Investigating the impact of the water‐vapour sounding observations from SAPHIR on board Megha‐Tropiques for the ARPEGE global model

Abstract

Data assimilation experiments of radiances from the microwave sounder SAPHIR on board the Megha‐Tropiques (MT) satellite were conducted using the global numerical weather prediction (NWP) model ARPEGE (Action de Recherche Petite Echelle Grande Echelle) operational at Météo‐France for the period June to August 2012. The combination of the MT's low‐inclined orbit with the SAPHIR's six channels was found to increase the number of assimilated humidity‐sensitive microwave observations in the Tropics by a factor of 3.8, compared with a control experiment using three existing Advanced Microwave Sounding Unit B and Microwave Humidity Sounder instruments. Impacts of SAPHIR on both analyses through the 4D‐Var system and forecasts are found to be positive with respect to radiosonde data and analyses by the European Centre for Medium‐range Weather Forecasts up to 12 h and neutral to positive up to 72 h. In particular, the improvement of innovation statistics for other humidity‐sensitive satellite observations when SAPHIR is assimilated indicates a positive synergy of the MT satellite with the present observing system. In addition to providing new humidity observations in the Tropics, the SAPHIR data can also be used to define new diagnostics for NWP models. In particular: (i) with the MT's orbit being non‐heliosynchronous, all local times are observed by the same sensor for a given area, which makes diurnal cycle diagnostics straightforward to set up; (ii) with SAPHIR providing more information on the vertical (six channels) than similar sensors, it can be easily used to study water‐vapour vertical structures in NWP models. The first diagnostic applied to the ARPEGE model reveals a good consistency between observed and modelled diurnal cycles in analyses over western Africa, but also reveals a 2–4 h time lag between mean observed and equivalent model brightness temperatures. The second diagnostic highlights a well‐represented vertical correlation of simulated brightness temperatures from the ARPEGE model with respect to SAPHIR over western Africa, but a too‐low vertical correlation over drier regions such as the southern Pacific Ocean during the period June to August 2012.