1183 A Probabilistic Tropical Cyclone Track Uncertainty “Cone” Using Multi-Model Ensembles

Wednesday, 25 January 2017
4E (Washington State Convention Center )
Derek Ortt, StormGeo, Inc, Houston, TX; and C. Hebert, B. Weinzapfel, and D. Eyre

The authors propose supplementing and eventually replacing the often-misinterpreted “cone of uncertainty” or “error cone” used worldwide with a probabilistic swath derived from members of multiple global ensembles.

For many years, the tropical cyclone forecast error cone has been produced by analyzing track error over the past five years and plotting that error in the form of a five-year, 66.7-percentile cone over the deterministic forecast track.  However, this cone represents neither the level of uncertainty nor the threat posed to a location near the projected path of a tropical cyclone.  Many people mistakenly interpret the cone as a threat area, feeling safe if they are outside the cone.   As track forecasts improve each year, the cone is steadily shrinking.  More and more of a tropical cyclone’s impacts are felt outside the cone.

An ensemble-based probabilistic swath provides an objective estimate of the true forecast uncertainty.  A narrower swath reflects a more certain forecast.  Bifurcated tracks (where a storm will likely take one path or the other) can also be clearly depicted using this technique.

The authors will introduce a product called StormGeo TRAC (Threatened Regions from Active Cyclones).  It shows a probability of a tropical cyclone’s center passing within 200 km of a location based on ensemble members of the ECMWF EPS, NCEP GEFS, and the MSC Canadian Ensemble Forecasts.  Color-shading between 5% of ensemble members and 100% allows a user to assess their likelihood of being near the storm path.  It can be drawn for a given timeframe, such as three, five, or seven days.

In this presentation, the decisions that went into creating TRAC are explained, and examples and verification of recent storms are shown.  Future work may involve the addition of other ensembles, creation of wind threshold probabilities, and an objective worst-case scenario for a given location.

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner