These images are based upon the randomly selected time step 05 UTC on 2008-04-07. The layer is always the one at the lowest altitude because only this is subjected to downscaling.

Fig. 1: Gravimetric humidity as supplied by the COSMO model: The coarse data.

Downscaled data

In the case of gravimetric humidity, there is no Schomburg rule, so the downscaling is limited to spline interpolation (cf. Fig. 2). Figures 3, 5, and 5.1 are missing for this reason.

Fig. 2: Gravimetric humidity after spline interpolation.

Fig. 4: Gravimetric humidity after spline interpolation and subsequent upscaling back to the coarse grid.

Fig. 4.1: Gravimetric humidity bias after spline interpolation, on the coarse grid. The axis limits have changed.

Even though places with elevated absolute bias of about 8*10^{–10 correspond to places with high gravimetric humidity of about 4*10}–3, the scale difference between the 2 fields is still large enough to attribute the remaining bias to numerical fluctuations. Something must have intensified the fluctuations in certain areas.