COAMPS-TC QPF as a function of track error and human impacts

Worldwide, the greatest threat to human life and property from tropical cyclones is due to flooding caused by extreme rainfall amounts. This is particularly true on the island of Taiwan, where mountainous terrain provides significant upslope enhancement of the precipitation produced by landfalling typhoons. This study examines the QPFs of three simulations by a special version of the Naval Research Laboratory's Coupled Ocean-Atmosphere Mesoscale Prediction System (COAMPS) designed for forecasting tropical cyclones (COAMPS-TC). A key difference between the cases is the size of the track error and the resulting QPF errors.

The best simulation is for Typhoon Jangmi (2008) which contains very low track errors and very accurate predictions of rainfall amounts, including the two highest totals which exceeded 1000 mm. On the other hand, the simulation for Typhoon Sinlaku (2008) resulted in a relatively small, but still pronounced southward deviation in the model track upstream of the island compared with the actual storm. This relatively small track error results in quite large QPF errors of both positive and negative sign along a confined band of latitudes. Multiple simulations were performed for the record-breaking Typhoon Morakot (2009). The first contains the largest track errors of any of the three cases, and predictably, precipitation amounts are significantly underforecast for the areas of heaviest flooding, as well as overpredicted in areas where light rainfall was observed. However, when the forecast track is improved through adjustment of the coarse domain boundaries, underprediction remains in place, though greatly reduced from the previous simulation. The portion of the QPF errors due to track displacements is examined more closely with a simple analytical model.

The QPFs from the two simulations of Typhoon Morakot are also examined in terms of alternative forecast verification methods, specifically probability of detection (POD) and false alarm ratio (FAR). Various thresholds for "significant flooding" are used to calculate these statistics at each model grid point. Using this metric shows that COAMPS-TC actually performs well over the regions which experienced the record breaking rainfall, but actually scores lower in regions on the eastern shoreline, both in terms of lower POD's and higher FAR's. This fact should be taken into account when further improvements of the model QPF are attempted.