Abstract
Data sensitivity experiments are carried out with a regional objective analysis and forecast model to study the influence of selected components of the FGGE observing system in the high latitude North Atlantic and eastern Arctic. The objective analysts are obtained from an adaptation of the Australian Numerical Meteorology Research Center (ANMRC) analysis scheme, which utilizes a combination of the Successive Correction Method and a variational approach. Similarly, the forecasts are produced by a ten-level ANMRC regional model adapted to the Northern Hemisphere. First-guess fields are obtained from the 6 h FGGE and NOSAT forecasts produced with the fourth-order general circulation model at NASA's Goddard Laboratory for Atmospheres.
Four experiments are conducted, each consisting of three analyses and forecasts. The experiments investigate how forecast skill is affected by the inclusion of (i) satellite temperature retrievals and surface buoy and marine reports; (ii) retrievals, but without the buoy and marine data; (iii) buoy and marine reports in the presence of the satellite temperature data; and (iv) buoy and marine reports in the absence of the satellite temperature data. In general, both the forecast and the analysis models perform favorably when compared to the ECMWF analyses.
The forecast results are case-dependent. In one case the inclusion of the satellite data significantly reduces the 500 mb height and sea level pressure forecast error at 24 and 48 h, independently of whether or not the buoy/marine data are included. In the two other cases, the satellite data provide a very small reduction of error and a very slight increase in forecast error, respectively. The inclusion of the buoy and marine data always reduces the error in the sea level pressure forecast, but the effects on the 500 mb height forecasts are case-dependent. In some cases the inclusion of the buoy data allows additional secondary circulation features to be identified.
An experiment is also conducted to illustrate how the quality of the data used at the boundaries of the regional model can potentially degrade or improve the forecast in the interior of the region.
