On the proper use of near-surface temperature observations in atmospheric models deployed over mountain regions

Near surface air temperature is a key meteorological parameter with high implications for the understanding and modelling of snow and water resource in mountain regions. Yet, it is hard to estimate and forecast accurately in these environments due to observational scarcity and model limitations in complex terrain.

In the present study, we analyze whether structural inhomogeneities in observational networks for temperature in mountain regions contribute to errors in their representations in numerical weather prediction (NWP) systems. Taking the case of the Arome-France NWP system over the French Alps, we analyze in particular the effects of the disparity in height above ground of the temperature sensors, of the inhomogeneous geographical distribution of stations that are preferentially located in valleys, and of the frequent altitude mismatch between stations’ real location and model grid points. We evaluate the consequences of these inhomogeneities in terms of model evaluation and data assimilation.

We especially show that measurement height is of high impact for model evaluation, providing a strong incentive to revisit model scores in mountain regions. It also carries strong implications for the assimilation, leading in the case of Arome-France to a negative impact of the assimilation of high-altitude temperature data if their height above ground is not properly considered. Inhomogeneities in data density between mountains and valleys also play a role that can be modulated depending on the assimilation system. This work paves the way for a better use of high-altitude near-surface observations within models deployed over mountain regions.