THE IMPACT OF SPATIAL AND TEMPORAL DISTRIBUTION OF SATELLITE-OBSERVATIONS ON TROPICAL CYCLONE DATA ASSIMILATION - PRELIMINARY-RESULTS

Preliminary work gauging the impact of varying the spatial and tempora l resolution of Cloud Drift Wind (CDW) data using different assimilati on techniques is presented, particularly within the framework of a gen eralised inverse data assimilation scheme. Results are presented for t he NW Pacific and Australian regions. There were three main findings f or the cases examined. Firstly, hourly and 12-hourly high-density CDWs produced lower mean forecast errors (relative to intermittent assimil ation) than those produced using operational CDWs from the Global Tele communication System (GTS). This is consistent with Le Marshall et al. (1994). Secondly, intermittent assimilation was significantly worse t han both the nudging and variational procedures. Finally, there was li ttle improvement using the variational as opposed to the nudging schem e when using 12-hourly data insertion and operational winds from the G TS. There, the variational procedure proved to be about 4 per cent sup erior to the nudging. This result is of note, given the variational pr ocedure takes about an order of magnitude longer to produce the initia l held than does nudging, and should be contrasted with an earlier fin ding by one of the authors (LML) where the variational procedure was f ound to be clearly superior to the nudging approach (Bennett et al., 1 993) where an enhanced CDW field was used.