Observing system experiments for monitoring and predicting seasonal-to-interannual climate variations
NEVILLE R. SMITH
RICHARD KLEEMAN
BMRC, Box 1289K,
Melbourne Vic. 3001 AUSTRALIA
Paper for International Workshop on Ocean Climate Variations from Seasons to Decades with Special Emphasis on Pacific Ocean Buoy Network, May 29-31 1996, Mutsu City, Japan.
ABSTRACT
This paper examines the level of redundancy in the systems which (potentially) provide information for initialisation of coupled models for seasonal to interannual climate forecasts. Results from analysis experiments using Tropical Atmosphere-Ocean (TAO) buoy array temperature data, Volunteer Observing System expendable bathythermograph (VOS XBT) data, and a combined data set, are presented. These studies indicate that there is an acceptable level of redundancy in the present systems and that, moreover, the separate systems complement each other in important respects.
Results from several experiments using an intermediate tropical Pacific Ocean coupled ocean-atmosphere forecast model are also presented. This model takes advantage of wind and subsurface observations for initialisation. The addition of all subsurface data clearly improves the model skill compared with the wind data only case, though the impact varies over the last 25 years. The results using TAO and VOS XBT data separately, compared with the combined case, suggest the TAO data are effective at improving the short-term skill (lead times around 4-6 months) while VOS XBT data appear more critical for longer lead times (out to 12 months). It is concluded that this is consistent with TAO and VOS sampling the rapid-response equatorial (Kelvin wave dominated) and slower sub-equatorial (Rossby wave dominated) regions, respectively. The degradation of the model skill after withdrawing either data set is in general relatively small. An experiment in which the total data set is severely degraded (70% of data removed) suggests that, for this model at least, the subsurface data requirements are much less than that presently in place. Several caveats on the interpretation of these results are discussed.
