The Extratropical Transition of Tropical Cyclones: Present-Day Climatology, Future Projections, and Statistical Prediction

This thesis addresses the extratropical transition (ET) of tropical cyclones. ET is the process by which a tropical cyclone, upon encountering a baroclinic environment at higher latitudes, loses its tropical characteristics and transforms into an extratropical cyclone. The three main goals of the thesis are to develop a historical climatology of global ET occurrence, to examine future projections of ET using a global climate model, and to advance the predictive understanding of ET. A global climatology of ET from 1979-2017 is presented, which explores frequency of occurrence, geographical and seasonal patterns, climate variability, and environmental settings associated with different types of ET in global ocean basins. ET is defined objectively by means of tropical cyclones' trajectories through the Cyclone Phase Space (CPS), which is calculated using storm tracks from best track data and geopotential height fields from reanalysis datasets. Two reanalysis datasets are used and compared for this purpose, the Japanese 55-year Reanalysis (JRA-55) and the ECMWF Interim Reanalysis (ERA-Interim). Results show that ET is most common in the western North Pacific and the North Atlantic, where about half of the tropical cyclones transition into extratropical cyclones. Coastal regions in these basins also face the highest rates of landfalling ET storms. In the Southern Hemisphere basins, ET percentages range from about 20% to 40%. Different "ET pathways" through the CPS are linked to different geographical trajectories and environmental settings: A majority of ETs start with the tropical cyclone becoming thermally asymmetric and end with the formation of a cold core. This pathway typically occurs over warmer sea surface temperatures and takes longer than the reverse pathway, in which a tropical cyclone undergoes ET by developing a cold core before becoming asymmetric. The classifications of tropical cyclones into "ET storms" (tropical cyclones that undergo at some point in their lifetimes) and "non-ET storms" (tropical cyclones that do not undergo ET) obtained from JRA-55 and ERA-Interim are evaluated against the classification obtained from the best track records. In contrast to the CPS definition of ET, which is automated and objective, the best track definition of ET is given by the subjective judgment of human forecasters who take into account a wider range of data. According to the F1 score and the Matthews correlation coefficient, two performance metrics that balance classification sensitivity and specificity, the CPS classification agrees most with the best track classification in the western North Pacific and the North Atlantic, and least in the eastern North Pacific. The JRA-55 classification achieves higher performance scores than does the ERA-Interim classification, mostly because ERA-Interim has a bias toward cold-core structures in the representation of tropical cyclones. Future projections of ET are examined using a five-member ensemble of a coupled global climate model, the Flux-Adjusted Forecast-oriented Low Ocean Resolution (FLOR-FA) version of CM2.5 developed at the Geophysical Fluid Dynamics Laboratory. First, CPS is applied to 1979-2005 FLOR-FA output to develop a historical ET climatology, which is compared to the 1979-2005 ET climatology obtained from JRA-55. This comparison shows that FLOR-FA simulates many unrealistic low-latitude ET events, due to strong local maxima in the geopotential height fields used as input to calculate the CPS parameters. These local maxima, which arguably result from strong grid-scale convective updrafts, mislead the CPS to detect an upper-level cold core where one is not present. Three solutions to this problem are examined: changing the algorithm to compute the CPS parameters such that it uses 95th percentile values of geopotential instead of the maxima, a temporal smoothing of the CPS parameters, and a combination of the previous two. All three modifications largely correct the misdiagnosed cases. Future (2071-2100) projections of ET activity under the Representative Concentration Pathway 4.5 are then explored. A number of changes between the future and historical simulations are robust with respect to the different modifications to the CPS described above, though few are statistically significant. A statistical model that predicts ET in the western North Pacific and the North Atlantic is introduced. The model, a logistic regression with elastic net regularization, was developed with a focus on predictive performance as well as physical interpretability and thus resides at the interface between machine learning and traditional statistics. It uses eight predictors that characterize the storm and its environment, the most important ones being latitude and sea surface temperature. The model is shown to have skill in forecasting ET at lead times up to two days, and it can predict the phase evolution of storms that undergo ET as well as of storms that remain tropical throughout their lifetimes. When used as an instantaneous diagnostic of a storm's tropical/extratropical status, the model performs about as well as the CPS in the western North Pacific and better than the CPS in the North Atlantic, and it predicts the timings of the transitions better than the CPS in both basins. The model can be integrated into statistical tropical cyclone risk models, or may be applied to provide baseline guidance for operational forecasts