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 (xi, 111 pages) : color illustrations, color maps.

Dissertation/Thesis Chair

Oliver Elison Timm

Committee Members

Justin Minder


Anthropogenic warming, Climate change, Climate variability, Dynamical downscaling, Pacific Decadal Oscillation, Regional climate modeling, Precipitation (Meteorology), Climatic changes, Ocean-atmosphere interaction, Weather forecasting, Rain and rainfall

Subject Categories

Atmospheric Sciences | Climate


As climate models continue to improve, the demand from resource managers and decision-makers for more accurate climate projections is increasing. However, natural climate variability poses a limit to the confidence in regional climate change projections, particularly for the mid-21st century. The unique geographic location of the Hawaiian Islands and its regional climate provide a challenging opportunity for climate modelers. The goal of this project is to examine both the Pacific Decadal Oscillation (PDO) and anthropogenic climate change for their impacts on near-term rainfall and temperature projections for the Hawaiian Islands. Of primary interest are the questions 1) is there a systematic difference between rainfall anomalies during positive and negative PDO phases? 2) how important is the natural variability compared to anthropogenic forcing for near-term rainfall projections? The Community Earth System Model (CESM) Large Ensemble (Kay et al. 2015) is used in conjunction with WRF for direct dynamical downscaling (i.e., 6-hourly CESM data supply the boundary conditions for the WRF model). An Empirical Orthogonal Function (EOF) analysis of North Pacific SSTs is performed with 35 CESM ensemble members. Ensemble members are grouped into PDO(+) and PDO(-) ensembles. Ensemble members with the five highest and five lowest PDO indices during the present-day period (1996–2005) are identified. These ensemble members are then used to prescribe the SSTs and atmospheric conditions for the downscaled PDO(+)/(-) simulations. Decade-long WRF simulations are composited according to PDO phase for the present-day (1996–2005) and future (2026–2035). The responses to the PDO and anthropogenic climate change are isolated and analyzed by comparing the future simulations to the present-day simulations and the PDO(+) simulations to the PDO(-) simulations. In response to anthropogenic warming, significant increases in surface air temperature (SAT) and SST are projected for the region, with the largest increases in SAT projected to occur at the highest elevations. Significant increases in wet season rainfall (~10–20%) are projected for the windward slopes of Big Island and Maui. Additionally, in a future climate, the PDO(+) phase will become significantly drier than the PDO(-) phase (by 10–30%) at many locations. It is also shown that changes in mean rainfall are strongly influenced by changes in the vertically integrated moisture flux convergence (MFC). Daily rainfall extremes are projected to increase in intensity (by up to 20 mm day-1) at many locations in a future climate. Daily rainfall extremes during PDO(+) conditions are also projected to become significantly more intense (by up to 20 mm day-1) on the windward slopes of Big Island. Significant decreases in rain days and increases in consecutive dry days are projected for central Big Island during the PDO(+) phase. In a future climate, daily temperature extremes are also projected to significantly increase (by up to 1.8 K) across the islands. Overall, results suggest that natural variability will continue to contribute to the uncertainty in near-term precipitation projections, masking the forced signal. The downscaled results also highlighted some critical limitations involved with working with ensembles from a single GCM. In order to fully assess the roles of anthropogenic climate change and internal variability on rainfall projections in Hawai‘i, additional downscaling efforts are needed (e.g. downscaling a different GCM or conducting a similar experiment using the pseudo-global warming method).