In this paper we will introduce the model system NOWVIV which was tested during a field campaign accomplished at the Fairchild-Dornier airfield in Oberpfaffenhofen, Germany during April/May 2001.  That campaign, named WakeOP, aimed at predicting, monitoring and characterizing the evolution of the wake of a given aircraft flying with prescribed configurations in various meteorological conditions. The model system is designed to provide forecasts of meteorological parameters to predict wake vortex transport and behavior. We  investigate the quality of the NOWVIV predictions in comparison to observations from a SODAR/RASS profiler.

A hierarchy of weather forecast models is combined within the model system "NOWVIV" - NOwcasting Wake Vortex Impact Variables -which run automatically at DLR in an operational mode during WakeOP. NOWVIV uses output data from the operational weather forecast model "LM" of the German Weather Service which runs at a resolution of 7 km. LM on the other hand is driven by the global model "GME" which runs at a resolution of 60 km. A core element of NOWVIV is the Penn State/NCAR mesoscale model "MM5" which is used to predict atmospheric state variables within a 2.1 km grid around the airfield with an increasing vertical spacing from 25 to 50m throughout the boundary layer. Detailed terrain and landuse information is provided to NOWVIV.  NOWVIV forecasts were successfully used to plan flight tracks during WakeOP.

NOWVIV is initialized every 12 hours, 12 UTC and 0 UTC. Locally measured WakeOP data have not been assimilated so far, however this is planned in the near future. Output variables are vertical profiles of horizontal and vertical wind, u,v,w, virtual potential temperature and turbulent kinetic energy. These variables are required to run a real-time wake vortex decay and transport model.

As an example the NOWVIV RMS error of cross wind (wind speed) is 2.5 m/s (3.4 m/s) compared to 3.5 m/s (5.2 m/s) for the LM.  This  result is based on the analysis of 10 days comparing predictions and observations every hour. The ten days comprise various weather situations and phenomena including e.g. a Foehn-event and a nocturnal low-level jet. In general LM overestimates the average wind speed by about a factor of two compared to the forecast of NOWVIV The better NOWVIV forecast quality can be attributed to the finer model resolution with a better representation of local orography and landuse. In addition, the forecast quality of turbulent kinetic energy and temperature is investigated placing emphasis on the foreast dynamics.  This is an importent aspect for a realtime wake vortex prediction system.