Impact of GOES/GMS North Pacific Wind Data on 0 to 5-day Northern Hemisphere Weather Forecasts

The North Pacific Experiment (NORPEX-98) took place between 14 January and 27 February 1998. Special observations collected in NORPEX-98 include approximately 700 "targeted" tropospheric soundings from GPS dropsondes deployed in 38 reconnaissance aircraft missions flown from bases in Hawaii and Alaska (Langland et al. BAMS, 1999, in-press). In addition, an extensive set of wind data was obtained over the entire north Pacific at 6-hr intervals using advanced and experimental techniques to extract information from multi-spectral geostationary (GOES-9 and GMS-5) satellite imagery. These satellite wind data provide a valuable resource for studies of predictability and error growth, and, in an operational context, are perhaps the best source of regular observations over the otherwise data-sparse Pacific basin. Wind observations are available at levels throughout the troposphere, except below cirrus-covered areas.

Here we consider the impact of GOES/GMS wind data, assimilated in various temporal and spatial configurations over the North Pacific, on model analysis and forecasts (up to 5 days) verified over the Northern Hemisphere. These studies are motivated by efforts to determine requirements for future permanent and targeted observing systems that are needed to improve forecasts of land-falling winter storms on the US West Coast and significant weather events over North America, Europe, and Asia. Analysis and forecast differences are displayed as 45-day averages over the NORPEX-98 observational period, in terms of 500 hPa height and total precipitation, with results obtained from the Navy Operational Global Atmospheric Prediction System NOGAPS).

The complete set of NORPEX-98 satellite wind data, assimilated at 6-hr intervals over the entire North Pacific reduces 2-day forecast error (measured by an energy-weighted error norm) over western North America by an average 21%, in comparison to a control with no geostationary wind data included. The January-February 1998 study period includes a strong El Niño event with numerous land-falling Pacific storms affecting California, Oregon, and Washington and record amounts of precipitation in some areas. We will also use satellite wind data from January and February 1999, to examine the effects of satellite wind data in the contrasting flow regime of a La Niña event.