Sunday, 7 January 2018
Exhibit Hall 5 (ACC) (Austin, Texas)
The Dvorak Satellite Technique is a relatively accurate, reproducible method for operationally estimating tropical cyclone intensity and position. This study examines the accuracy and potential biases of the method across two different agencies (the Tropical Analysis and Forecast Branch and the Satellite Analysis Branch), as well as one objective technique (the Automated Dvorak Technique) for Atlantic and East Pacific storms from 2009 to 2016. It is hypothesized that each agency contributes significant skill to a multi-group consensus. To address this theory, Best Track data were collected from the National Hurricane Center’s HURDAT2 database and fix data were collected from the Automated Tropical Cyclone Forecast System as well as the University of Wisconsin’s Cooperative Institute for Meteorological Studies. Best Track data were utilized as verification if there were either aircraft reconnaissance or scatterometer passes within two hours of a given Dvorak fix. Error and bias statistical analyses were completed using Microsoft Excel, while neural network assessments were performed using Weka. Results suggest that the combination of the three helps lower some consensus error metrics. For instance, the consensus’ mean absolute intensity error was 2.63%, 20.6%, and 35.9% smaller than TAFB, SAB, and ADT, respectively. Future studies will be needed to analyze the implications of GOES-16 on the Dvorak technique’s skill, as well as determine any other potential links in Dvorak features to certain error and bias patterns.
- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner