Statistical-Dynamical Forecasting of Tropical Cyclogenesis in the North Atlantic at Intraseasonal Lead Times
Abstract
We have created a combined statistical-dynamical model to predict the probability of tropical cyclone (TC) formation at daily, 2.5 degree horizontal resolution in the North Atlantic (NA) at intraseasonal lead times. Based on prior research and our own analyses, we chose five large-scale environmental factors (LSEFs) to represent favorable environments for TC formation. The LSEFs include 850 mb relative vorticity, sea surface temperature, vertical wind shear, Coriolis, and 200 mb divergence. We used logistic regression to create a statistical model that depicts the probability for TC formation based on these LSEFs. Through verification of zero-lead hindcasts, we determined that our regression model performs better than climatology. For example, these hindcasts had a Brier skill score of 0.04 and a relative operating characteristic skill score of 0.72. We then forced our regression model with LSEF fields from the National Centers for Environmental Prediction (NCEP) Climate Forecast System (CFS) to produce non-zero lead hindcasts and forecasts. We conducted a series of case studies to evaluate and study the predictive skill of our regression model, with the results showing that our model produces promising results at intraseasonal lead times.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jun 01, 2009
- Accession Number
- ADA501774
Entities
People
- Chad S. Raynak
Organizations
- Naval Postgraduate School