This dissertation explores the influence of extreme precipitation and the potential impacts of climate change on the vulnerable water resources of the southwestern United States. Specifically, it focuses on 1) the characteristics, origins, and impacts of historical extreme warm-season precipitation in the Lake Mead watershed, 2) improving existing bias correction techniques for projected future streamflows, and 3) investigating the vulnerability of California’s largest reservoir, Lake Shasta, to climate change under existing and adaptive operating protocols. Although the North American Monsoon (NAM) is the main driver of summertime climate variability in the American southwest, considerable knowledge gaps exist regarding its impact at the northern extent of the core region (northwestern Mexico, southern Arizona, and New Mexico). The first part of this dissertation catalogues historical extreme precipitation events in the Lake Mead watershed (located at the NAM’s northern boundary) and identifies unique synoptic drivers of extreme precipitation between the canonical NAM region and watersheds to the north. From here, the dissertation shifts its focus from the historical period to future climate projections. Motivated by a desire to connect bias correction techniques to the underlying dynamics within earth systems models, a novel statistical method is developed for projected streamflow wherein data are windowed based on hydrograph-relative time, rather than Julian day. This method yields improved preservation of original climate model data for both extreme and non-extreme events. Utilizing these bias corrected streamflow projections, and a simplified model of operations at California’s largest reservoir, Lake Shasta, developed by the author, coming threats to water supply and flood risk under existing operations and several forms of adaptive responses to climate change are analyzed. Compared to the historical period, we simulate 27% declines in carryover storage (storage on September 30th) at the end of the 21st century under a severe warming scenario if operations are left unchanged. Despite many simulated interventions favoring water supply over flood risk, historical levels of carryover storage were irretrievable at the end of the century under the warmer of the two warming scenarios examined in this study.

The purpose of this thesis is to examine and advance North American weatherpredictability from weather to subseasonal time-scales. Specifically, it focuses on 1) developing machine learning/deep learning methods and models to improve predictability through numerical weather prediction (NWP) post-processing on weather time-scales (0-7 days) and 2) examining the physical mechanisms which govern the evolution of the predictable components and noise components of teleconnection modes on subseasonal time-scales (7 days - 1 month). NWP deficiencies (e.g., sub-grid parameterization approximations), nonlinear error growth associated with the chaotic nature of the atmosphere, and initial condition uncertainty lead initial small forecast errors to eventually result in weather predictions which are as skillful as random forecasts. A portion of these forecast errors are inherent to the NWP models alone, systematic biases. The first two chapters develop cutting-edge vision-based deep-learning algorithms to advance the current state-of-the-art NWP post-processing and correct these systematic biases. Using dynamic forecasts of North Pacific integrated vapor transport (IVT) as a test case, we develop post-processing systems which are spatially aware, readily encode non-linear predictor interaction, easily ingest ancillary weather variables, and have state of the art training methods that systematically prevent model overfitting. Further, we outline a framework to quantify uncertainty in single-point (deterministic) forecasts using neural networks. The uncertainty is shown to be probabilistically rigorous, leading to calibrated probabilistic forecasts which outperform or compete with calibrated dynamic NWP ensemble systems for IVT under atmospheric river conditions.

The second half of this thesis shifts focus to subseasonal time scales and examinespredictability in the Pacific North American (PNA) sector in boreal winter. Particularly, it investigates the physical mechanisms involved in the intraseasonal modulation of atmospheric Signal-to-Noise (SN), and how it is affected by slowly varying climate modes (ENSO and MJO). These mechanisms are further explored using a fully-coupled hindcast of the 20th century, showing that the increased SN leads to high model forecast skill at subseasonal timescales in particular forecast windows of opportunity. Additionally, we reveal the MJO as the largest growing mode of tropical forecast uncertainty which directly influences PNA forecast certainty.

Atmospheric rivers (AR) have been found to substantially contribute to the annual precipitation in many mid-latitude regions, but while their synoptic scale dynamics have been studied for a century (through extratropical cyclone research), the mesoscale dynamics are less understood and present forecasting challenges. In this dissertation we focus on i) the pre-cold-frontal low-level jet (LLJ), ii) moisture effects on the orographic and dynamic precipitation components, and iii) physical processes associated with AR initial condition sensitivity.

An analysis using dropsonde observations from 24 cross-AR transects found that the LLJ is strongly supergeostrophic with about 20% of the wind being ageostrophic. The ERA5 reanalysis product is used to investigate the ageostrophic forcing mechanism finding that the pressure tendency term serves to accelerate the ageostrophic jet, and the Coriolis torque and advective tendency terms serve to propagate the LLJ. Therefore, to accurately simulate an LLJ, a model must adequately resolve the pressure tendencies along the cold front.

Next, an analysis of a strong landfalling AR is presented comparing the evolution of a control simulation with an optimally perturbed simulation using the moist adjoint tool from the Coupled Ocean/Atmosphere Mesoscale Prediction System. The perturbed simulation shows a strengthened role of the orographic and dynamic components of the precipitation both resulting from enhanced latent heating effects by which: i) a stronger diabatically driven low-level potential vorticity anomaly strengthens the water vapor transport and thereby the orographic precipitation, and ii) a greater moist diabatic forcing enhances the transverse circulation thereby increasing ascent and the dynamic precipitation.

An idealized two-dimensional moist frontogenesis model is then used to quantify the transverse circulation response to varying amplitudes of optimal moisture perturbations demonstrating that small changes in moisture leads to comparatively larger changes in the frontal ascent. This non-linear growth is due to a feedback process in the transverse circulation by which small changes in moisture cause greater latent heating which ultimately lead to a strengthened transverse circulation and the cycle repeats.

The research presented in this dissertation improves the understanding of AR mesoscale dynamics through investigation of LLJs, effects of moisture perturbations, and by identifying an important feedback process.

California has extremely high interannual variability in precipitation, and this is projected to increase with climate change. During northern California’s wettest year on record (2017), successive atmospheric river (AR) events associated with extreme precipitation served both to alleviate drought conditions and contribute to the Oroville Dam crisis. This dissertation investigates the characteristics, synoptic patterns, and drivers of these successive ARs.

First, this dissertation introduces the concept of “AR families” using an objective method to identify successive AR events at a single location (Bodega Bay, CA). Compared to single AR events, AR families exhibit unique, identifiable and semi-stationary patterns across the North Pacific. The algorithm was then applied to all AR events impacting California over 39 years, 1981-2019. K-means clustering identified six distinct clusters of large-scale patterns associated with AR families. These clusters represent two types of predominant patterns: meridional and zonal, and two show strong relationships with the El Niño/Southern Oscillation.

Persistent anomalies (PAs) across the North Pacific are evaluated for their association with AR families and drought conditions. This analysis is sensitive to methodology including detrending and determining the appropriate magnitude for the PA, ultimately impacting the identification of PAs associated with drought conditions or AR families in California. Traditional methodologies lack a strong correlation between PA frequency and precipitation in California, therefore new methods are explored to meaningfully relate the frequency and location of PAs to anomalous precipitation statewide.

Two satellite sea surface temperature (SST) datasets are used to assess the relationship between wintertime SST variability and AR activity, showing low levels of spatial SST variability co-occur with low AR frequency. Within AR events, near shore buoy data shows that the air-sea temperature difference prior, during, and post AR passage is consistent with an approaching warm front, AR conditions and cold air advection, respectively. Case studies of AR families further investigate the variability between AR activity, SST, and precipitation.

These findings identify the importance of successive AR events to California’s climate and characterize their synoptic precursors and meteorological drivers that provide the foundation for increased situational awareness and predictability of their occurrence and impacts.