Page Banner

United States Department of Agriculture

Agricultural Research Service

Research Project: INTEGRATION OF CLIMATE VARIABILITY AND FORECASTS INTO RISK-BASED MANAGEMENT TOOLS FOR AGRICULTURE PRODUCTION AND RESOURCE CONSERVATION

Location: Great Plains Agroclimate and Natural Resources Research Unit

Title: Assessment of the scale effect on statistical downscaling quality at a station scale using a weather generator-based model

Authors
item Chen, Jie -
item Zhang, Xunchang
item Brissette, Francois -

Submitted to: International Journal of Climatology
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: March 30, 2013
Publication Date: May 20, 2013
Citation: Chen, J., Zhang, X.J., Brissette, F.P. 2013. Assessment of the scale effect on statistical downscaling quality at a station scale using a weather generator-based model. International Journal of Climatology. DOI:10.1002/joc.3717.

Interpretive Summary: The resolution of Global Climate Models (GCMs) is too coarse to assess the fine scale or site-specific impacts of climate change on agricultural production. Downscaling methods have been developed to fill the gaps. As the resolution of climate model increases, it is imperative to know whether the finer resolution of Regional Climate Models (RCMs) would result in any improvement in statistical downscaling quality at the station scale. The objective of this study is to assess the scale effect of GCM/RCM output on statistical downscaling quality of precipitation using a weather generator-based model. The downscaling is conducted across three scales, from 300-km, 50-km, and 15-km resolutions to a station scale for two Quebec stations. Observed precipitation gridded to the corresponding scales is also studied in parallel. The results show that the frequencies of dry and wet spells, relative to those of observed data, are reasonably well downscaled for all spatial scales. The statistics of downscaled precipitation are somewhat overestimated for the Sept-Iles station, but underestimated for Bonnard station at all daily, monthly and annual scales. The number of wet days per year is well preserved for all downscaling experiments and two stations. There are no notable differences in downscaling quality between using gridded and modeled data for both stations. Using the gridded monthly precipitation as a predictor, as the resolution increases, the downscaling quality is slightly improved for Bonnard station. However, there is no apparent improvement, when using projected monthly precipitation as the predictor. Overall, there is no noteworthy scale effect on downscaling quality when the weather generator-based downscaling model is used to downscale GCM/RCM precipitation to a station, indicating that the downscaling method works equally well for both high and low resolution precipitation data. The results will be useful to scientists and engineers who are interested in assessing the impact of climate change on natural resource conservation and agricultural production.

Technical Abstract: The resolution of General Circulation Models (GCMs) is too coarse to assess the fine scale or site-specific impacts of climate change. Downscaling approaches including dynamical and statistical downscaling have been developed to meet this requirement. As the resolution of climate model increases, it is imperative to know whether the finer resolution of Regional Climate Models (RCMs) would result in any improvement in statistical downscaling quality at the station scale. The objective of this study is to assess the scale effect of GCM/RCM output on statistical downscaling quality of precipitation using a weather generator-based model. The downscaling is conducted across three scales, from GCM, and mid and high resolution RCMs to a station scale for two Quebec stations. Observed precipitation gridded to the corresponding scales is also studied in parallel, totally six downscaling experiments. The results show that the frequencies of dry and wet spells, relative to those of observed data, are reasonably well downscaled for all six experiments. The statistics of downscaled precipitation are somewhat overestimated for the Sept-Iles station for all six downscaling experiments, but underestimated for Bonnard station at all daily, monthly and annual scales. The number of wet days per year is well preserved for all downscaling experiments and two stations. There are no notable differences in downscaling quality between using gridded and modeled data for both stations. Using the gridded monthly precipitation as a predictor, as the resolution increases, the downscaling quality is slightly improved at all daily, monthly and annual scales for Bonnard station. However, there is no apparent improvement, when using projected monthly precipitation as the predictor. Overall, there is no noteworthy scale effect on downscaling quality when the weather generator-based downscaling model is used to downscale GCM/RCM precipitation to a station, indicating that the downscaling method works equally well for both high and low resolution precipitation data.

Last Modified: 9/21/2014
Footer Content Back to Top of Page