Assimilation of zenith total delays in the AROME France convective scale model: a recent assessment

Abstract

The impact of assimilating GPS zenith total delays (ZTD) in the convective scale model AROME is assessed over a 1-month period in summer 2013. The experimental set-up is similar to the current operational usage at Météo-France where the observing system has been expanded in July 2013 in a three-dimensional variational (3D-Var) data assimilation scheme with a 3-hour cycling. Three experiments are performed. In a baseline experiment the GPS ZTD provided through the E-GVAP programme are withdrawn from the observing system (NOGPS). In a second experiment, GPS ZTD from E-GVAP are included in the observing system, representing the operational configuration at Météo-France (EGVAP). The last experiment is similar to EGVAP but new ZTD observations processed by the University of Luxembourg are also assimilated on top of all other observations (UL01). In the first stage, it has been verified through a systematic comparison with model counterparts that the quality of ZTD data processed by the University of Luxembourg is similar to the one provided by other analysis centres from the E-GVAP programme. After a number of quality controls, it has been possible to assimilate around 90 additional observations on top of around 600 stations from E-GVAP every 3 hours. Despite the small fraction of observations assimilated in AROME that ZTD represent (<2%), it is shown that they systematically improve the atmospheric humidity short-range forecasts by a comparison with other observing systems informative about water vapour (radiosoundings, satellite radiances, surface networks) even though it is by small amounts. When examining objective precipitation scores over France, the improvement brought by the UL01 stations on top of E-GVAP is systematic for all daily precipitation thresholds. Examination of several case studies reveals the ability of the ZTD observations to modify the intensity and location of precipitating areas in accordance with previous studies. The addition of ZTD from UL01 is also found to be beneficial, by improving rainfall patterns. Planned improvements to the AROME forecasting and assimilation systems with higher horizontal resolution and hourly cycling of 3D-Var assimilation will be of benefit to ZTD observations.

Keywords

• mesoscale modelling
• zenith total delays
• data assimilation
• numerical weather prediction