A Study of a Monsoon Depression Bringing Record Rainfall over Australia. Part I: Numerical Predictability Experiments

Abstract

Between 11 and 15 March 1989, a low pressure system developed in the monsoon trough over the northwest coast of Western Australia. Subsequently an intense precipitating depression moved southeastwards across Australia, bringing flood rainfall to a wide area of the driest parts of the continent. The synoptic?dynamic aspects of this case are described in a companion paper. A series of studies are presented in this paper, examining various aspects of the numerical predictability of this event. Forecasts are compared from a static cold start analysis (the operational system at the time), a limited-area data assimilation system with operational (2.5 h) data cutoffs, and the same assimilation system with an infinite data cutoff. It is shown that the operational forecasts of the cyclone's position, intensity, and precipitation were clearly inferior to both assimilation forecasts. The assimilation forecasts of rainfall subjectively showed a greater margin of improved skill over the operational forecasts based at 1200 UTC than they did for the 0000 UTC-based forecasts. This suggests that the assimilation system, by the use of model feedback in the first-guess fields and the greater use of remotely sensed data, is less sensitive to the variations in radiosonde data density at 1200 UTC than was the operational system. The late data cutoff assimilation forecasts were superior to those with the operational cutoffs, and it was shown by a sensitivity experiment that most of this improvement was due to the inclusion of satellite sounding data from the TIROS series of orbiting satellites (TOVS data), which do not arrive in time for the operational 0000 and 1200 UTC analyses. This suggests that when locally processed TOVS data become available, an improvement in forecast skill should be realized. A series of model sensitivity experiments, in which forecasts are compared with and without the precipitation parameterization activated show that the forecast intensification of the low as it crossed the continent was significantly aided by feedback from latent heat release. Another interesting point resulting from these sensitivity experiments is that latent heat feedback contributed to the forecast intensification of the jet stream well downstream of the precipitating depression.