Hi AFT, This was part of an email I had in response to my questions re the GFS. To be honest I suspect as Paul quite rightly states above most of the data would try to be incorporated in each run. The NOAA email makes it even more confusing though...sorry! <_< 'Model runs are generally made four times a day at 00Z (7:00 pm EST), 06Z (1:00 am EST), 12Z (7:00 am EST) and 18Z (1:00 pm EST). Since radiosonde observation balloons are released throughout the world at 00Z and 12Z, the models are "initialized" at these times with real data for subsequent runs. When there is little or no data, for example for the "off hour" 06Z and 18Z runs, the models are initialized from previous model run results. ' PS - It was New Years Eve when I asked the question though, so maybe he thought I was strange asking that question around 5pm, so he could have been waiting to go home for a drink! :lol: Kind regards CV