Understanding and predicting temperature variability in the observational record

Abstract

Temperature variability and change over land and ocean exhibit characteristic spatial and temporal structures. Understanding the physical mechanisms underlying these structures provides information about the movement and storage of heat in the climate system. In this thesis, I first analyze, and present an energy balance model for, seasonal temperature variability in the extratropics, which supports the idea that the advection of heat between land and ocean by the mean atmospheric circulation can explain the regional characteristics of seasonal variability over both land and ocean. The model is subsequently combined with a large, representative ensemble of Lagrangian atmospheric trajectories to provide a realistic model of the seasonal cycle in the Northern Hemisphere mid-latitudes. Second, based on the Lagrangian trajectories, a new spatially-resolved metric, termed Relative Land Influence, is developed. The metric quantifies the role of land as compared to ocean in influencing the temperature variability at a given location. In addition to explaining the majority of the spatial pattern of seasonal variability, Relative Land Influence is a significant predictor of the observed temperature change over both land and ocean independently since 1950, suggesting that similar physical processes influence temperature variability on seasonal and decadal timescales. Finally, I explore the tails of temperature distributions in the context of identifying the causes of anomalously hot days in the Eastern United States during peak summer. A coupled ocean-atmosphere mode in the central mid-latitude Pacific is identified, which evolves on a characteristic timescale and ultimately leads to the amplification of a mid-latitude wave train that includes a blocking high over the Eastern United States. The early identification of the sea surface temperature precursors to this mode allows for skillful prediction of heat events at lead times greater than 40 days. The identification of physical processes underlying temperature variability on a range of timescales can inform predictions of how temperature variability may change in the future.