Date of Award

12-1-2022

Language

English

Document Type

Master's Thesis

Degree Name

Master of Science (MS)

College/School/Department

Department of Atmospheric and Environmental Sciences

Content Description

1 online resource (xiii,112 pages) : illustrations (some color), color maps.

Dissertation/Thesis Chair

Ryan D Torn

Committee Members

Kristen Corbosiero

Keywords

Cyclone forecasting, Dropwindsondes

Subject Categories

Atmospheric Sciences | Meteorology

Abstract

Tropical cyclone (TC) hazards are primarily dictated by the TC position; thus, it is important to produce accurate TC track forecasts. Many different features can influence a TC’s motion, yet these features are not always well sampled by in-situ observations over the open ocean, creating a need for supplemental observation collected via aircraft, including the deployment of dropsondes. In 1997, the National Hurricane Center (NHC) began operational synoptic surveillance missions in the near-storm environments of TCs using the Gulfstream IV-SP jet aircraft (G-IV), with the goal of reducing track forecast errors. In the first 10 years of operational missions, the dropsonde data collected during 176 G-IV missions led to a 10–15% improvement in 0–60-h track forecast errors for forecasts initialized at mission nominal times (Aberson 2010). However, despite the addition dropsondes deployed around the TC core to operations and additional research into optimal targeting strategies, no further research has been published regarding the impacts of G-IV dropsondes on recent TC track forecasts. Therefore, this thesis expands on previous studies by investigating the impacts of G-IV dropsonde data on Atlantic basin TC track forecasts from 2018–2021 when ensemble-based sensitivity was utilized for creating aircraft tracks. This research investigates the impacts of dropsonde data collected during NOAA G-IV synoptic surveillance missions on position forecasts for Atlantic basin TCs from 2018–2021 by comparing forecasts initialized with dropsonde data available against forecasts without dropsondes and to the forecast initialized 12-h before dropsondes. ECMWF EPS, GEFS, and CLP5 position forecasts are analyzed for 675 forecast initialization times, of which 56 had dropsonde data assimilated at the time of initialization. When compared to the 619 forecast initialization times without dropsondes, the forecast initialization times with dropsondes are found to have lower average position errors and higher skill relative to CLP5, suggesting that the dropsonde data is likely adding skill to position forecasts. In contrast, when forecasts initialized with dropsondes are compared to forecasts initialized 12-h before dropsondes, the results indicate that differences in skill and average position error are similar to differences seen in randomly selected pairs of forecast initializations from climatology. Furthermore, four cases studies are conducted to diagnose the potential sources of significant position error reductions in track forecasts initialized with the assimilation of dropsondes. Two potential sources are identified: changes in the environmental steering flow that yield differences in TC motion (hurricanes Zeta (2020) and Jerry (2019)) and a change in the initial position of the TC that placed the TC into a different steering flow regime (hurricanes Marco (2020) and Dorian (2019)).

Download

Share

COinS