Rasmus E Benestad, Christine Achberger & Elodie Fernandez

Empirical-statistical downscaling of distribution functions for daily precipitation

Regclim results

A relatively new approach based on predictions of the parameters of a variable's distribution functions is a promising method for downscaling daily precipitation.

Here, two different approaches are investigated, one assuming that the precipitation follows a Gamma distribution and one assuming a simpler exponential distribution law requiring one parameter only.

The scale and shape parameters for the Gamma distribution and the slope parameter for the exponential function are tested for dependencies against large-scale climatic conditions. The number of rainy days and shape parameters can be associated with the large-scale circulation, but the relationship between the scale and the atmospheric flow is weak at best. The slope parameter for the exponential distribution exhibits a strong dependency to local mean temperature, mean precipitation and altitude.

The dependency between these parameters and climate variables is utilised in a projection of the frequency distribution for 24hr precipitation for an SRES-A1b-based climate-change scenario for 2070.

Read the report: pdf file (43 pages, 4.8 Mb )