Keywords
extremes, regional downscaling, precipitation, climate change, extreme value analysis
Start Date
1-7-2010 12:00 AM
Abstract
The ability of regional dynamically-downscaled general circulation models (GCMs) to assess changes to future extreme climatic events was investigated by comparing hindcast model outputs with observations. Projections were generated on a 0.1o grid across Tasmania using the CSIRO Conformal Cubic Atmospheric Model (CCAM). Two future SRES emission scenarios (A2 and B1) and multiple boundary conditions from GCMs were used for the period 1961-2100. A bias-adjustment procedure was employed to spatially correct extreme magnitudes. Events were fitted to a Generalised Pareto Distribution (GPD) using an automated threshold selection procedure developed for gridded precipitation datasets. Estimates of precipitation average recurrence intervals (ARIs) were calculated using extreme value analysis and compared to gridded observations. Spatial patterns were found in gridded precipitation extremes that closely matched observations. Projections of future changes to precipitation extremes were found to vary spatially between models, correlating with projected changes to regional climate drivers. Results demonstrate that dynamical downscaling captures regional climate variability (particularly relevant for precipitation) and displays significant ability in modelling future changes to the intensity, magnitude and frequency of extreme events at the local scale for use in adaptation and emergency planning applications.
Modelling Extreme Events in a Changing Climate using Regional Dynamically- Downscaled Climate Projections
The ability of regional dynamically-downscaled general circulation models (GCMs) to assess changes to future extreme climatic events was investigated by comparing hindcast model outputs with observations. Projections were generated on a 0.1o grid across Tasmania using the CSIRO Conformal Cubic Atmospheric Model (CCAM). Two future SRES emission scenarios (A2 and B1) and multiple boundary conditions from GCMs were used for the period 1961-2100. A bias-adjustment procedure was employed to spatially correct extreme magnitudes. Events were fitted to a Generalised Pareto Distribution (GPD) using an automated threshold selection procedure developed for gridded precipitation datasets. Estimates of precipitation average recurrence intervals (ARIs) were calculated using extreme value analysis and compared to gridded observations. Spatial patterns were found in gridded precipitation extremes that closely matched observations. Projections of future changes to precipitation extremes were found to vary spatially between models, correlating with projected changes to regional climate drivers. Results demonstrate that dynamical downscaling captures regional climate variability (particularly relevant for precipitation) and displays significant ability in modelling future changes to the intensity, magnitude and frequency of extreme events at the local scale for use in adaptation and emergency planning applications.
