Coherent Observation Operators for Surface Data Assimilation with Application to Snow Depth

Abstract

A method to derive observation operators for surface data assimilation is presented. It is based on a statistical modeling of the physical processes behavior near the surface. It provides a general framework to design such observation operators. The method avoids the problems that arise with vertical interpolation and extrapolation due to the difference between real orography and model orography. Examples are given for surface temperature and snow depth analysis. The last example is then implemented in the French global operational model Arpège, and full data assimilation experiments are presented with and without the snow-depth observation operational. The impact of the method is positive, and its amplitude is compared with the one produced by another common model change, namely physical parameterizations modification.