Saturation limiters for water vapour advection schemes: impact on orographic precipitation

A common fault of atmospheric general circulation models (AGCMs) is an overestimation of orographic precipitation. One basic reason is that water vapour advection schemes do not use information about the local temperature. When water vapour is advected from a warm grid point to a colder one, supersaturation may occur on the way, and the water vapour advected may partly precipitate before reaching the latter. This process is particularly important when moisture is advected upward mountain slopes along terrain‐following coordinates. Not taking it in account, i.e., letting all the moisture reach the colder point, leads to artificial drying of the windward valleys and foothills, and to overestimation of rainfall over summits and plateaux. This spurious behaviour is amplified by the resulting biases in the circulation, due to misplacement of the moisture convergence. It is a general bias, although its magnitude may be reduced, for instance when σ-coordinates are replaced by hybrid coordinates, or increased by highly diffusive schemes such as the upstream finite differencing. A simple way of correcting this bias is to test the advected water vapour with respect to saturation values, and redistribute it accordingly over the grid points found along the advecting path. This method is tested on a finite difference model using σ-coordinates and an upstream advection scheme. The effect on the distribution of moisture and rainfall is dramatic: precipitation is displaced from summits and plateaux to slopes and foothills, leading to much more realistic patterns, in particular for the Indian and Amazonian monsoons.