An evaluation of thermodynamic estimates of climatological maximum potential tropical cyclone intensity

This paper investigates the performance of two recently developed thermodynamic models of maximum tropical cyclone intensity (K. A. Emanuel's referred to here as El and G. J. Holland's referred to here as H1), which are designed to estimate the most intense storm possible given the ambient environmental conditions. The study involves estimating the maximum potential tropical cyclone intensity (MPI) from climatological information in three ocean regions, where relatively reliable atmospheric soundings and tropical cyclone intensity data exist. The monthly MPI was estimated for 28 locations across the northwest Pacific, southwest Pacific, and North Atlantic Ocean regions. Empirically derived relationships between observed maximum storm intensity and sea surface temperature were also utilized in the examination of regional MPI model performance. Derived MPIs generally agreed well with observed maximum intensities during the tropical cyclone season. The H1 model tended to underestimate the maximum intensity of storms early and late in the tropical cyclone season and at stations between 10 degrees and 20 degrees N in the northwest Pacific, where the effect of continental air led to weak model estimates for the given surface energy conditions. Additionally, extremely intense H1 estimates were predicted at some stations in the Australian/southwest Pacific region where particularly unstable atmospheric conditions and low ambient surface pressure values are observed. These features of model performance are largely due to the sensitivity of H1 to warm environmental upper-level temperatures. The El model displayed a poor seasonality, frequently predicting the occurrence of storms during winter months. Emanuel MPI estimates were at times underestimated for stations in the North Atlantic and northwest Pacific. The El model estimates in the northwest Pacific were affected by particularly warm upper-level conditions, while relatively high ambient surface pressures in the North Atlantic at 25 degrees N lead to MPI estimates, which are weaker than observed in this region.