delft, Netherlands; and R. McGrath, P. Lynch, T. Semmler, S. Wang, J. A. Hanafin, and P. Nolan
In the latest IPCC report, recent observations confirm increases in global mean temperatures, rising global average sea level and diminishing snow and ice cover. This warming, and the consequent rise in atmospheric water vapor have led to an increase in mean precipitation over northern Europe as well as an increase in the frequency of heavy precipitation events over most land areas. It is expected that global average surface air temperatures will continue to increase into the 21st century, and that hot extremes, heat waves and heavy precipitation events will continue to increase in frequency. One of the goals of the C4I project is to determine which Irish catchments are vulnerable to increased flood risk due to the predicted changes in precipitation patterns. Boundary data from the European Centre Hamburg Model Version 5 (ECHAM5) global climate model are used to force the Rossby Centre Atmosphere Model (RCA3) regional climate model, producing dynamically downscaled precipitation and temperature data under past and future climate scenarios. These data are used to force the HBV-light rainfall/run-off model to simulate streamflow in the validation period (1961-2000), and in the future (2021-2060) under the SRES-A1B scenario. Here, it is demonstrated that the dynamically downscaled precipitation data include a significant, positive bias. This results in a considerable overestimation of streamflow in all catchments studies. The spatial and temporal variability of the bias is explored by comparing simulated precipitation to precipitation observed at seven gauge locations for the full reference period. Observed and simulated circulation patterns are analyzed to identify potential sources of the bias. It is shown that even implementation of a simple bias correction scheme can lead to improved simulations of both monthly mean flow and annual maximum daily mean flow for the validation period. In the short term, this correction method provides us with adequate data with which to force the hydrological model to reliably reproduce past streamflow. In the longer term, this study demonstrates the existence of this significant bias and underscores how it greatly diminishes our ability to accurately predict the hydrological response to modeled climate if it is uncorrected.
Poster Session 3, Climate Model and Prediction Poster Session
Wednesday, 23 January 2008, 2:30 PM-4:00 PM, Exhibit Hall B
Previous paper Next paper