A new bias correction scheme for assimilating GPS zenith tropospheric delay estimates

The impact of GPS zenith tropospheric delay estimated from the groundbased network of GPS receiving stations in the Norwegian regional model is assessed in this work. The data are screened from the stations with irregular or insufficient dissemination of observations. A normality test on the observation minus modelequivalent of the remaining dataset led to 29 assimilated pairs of stations/processing centers from the initial 668. A site-dependent multi-linear scheme that uses model surface pressure, skin temperature, integrated water vapor, and tropospheric thickness as predictors to correct the observation departure biases has been implemented. This scheme is proved to successfully remove the observational error bias. A two-month impact study during the winter of 2008 showed that the selected and bias-corrected dataset was of benefit for geopotential and short-range humidity fields. The impact on temperature fields and precipitation forecasts was neutral.