Applying 2D Bias Correction Method to Gridded Simulations of Precipitation and Temperature over Southeastern South America.

Abstract:

The two dimensional bias correction methodology for temperature and precipitation, developed by Piani et al. (2012) for station data, was applied to the CCSM4 (NCAR) model gridded output from the CMIP5 dataset and a 40 year gridded dataset over Southeastern South America (Tencer et al., 2011; Jones et al., 2012). Copula density functions of observed temperature and precipitation showed significant structure when subsets of sixteen gridpoints were pooled together. By contrast no structure is detectable in copulas of GCM data. By construction, independent one dimensional bias correction of temperature and precipitation cannot correct copula density distributions hence, the 2D method is applied. The method is validated, as customary, by calibrating and subsequently validating the methodology with non-overlapping 20 year time periods. Visual inspection of single copula density functions for all grid points is unfeasible. Hence the bias correction method is validated by calculating a Kolmogorov-Smirnoff (KS) type statistic measuring the distance between observed and simulated and between observed and corrected copulas at each grid point. Results for the KS statistic are plotted in the figure shown. The methodology clearly shows great potential for application to climate impact modeling.

References

Jones, P. D., Lister, D. H., Harpham, C., Rusticucci, M. and Penalba, O. (2013), Construction of a daily precipitation grid for southeastern South America for the period 1961–2000. Int. J. Climatol., 33: 2508–2519. doi: 10.1002/joc.3605

Piani, C., &Haerter, J. O. (2012). Two dimensional bias correction of temperature and precipitation copulas in climate models. Geophysical Research Letters, 39(20).

Tencer, B., Rusticucci, M., Jones, P., & Lister, D. (2011).A Southeastern South American Daily Gridded Dataset of Observed Surface Minimum and Maximum Temperature for 1961-2000. Bulletin of the American Meteorological Society, 92(10).

Figure. Kolmogorov-Smirnoff type statistic measuring the distance between observed and simulated (left) and between observed and corrected (right) copulas at each grid point.