Adaptation of Model Output Statistics to Tropical Cyclone Forecasting (Invited Presentation)

Bob Glahn was instrumental in the development of Model Output Statistics (MOS), which have provided valuable forecast information for more than half a century. MOS utilizes statistical methods to relate parameters that are well represented by numerical weather prediction models to forecast variables that are not well represented. For example, due to the complexity of land surface properties and other local effects, the direct prediction of surface temperatures may have limited accuracy at some locations. However, the low level thickness and other parameters that affect the surface are better predicted, and can be used as input to MOS to provide a more accurate surface temperature forecast. MOS techniques have also been applied to tropical cyclone forecasts since the late 1950s. The early techniques concentrated on both track and intensity, but the emphasis in the 1960s through 1980s was on track. The direct prediction of track by dynamical models became much more accurate in the 1990s so statistical track forecast techniques were no longer needed. Direct prediction of intensity was much more problematic and a number of statistical intensity methods have been utilized since the early 1990s. More recently, statistical post processing methods for TC tracks and intensity have regained favor in the context of providing optimal combinations of ensemble forecasts. The history of statistical tropical cyclone forecast methods is briefly reviewed, current methods are summarized and a future outlook is presented.