SEARCH

SEARCH BY CITATION

Keywords:

  • statistical downscaling;
  • atmospheric humidity;
  • Czech Republic

Abstract

The potential for surface humidity variables to be downscaled from large-scale upper air fields is examined. Downscaling studies are rarely aimed at humidity variables, although some climate-change impact models, e.g. the crop models, require them as an input. The surface humidity variables considered are water vapour pressure, dew-point temperature, dew-point deficit, and relative humidity. The potential predictands include several humidity variables, circulation fields (geopotential heights, wind speed, vorticity), and temperature, mainly at 850 hPa. As a downscaling method, multiple linear regression of grid-point values is used. The study is performed for the summer season at four stations in the Czech Republic. Downscaling models explain more variance for the variables that reflect only the water content (vapour pressure and dew-point temperature) than for those depending on temperature (dew-point deficit and relative humidity). The performance for the former two variables exhibits a much smaller (or virtually no) diurnal cycle and much smaller spatial variability. The most efficient predictors are the upper air humidity variables; adding circulation and/or temperature variables to the predictors brings a marginal or even no improvement over the downscaling models based on the humidity quantities only. Copyright © 2005 Royal Meteorological Society