Publication Date

2018-04-23

Availability

Open access

Embargo Period

2018-04-23

Degree Type

Dissertation

Degree Name

Doctor of Philosophy (PHD)

Department

Meteorology and Physical Oceanography (Marine)

Date of Defense

2018-04-05

First Committee Member

Ben P. Kirtman

Second Committee Member

Amy Clement

Third Committee Member

Brian Soden

Fourth Committee Member

Shimon Wdowinski

Abstract

In this manuscript our goal is to increase our understanding of the components of the climate system that contribute to coastal flooding along the east coast of the United States (US) with a particular focus on the state of Florida. To achieve this goal we use a high resolution (HR) configuration of the Community Climate System Model Version 4 (CCSM4) with a 0.1° resolution in the ocean and a 1.0° resolution in the atmosphere that is able resolve currents, eddies, and other small scale features in the ocean as well as accurately depict the shape of the coastline. We compare the results of unforced HR runs as well as a HR climate change run and HR hosing experiments to results from low resolution (LR, 1.0° in the ocean) runs of CCSM4, and to observations and reanalysis in to highlight the differences between the behavior of the HR and LR models and to validate the HR models as a better representation of the climate system. We begin by assessing the predictability of regional climate trends using the hindcasts of the North American Multi-Model Ensemble (NMME), a seasonal forecasting system with short lead times out to a maximum of only 9 months. The NMME uses general circulation models with a typical, 1.0° resolution in the ocean. Linear trends of SST and precipitation were calculated on a point-by-point basis for 28 years of hindcasts and subsequently a Monte Carlo Method (MCM) was implemented to correlate the spatial pattern of the simulated trends at 1-month, 3-month, and 6-month lead times to observed trends for thousands of randomly selected regions of varying sizes. With a 1-month lead-time the models have some degree of forecast skill on a regional scale; however, there is little to no long-lead (6 months or longer) forecast skill for randomly selected regions. However, there are specific areas where trends are well forecasted. The MCM is used to identify these regions and to determine how pre-selected regions perform against all others of that particular size. We then use a series of unforced runs of HR CCSM4 to characterize the natural variability of SSH along the Atlantic coast of the US, and to try and determine the driver of coherent coastal sea level along the entirety of the Atlantic coast from Canada to South America. The relationship between coastal SSH and other components of the climate system is compared by correlating coastal SSH with SST, surface KE, and SSH at all other points in the North Atlantic. The same procedures were applied to an unforced LR run of CCSM4 and Simple Ocean Data Analysis (SODA) reanalysis. We find that the HR, unlike the LR, reproduces these connections as they occur in reanalysis. We also reveal a secondary component of coastal SSH that is systematically out of phase between coastal SSH to the south of the Gulf Stream separation and the overall mean of coastal SSH. The same natural variability runs of CCSM4 were used to identify a pattern of precipitation variability that links a positive precipitation over the southern US and Gulf of Mexico in the boreal fall and winter to elevated SST and SSH along the US Atlantic coast at low-frequencies. This mode of precipitation variability is the leading EOF in the Gulf of Mexico and western north Atlantic region in models and satellite observations. Enhanced precipitation associated with positive SSH in this region has potential negative implications on coastal flooding. Finally, HR hosing experiments with 0.1 Sv and 1.0 Sv of constant freshwater forcing in the north Atlantic were compared to a LR 0.1 Sv hosing experiment and a climate change run of HR CCSM4 with historical levels of CO2 and no freshwater forcing. The results of the LR hosing experiment are consistent with many previous studies and indicate that 0.1 Sv of hosing will result in a rapid slowing of the AMOC and a dramatic increase in coastal SSH along the coast of the US, particularly to the north of Cape Hatteras. Surprisingly, the SSH differences in the HR 0.1 Sv and 1.0 Sv hosing experiment fall within the natural variability of coastal SSH. We are currently limited to 45 years of output for each of the HR hosing experiments. Therefore, it is possible that the freshwater forcing will eventually lead to SSH changes that are outside the range of natural variability. Given that 1.0 Sv is a massive, unrealistically large freshwater forcing the results of these HR hosing experiments suggest that the dynamic SSH response to melting the Greenland ice sheet may be smaller than previously hypothesized or will at least take much longer to develop than LR models have projected.

Keywords

High-resolution Climate Modeling; Coastal Flooding; Regional Sea Level Rise; Natural Variability of Coastal Sea Surface Height

Share

COinS