Abstract
A study was made of the performance of the Limited-Area Fine-Mesh (LFM) operational forecasts for cases of explosive cyclogenesis in the west-central North Atlantic Ocean during 1981–84. For 51 instances in which the observed 12-h deepening was at least 10 mb, the LFM forecasts for 12–24 h range indicated 58% of this deepening on the average, but accounted for only 30% of the variance in individual cases. In a comparison of central pressures from manual analyses with those in LFM initializations and predictions, the latter were insufficiently deep once rapid intensification began, the discrepancy increasing from about 4 mb in the initializations to about 10 mb in the forecasts from 36 to 48 h. In a number of instances the LFM did not detect the initial appearance of the cyclone. Mean position errors increased from about 75 n mi (140 km) initially to about 185 n mi (340 km) at 48-h range. Mean vector errors were shortly southeast of the analyzed center initially and about 65 n mi (120 km) east-northeast finally, indicating a forecast track slightly too fast and slightly too far to the right. Two individual case studies showed that even when there are large quantitative discrepancies between events in the LFM and real atmospheres, the model is qualitatively correct. These results indicate the essentially baroclinic nature of the cyclogenesis, but the intensity of response to the baroclinic forcing remains intractable.