Diagnosing the Predictability of Tropical Cyclogenesis via the ECMWF Ensemble

The predictability of tropical cyclogenesis as modeled by the operational ECMWF ensemble is quantified using a number of different techniques, including traditional Lorenzian error-growth calculations, as well as non-traditional methods such as a circulation-following predicted forecast error variance and a newly-developed storm-relative Predictive Power. Predicted forecast error variance is compared against the square of the actual error to diagnose the ability of the ensemble to predict error variance. Predictive Power quantifies predictability as having been lost once the spread of the forecast probability density function (PDF) exceeds the climatological PDF. In this study, Predictive Power is calculated within a wave-following domain of prescribed radius that follows maxima in circulation and developing warm cores in order to isolate genesis events in the ensemble forecast.

It is found that errors associated with vorticity, upper-level divergence, low-level convergence, and moisture within and local to easterly waves is a significant source of uncertainty in the genesis forecast. Much of the small-scale error growth and error variance of the ensemble forecast is associated with localized bands of greater vorticity and divergence, likely related to the difficulty in representing convective bands within the tropical wave.. We investigate the degree to which these errors in the mesoscale limit the predictability of genesis forecasts. Lastly, correlations and lag-correlations between errors are presented. .