Date of Award




Document Type

Master's Thesis

Degree Name

Master of Science (MS)


Department of Atmospheric and Environmental Sciences

Content Description

1 online resource (x, 69 pages) : color illustrations, color maps.

Dissertation/Thesis Chair

Ryan D Torn

Committee Members

Lance F Bosart


Numerical weather forecasting, Cyclone forecasting, Atmospheric waves, Rossby waves, Ensemble learning (Machine learning)

Subject Categories

Atmospheric Sciences


Baroclinic cyclogenesis can act as an impulsive disturbance along the midlatitude waveguide and promote the dispersion of Rossby waves, potentially impacting the predictability of the downstream midlatitude flow. Indeed, several studies have traced medium-range forecast errors and uncertainties to upstream cyclone development, especially when diabatic processes are involved in modifying the upper-tropospheric potential vorticity (PV) distribution. However, it remains unclear why some forecasts downstream of cyclogenesis are particularly uncertain while others are not. This study uses a 30-yr (1985–2014) sample of cool-season North Pacific and North Atlantic baroclinic cyclones and forecasts from the second-generation Global Ensemble Forecast System (GEFS) reforecast dataset to assess whether certain midlatitude flow configurations are systematically associated with large and small downstream forecast uncertainty. The degree of forecast uncertainty is quantified using normalized 500-hPa geopotential height forecast standard deviation (SD) anomalies, which are a function of geographic location, time of year, and forecast lead time. Composite analyses reveal that in each basin large downstream forecast SD tends to be associated with prominent ridge amplification east of the cyclone and uncertainty in the evolution of a downstream trough. In contrast, small forecast SD is associated with more coherent Rossby wave packets that propagate along the waveguide and less persistent flow anomalies. Self-organizing maps (SOMs) are used to assess the probability of large downstream forecast SD being observed when certain flow configurations and near-storm environmental conditions are present. In particular, composites are constructed to diagnose the potential role of diabatic activity in the warm sector of the developing cyclone and interaction with the upper-tropospheric waveguide. In the North Pacific, flow configurations that favor large and small forecast SD are characterized by a similar degree of upper-tropospheric PV advection by the irrotational wind at the time of cyclogenesis, which suggests differences in downstream predictability are not, on average, explained by differences in the upstream waveguide perturbation. In the North Atlantic, strong (weak) latent heat release and perturbing of the upper-tropospheric waveguide in conjunction with cyclogenesis is associated with a reduced likelihood of small (large) forecast SD downstream. Additional research should be undertaken to understand whether these associations are indicative of systematic links.