Tropical Island Convection in the Absence of Significant Topography. Part II: Nowcasting Storm Evolution

Abstract

This paper examines influences on the short-range prediction of organized convection under conditions of strong diurnal forcing. The analyses are based on data provided by the Maritime Continent Thunderstorm Experiment, which was conducted in 1995 over the Tiwi Islands (11°S) 50?100 km north of the Australian continent. Organized convection over the Tiwi Islands is often dubbed ?Hector? by residents and researchers alike. The authors' purpose is to utilize results from these analyses to improve convective storm ?nowcast? systems and their associated forecast and warning products. The environmental near-surface wind direction is shown to be singularly influential in statistically predicting the location of convection and its time of occurrence over the Tiwi Islands. This finding is robust, despite the fact that mean wind speeds were merely 1?4 m s?1. The island boundary layer water vapor mixing ratio, derived from a morning sounding and inland surface stations, is shown to be correlated with measures of overall convective activity. Mesonet inferences of ambient water vapor anomalies, together with geostationary satellite cloud imagery and radar data, point to favored locations for individual thunderstorm initiation nowcasts. The detailed evolution of Hector thunderstorms on any given day is dependent upon the location and movement of quasi-chaotic interactions among sea breezes, gust fronts, cumulus clouds, and existing storms. The primary mechanism for increasing the size of Hector was an excitation of new convection and merging with existing storms in response to forcing by a westward-propagating gust front. This is fully consistent with previous works, which have examined the effects of convergence lines, cold pools, shear, and the merging of radar echoes. Dissipation of Hector may be predicted as it moves westward from land to ocean and, occasionally, when it moves over land to areas that have been cooled by earlier storms. The results of this study have important implications for expert and numerically based forecasting methods concerned with thunderstorm prediction in the 0?6-h range. A twofold approach exhibits promise in the Tiwi Islands: 1) use of statistical information provided by a dynamically based climatology (?6 h forecast) and 2) monitoring and extrapolation of existing convergence lines, storms, and cumulus clouds for individual thunderstorm predictions (0?2 h). Variational assimilation of such information into high-resolution forecast models should lead to improved dynamical predictions in the 2?6-h range.