The El Niño-Southern Oscillation (ENSO) dominates interannual climate variability in the tropical Pacific, with sea surface temperatures (SSTs) oscillating between cooler (La Niña) and warmer (El Niño) conditions. ENSO affects midlatitude weather and climate through reorganizing winds and circulation, shifts known as teleconnections for their far-reaching impacts.

ENSO-induced teleconnections were once considered stationary and predictable, but observations show they differ widely between individual events. Internal climate variability produces distinct ENSO events with diverse amplitudes, SST patterns, and ocean-atmosphere feedbacks, causing large uncertainties in ENSO rainfall prediction. Moreover, limited instrumental observations, based on fewer than 50 ENSO events in the 20th century, are insufficient to characterize ENSO complexity and its teleconnections.

This thesis uses highly-resolved paleoclimate data spanning the Last Millennium (LM), including climate models and paleoclimate data assimilation (DA) techniques, to expand 20th-century data by thousands of years. We examine the changes in ENSO characteristics and North American hydroclimate teleconnections over the LM, investigating rainfall shifts due to background temperature changes and ENSO teleconnection variability on decadal-to-centennial timescales.

We first evaluate ENSO diversity and North American teleconnections over the LM. DA products show intensified Eastern Pacific El Niño in the 20th century, but no change in Central Pacific El Niño frequency. Shifts in El Niño hydroclimate teleconnections over North America are variable in magnitude but consistent in sign over the LM. Then, we present a regional focus on Mississippi River basin hydroclimate extremes, revealing that strong El Niño-like warming over the tropical Pacific and North Atlantic SST patterns jointly contribute to wet extremes across the basin and its major tributaries. Finally, we assess the stationarity of North American hydroclimate responses to various climate modes, finding that teleconnections vary on multi-decadal and multi-centennial timescales throughout the LM, driven partly by predictable SST patterns.

Natural variability in rainfall patterns characterized in this work will compound with global warming to shape 21st-century conditions. This research provides new insights from paleoclimate data, expanding statistical constraints on the stationarity and predictability of ENSO impacts, and serves as a last-millennium benchmark for climate prediction and water risk assessments under the 21st-century anthropogenic warming.