Tuesday, 17 April 2018
Champions DEFGH (Sawgrass Marriott)
In the context of hurricane model development, model evaluation is largely based on numerical integrations of past events, where integrations including a proposed model change are evaluated against control simulations and the best available observations. Proposed model changes that are promising on the face of the resulting model evaluation are subject to evaluation based on progressively larger samples of storms, however there are severe constrains on the number of storms available for model development, usually limited to a couple of years of the most recent historical record. This limitation is believed to be related to the different performance that retrospective runs often have when compared with real time model performance, once model changes are adopted. In this work we begin to explore how storm sample choice may influence results of specific model changes. Our work is presented in the context of an emerging objective criteria of storm choice for model development.
- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner