The sensitivity of data assimilation on water vapour fields on convection-permitting WRF simulations over the GNSS Upper Rhine Graben Network (GURN), Germany

The assimilation of observations in local area models (LAMs) assures that the states of meteorological variables are as close to reality as possible. Water vapor is an important constituent in terms of cloud and precipitation formation. Its highly variable nature in space and time is often insufficiently represented in models.

The aim of our work is to improve the simulation of water vapour in the Weather Research and Forecasting model WRF by assimilation of different observations. At the current stage, temperature, relative humidity, and surface pressure derived from climate stations are applied as well as zenith total delay (ZTD) data from global navigation satellite system (GNSS) stations. We try to identify the best setup of assimilation parameters which all of them directly or indirectly influence water vapour simulations. We will show case studies of high-resolution WRF simulations (2.1 km) between 2016 and 2018 for different seasons in southwest Germany. The impact of assimilation (3D-VAR) of different variables, combinations of variables, background error option as well as the temporal resolution of assimilation is evaluated. We look at column values and also at profiles derived from radiosondes. Our results show a positive impact when assimilating measured data, but deteriorations are also possible. A distinct influence of assimilation is only apparent for a few time steps. If the temporal resolution of the assimilated variables is too coarse and there is no assimilation close to these time steps, the positive effect vanishes.