Towards a Comparative Index of Seaport Climate Vulnerability: Developing Indicators from Open-Data

This work was motivated in part by Austin Becker’s 2013 dissertation, Building Seaport Resilience for Climate Change Adaptation: Stakeholder Perceptions of the Problems, Impacts, and Strategies, which surveyed global port authorities’ perceptions and plans for climate change adaptation and found a disconnect between perceptions of climate impacts and a lack of policies to address them. That work called for the development of a nationwide risk and vulnerability index for ports as a next step in the climate adaptation process for seaports. Climate change adaptation was found to be in the early planning phase for most ports globally, and assessing vulnerabilities is a recommended first step in risk-reduction. In the face of climate change impacts projected over the coming century, seaport decision makers have the responsibility to manage risks for a diverse array of stakeholders and enhance seaport resilience against climate and weather impacts. At the single port scale, decision makers such as port managers may consider the uninterrupted functioning of their own port the number one priority. But, at the multi-port (regional or national) scale, policy-makers will need to prioritize competing port climateadaptation needs in order to maximize the efficiency of limited physical and financial resources and maximize the resilience of the marine transportation system as a whole. Such multi-port decisions can be supported by information products such as indicatorbased composite indices that allow for objective assessment of relative vulnerabilities among a sample of ports. To that end, this work, consisting of three distinct but theoretically related manuscripts, advances the state of data-driven Climate Impact Adaptation and Vulnerability (CIAV) decision-support products for the seaport sector by assessing the current state of vulnerability assessments for seaports (manuscript 1), compiling and refining a set of candidate indicators of seaport climate and extreme-weather vulnerability from open-data sources for 23 major seaports of the United States’ North Atlantic region and creating and applying a Visual Analogue Scale (VAS) instrument for expert-evaluation of the candidate indicators (manuscript 2), and finally by applying the Analytic Hierarchy Process (AHP) with port-experts to weight a selection of the indicators to examine the suitability of the indicator-based vulnerability assessment (IBVA) approach and available open-data to create a composite index of relative climate and extreme-weather vulnerability for the sample of ports. The first manuscript in this work provides an overview of a variety of approaches that set out to quantify various aspects of seaport vulnerability. It begins with discussion of the importance of a “multi-port” approach to complement the single case study approach more commonly applied to port assessments. It then addresses the components of climate vulnerability assessments and provides examples of a variety of approaches. Finally, it suggests an opportunity exists for further research and development of standardized, comparative CCVA methods for seaports and the marine transportation system that can support CIAV decisions and allow decision-makers to compare mechanisms and drivers of climate change across multiple ports. When comparing vulnerabilities of multiple disparate systems such as ports in a region, IBVA methods can yield standardized metrics, allowing for high-level analysis to identify areas or systems of concern. To advance IBVA for the seaport sector, the second manuscript in this work investigates the suitability of publicly available opendata, generally collected for other purposes, to serve as indicators of climate and extreme-weather vulnerability for 23 major seaports in the Northeast United States, addressing the question: How sufficient is the current state of data reporting for and about the seaport sector to develop expert-supported vulnerability indicators for a regional sample of ports? To address this question, researchers developed a framework for expert-evaluation of candidate indicators that can be replicated to develop indicators in other sectors and for other purposes. Researchers first identified candidate indicators from the CCVA and seaport-studies literature and vetted them for data-availability for the sample ports. Candidate indicators were then evaluated by experts via a mindmapping exercise, and finally via a visual analogue scale measurement instrument. Researchers developed a VAS instrument to elicit expert perception of the magnitude and direction of correlation between candidate indicators and each of the three dimensions of vulnerability that have become standard in the CCVA literature, e.g., exposure, sensitivity, and adaptive capacity. For candidate indicators selected from currently available open-data sources, port-expert respondents found notably stronger correlation with the exposure and sensitivity of a port than with the adaptive capacity. Results suggests that better data reporting and sharing within the maritime transportation sector will be necessary before IBVA will become feasible for seaports. The third manuscript in this work describes a method of weighting indicators for assessing the exposure and sensitivity of seaports to climate and extreme-weather impacts. To examine the suitability of IBVA methods and available data to discriminate relative vulnerabilities among a sample of ports, researchers employed AHP to generate weights for a subset of expert-selected indicators of seaport exposure and sensitivity to climate and extreme-weather. The indicators were selected from the results of the VAS survey of port-experts who ranked candidate indicators by magnitude of perceived correlation with the three components of vulnerability; exposure, sensitivity, and adaptive capacity. As those port-expert respondents found significantly stronger correlation between candidate indicators and the exposure and sensitivity of a port than with a port’s adaptive capacity, this AHP exercise did not include indicators of adaptive capacity. The weighted indicators were then aggregated to generate composite indices of seaport exposure and sensitivity to climate and extreme weather for 23 major ports in the North East United States. Rank order generated by AHP-weighted aggregation was compared to a subjective expert-ranking of ports by perceived vulnerability to climate and extreme weather. For the sample of 23 ports, the AHP-generated ranking matched three of the top four most vulnerable ports as assessed subjectively by portexperts. These results suggest that a composite index based on open-data may eventually prove useful as a data-driven tool for identifying outliers in terms of relative seaport vulnerabilities, however, improvements in the standardized reporting and sharing of port data will be required before such an indicator-based assessment method can prove decision-relevant. Overall, this body of work began with a call to develop a method to assess the relative vulnerabilities of seaports to climate and extreme-weather impacts. In the first of three manuscripts, this research identifies an opportunity to contribute to the CCVA literature for the seaport sector by piloting a multi-port vulnerability assessment method based on the use of indicators. The second manuscript in this work contributes to the field of IBVA for seaports by identifying from open-data sources and refining via expert-elicitation methods a set of expert-evaluated candidate indicators of seaport climate and extreme-weather vulnerability. This indicator-evaluation resulted in the finding that adaptive capacity is considered by port-experts as the most difficult of the three components of vulnerability (i.e., exposure, sensitivity, and adaptive capacity) to represent with quantitative data. The final manuscript of this work contributes to the body of CCVA and seaport-studies literature by building and trialing a composite-index of seaport climate and extreme-weather vulnerability based on the evaluated indicators and using AHP to generate component weights. By modeling seaport vulnerability with an indicator-based composite index and comparing results to expert expectations, this work has shown the potential of indicator-based methods to bring a data-driven approach to the CIAV decision-making process, however, results suggest that the current state of publicly available data for and about the seaport sector is not currently sufficient for a robust, expert-supported index.

In the face of climate change impacts projected over the coming century, seaport decision makers have the responsibility to manage risks for a diverse array of stakeholders and enhance seaport resilience against climate and weather impacts. At the single port scale, decision makers such as port managers may consider the uninterrupted functioning of their own port the number one priority. But, at the multi-port (regional or national) scale, policy-makers will need to prioritize competing port climateadaptation needs in order to maximize the efficiency of limited physical and financial resources and maximize the resilience of the marine transportation system as a whole.
Such multi-port decisions can be supported by information products such as indicatorbased composite indices that allow for objective assessment of relative vulnerabilities among a sample of ports.
To that end, this work, consisting of three distinct but theoretically related manuscripts, advances the state of data-driven Climate Impact Adaptation and Vulnerability (CIAV) decision-support products for the seaport sector by assessing the current state of vulnerability assessments for seaports (manuscript 1), compiling and refining a set of candidate indicators of seaport climate and extreme-weather vulnerability from open-data sources for 23 major seaports of the United States' North Atlantic region and creating and applying a Visual Analogue Scale (VAS) instrument for expert-evaluation of the candidate indicators (manuscript 2), and finally by applying the Analytic Hierarchy Process (AHP) with port-experts to weight a selection of the indicators to examine the suitability of the indicator-based vulnerability assessment (IBVA) approach and available open-data to create a composite index of relative climate and extreme-weather vulnerability for the sample of ports.
The first manuscript in this work provides an overview of a variety of approaches that set out to quantify various aspects of seaport vulnerability. It begins with discussion of the importance of a "multi-port" approach to complement the single case study approach more commonly applied to port assessments. It then addresses the components of climate vulnerability assessments and provides examples of a variety of approaches. Finally, it suggests an opportunity exists for further research and development of standardized, comparative CCVA methods for seaports and the marine transportation system that can support CIAV decisions and allow decision-makers to compare mechanisms and drivers of climate change across multiple ports.
When comparing vulnerabilities of multiple disparate systems such as ports in a region, IBVA methods can yield standardized metrics, allowing for high-level analysis to identify areas or systems of concern. To advance IBVA for the seaport sector, the second manuscript in this work investigates the suitability of publicly available open-data, generally collected for other purposes, to serve as indicators of climate and extreme-weather vulnerability for 23 major seaports in the Northeast United States, addressing the question: How sufficient is the current state of data reporting for and about the seaport sector to develop expert-supported vulnerability indicators for a regional sample of ports? To address this question, researchers developed a framework for expert-evaluation of candidate indicators that can be replicated to develop indicators in other sectors and for other purposes. Researchers first identified candidate indicators from the CCVA and seaport-studies literature and vetted them for data-availability for the sample ports. Candidate indicators were then evaluated by experts via a mindmapping exercise, and finally via a visual analogue scale measurement instrument.
Researchers developed a VAS instrument to elicit expert perception of the magnitude and direction of correlation between candidate indicators and each of the three dimensions of vulnerability that have become standard in the CCVA literature, e.g., exposure, sensitivity, and adaptive capacity. For candidate indicators selected from currently available open-data sources, port-expert respondents found notably stronger correlation with the exposure and sensitivity of a port than with the adaptive capacity.
Results suggests that better data reporting and sharing within the maritime transportation sector will be necessary before IBVA will become feasible for seaports.
The third manuscript in this work describes a method of weighting indicators for assessing the exposure and sensitivity of seaports to climate and extreme-weather impacts. To examine the suitability of IBVA methods and available data to discriminate relative vulnerabilities among a sample of ports, researchers employed AHP to generate weights for a subset of expert-selected indicators of seaport exposure and sensitivity to climate and extreme-weather. The indicators were selected from the results of the VAS survey of port-experts who ranked candidate indicators by magnitude of perceived correlation with the three components of vulnerability; exposure, sensitivity, and adaptive capacity. As those port-expert respondents found significantly stronger correlation between candidate indicators and the exposure and sensitivity of a port than with a port's adaptive capacity, this AHP exercise did not include indicators of adaptive capacity. The weighted indicators were then aggregated to generate composite indices of seaport exposure and sensitivity to climate and extreme weather for 23 major ports in the North East United States. Rank order generated by AHP-weighted aggregation was compared to a subjective expert-ranking of ports by perceived vulnerability to climate and extreme weather. For the sample of 23 ports, the AHP-generated ranking matched three of the top four most vulnerable ports as assessed subjectively by portexperts. These results suggest that a composite index based on open-data may eventually prove useful as a data-driven tool for identifying outliers in terms of relative seaport vulnerabilities, however, improvements in the standardized reporting and sharing of port data will be required before such an indicator-based assessment method can prove decision-relevant.
Overall, this body of work began with a call to develop a method to assess the relative vulnerabilities of seaports to climate and extreme-weather impacts. In the first of three manuscripts, this research identifies an opportunity to contribute to the CCVA literature for the seaport sector by piloting a multi-port vulnerability assessment method based on the use of indicators. The second manuscript in this work contributes to the field of IBVA for seaports by identifying from open-data sources and refining via expert-elicitation methods a set of expert-evaluated candidate indicators of seaport climate and extreme-weather vulnerability. This indicator-evaluation resulted in the finding that adaptive capacity is considered by port-experts as the most difficult of the three components of vulnerability (i.e., exposure, sensitivity, and adaptive capacity) to represent with quantitative data. The final manuscript of this work contributes to the body of CCVA and seaport-studies literature by building and trialing a composite-index of seaport climate and extreme-weather vulnerability based on the evaluated indicators and using AHP to generate component weights. By modeling seaport vulnerability with an indicator-based composite index and comparing results to expert expectations, this work has shown the potential of indicator-based methods to bring a data-driven approach to the CIAV decision-making process, however, results suggest that the current state of publicly available data for and about the seaport sector is not currently sufficient for a robust, expert-supported index.
vii ACKNOWLEDGMENTS I have many people to thank for teaching me, believing in me, and offering support when I needed it. I am grateful to many people for helping to make this possible, especially those who gave their time participating in this research.
I would like to thank my advisor, Austin Becker, for taking me in as a new PhD student, providing me with opportunities to pursue interesting and interdisciplinary research, and continuously supporting me along the way. It has been a pleasure working alongside, travelling with, teaching with, and learning from him. He set an excellent example as a professor and advisor, and I am grateful to consider him a friend. I want to thank my mother, for without her support I would not be here. She supported me throughout this challenge in numerous ways, large and small. She made this possible. I am grateful to all my family for their support during this process.
Lastly, I thank my daughter, Galusina. She provided me with inspiration and support during these years more than she will ever know.
xiii Table 1 Indicators, categories and data sources used in  ............. 31 Table 2 Indicators, categories, and data sources used in (Hsieh, Tai, and Lee 2013) . 33 Table 3 Standardized indicators showing threshold values from (Hsieh, Tai Table 4 Results of port vulnerability analysis from (Hsieh, Tai, and Lee 2013) ......... 35 Table 5 Table 8 Expert-suggested candidate indicators of seaport vulnerability to climate and extreme weather impacts. While these suggested candidate indicators lacked the readily available data required to be included in the VAS survey, they may hold promise for further development provided data can be synthesized or compiled from

Seaports Are Critical, Constrained, and Exposed
Seaports represent an example of spatially defined, large scale, coast-dependent infrastructure with high exposure to projected impacts of global climate change (Becker et al. 2013, Melillo, Richmond, and Yohe 2014. Seaports play a critical role in the global economy, as more than 90% of global trade is carried by sea (IMO 2012). A disruption to port activities can interrupt supply chains, which can have far reaching consequences (Becker, Newell, et al. 2011, Becker et al. 2013. Seaports are inextricably linked with land-based sectors of transport and trade, and serve both the public and private good. Globally, climate change adaptation is still in the planning stages for most seaports (Becker, Inoue, et al. 2011), yet the inevitable imperative for climate resiliency looms, as atmospheric concentrations of greenhouse gasses, the primary driver of climate change (IPCC 2013), continue to accumulate (WMO 2015). Indeed, most aspects of climate change will persist for centuries even if anthropogenic emissions of carbon dioxide were halted today (IPCC 2013).
Functionally restricted to the water's edge, seaports will face impacts driven by changes in water-related parameters like mean sea level, wave height, salinity and acidity, tidal regime, and sedimentation rates, yet they can also be affected directly by changes in temperature, precipitation, wind, and storm frequency and intensity (Koppe, Schmidt, and Strotmann 2012). The third U.S. National Climate Assessment (NCA) (Melillo, Richmond, and Yohe 2014) of the U.S. Global Change Research Program notes that impacts from sea level rise (SLR), storm surge, extreme weather events, higher temperatures and heat waves, precipitation changes, and other climatic conditions are already affecting the reliability and capacity of the U.S. transportation system. While the U.S. NCA predicts that climate change impacts will increase the total costs to the nation's transportation systems, the report also finds that adaptive actions can reduce these impacts.
In the face of these challenges, port decision makers have the responsibility to manage risks for a diverse array of stakeholders and enhance seaport resilience against climate and weather impacts. At the single port scale, decision makers such as port managers may consider the uninterrupted functioning of their port the number one priority. But, at the multi-port (regional or national) scale, policy-makers will need to prioritize competing port climate-adaptation needs in order to maximize the efficiency of limited physical and financial resources and maximize the resilience of the marine transportation system as a whole. Recognizing a regional or national set of ports and waterways as part of an interconnected marine transportation system (MTS) 1 , how should responsible decision makers prioritize the climate adaptation decisions for systems that involve multiple ports? This chapter provides an overview of a variety of approaches that set out to quantify various aspects of seaport vulnerability. It begins with discussion of the importance of a "multi-port" approach to complement the single case study approach more commonly applied to port assessments. It then addresses the components of climate vulnerability assessments and provides examples of a variety of approaches.
Finally, it concludes with recommendations for next steps.

Impediments to Multi-Port Adaptation
A 2016 study which quantified the resources, time and cost of engineering minimum-criteria "hard" protections against sea level rise for 223 of the world's most economically important seaports, suggested insufficient global capacity for constructing the proposed protective structures within 50-60 years (Becker et al. 2016). As individual actors and governments consider climate-adaptation solutions for seaports, a global uncoordinated response involving heavy civil infrastructure construction may be unsustainable simply from a resource availability perspective (Becker et al. 2016, Becker, Newell, et al. 2011, Peduzzi 2014. Given limited financial and construction resources for the implementation of engineered protection across many ports, some form of prioritization for national and regional-scale climate-adaptation will likely be necessary. Port authorities have expressed that although general concern for climate change exists, awareness of sea level rise is limited and the planning for adaptation is lacking (Becker et al. 2010).
The implementation of strategic adaptation on a multi-port scale is further challenged by complex and dynamic regional differences defined by varying landscapes and geographies that are far from uniform in their climate change vulnerability. Some ports, for example, may by surrounded by lowlands at risk to inundation from sea level rise. For these ports, the ground transportation systems may by more threatened than the port itself (e.g., Port of Gulfport, MS). In other areas, storm surge might be amplified by the geomorphology of an estuarine system (e.g., Providence, RI).
At the single port scale, the design of engineering protection during a port's expansion can benefit by estimating how long the infrastructure will last and withstand future impacts (Becker, Toilliez, and Mitchell 2015). However, justifying major investments is challenged by the uncertainty involved in projecting the extent to which ports will be impacted this century (Becker and Caldwell 2015). In the following section, we first discuss the concept of measuring vulnerability, risk, and resilience, then describe assessment methods employed by individual ports. Following, we discuss the need for multi-port assessment approaches and work in this area to date.

Assessing Climate Vulnerabilities to Facilitate Far-Sighted Resilience Planning
Vulnerability and resilience are two theoretical concepts, sometimes defined complementarily, other times described as opposite sides of the same coin, , Linkov et al. 2014) that have gained increasing attention in the climate change adaptation and hazard risk reduction literature. As theoretical notions, resilience and vulnerability are not directly measurable, and some researchers (Barnett, Lambert, and Fry 2008, Eriksen and Kelly 2007, Klein 2009, Gudmundsson 2003 have criticized attempts to assess them as unscientific and or biased. However, policymakers are increasingly calling for the development of methods measure relative risk, vulnerability, and resilience . The International Association of Ports and Harbors (IAPH) defines seaport vulnerability using three components: exposure, sensitivity, and adaptation capacity (Koppe, Schmidt, and Strotmann 2012). Measuring a port's exposure requires downscaled regional climate projections which may not yet be available for some port regions, and where they are available, necessarily contain uncertainty. A port on the west coast of the U.S., for example, may be considered less exposed to hurricanes than a port on the east coast. Port exposure, then, may be analyzed using a multiple scenario approach, with a range of values for the applicable climate variables. Measuring port sensitivity and adaptation capacity generally requires site-specific analyses. By analyzing the impacts of projected changes in regional or even local climate variables and evaluating a port's design criteria in light of those impacts, the sensitivity to those changes can be determined for a port and its assets. Recently constructed infrastructure designed for higher intensity storms, for example, may be considered as less sensitive to a given storm event than infrastructure that is in a state of disrepair already. An assessment of a port's adaptive capacity, taking into account the port system's planning parameters, management flexibility and existing stresses, can reveal obstacles to a port system's ability to cope with climate change impacts. A port with robust planning procedures and more wealth, for example, may be considered to have a higher adaptive capacity than a port that has lesser planning and resources. In 2011, Becker and collaborators made a first attempt at quantifying international seaport adaptive capacity by developing a scoring system based on port authority responses regarding climate adaptation policies currently in place (Becker, Inoue, et al. 2011).
Because exposure and vulnerability are dynamic (IPCC 2012), varying across spatial and temporal scales, and individual ports are differentially vulnerable and exposed, assessments should be iterative with multiple feedbacks, shaped by people and knowledge (IPCC 2014a), and take a "bottom up" approach by including input from a diverse stakeholder cluster to ensure that the variables representing exposure, sensitivity and adaptive capacity are empirically identified by and important to the stakeholders, rather than presupposed by the researchers or available data .
A concept related to vulnerability, risk is a measure of the potential for consequences where something of value is at stake and where the outcome is uncertain (IPCC 2014b). Risk can be quantitatively modeled as Risk = p(L), where L is potential loss and p the probability of occurrence, however, both can be speculative and difficult to measure in the climate-risk context. Risk, in the context of climate change, is often defined similarly to vulnerability , but with the added component of probability, thus making vulnerability a component of risk.
Resilience, another closely related term with a more positive connotation than vulnerability, is defined by the IPCC as "the capacity of social, economic and environmental systems to cope with a hazardous event or trend or disturbance, responding or reorganizing in ways that maintain their essential function, identity and structure, while also maintaining the capacity for adaptation, learning and transformation" (IPCC 2014b). The National Academy of Science (The National Academies 2012) and the President of the United States (Obama 2013) define critical infrastructure resilience as, "the ability to prepare, resist, recover, and more successfully adapt to the impacts of adverse events." With resilience defined in terms of ability, and vulnerability defined in terms of susceptibility, it is tempting to consider them polar opposites , however, resilience can also be considered a broader concept than vulnerability. Most working definitions of resilience involve a process that begins before a hazardous impact, but also includes temporal periods during and after the impact. Resilience, like vulnerability, can also encompass coping with adverse effects from a multitude of hazards in addition to climate change. By increasing our understanding of the distribution of seaport climate vulnerabilities, the overall resilience of the MTS may be enhanced.

CIAV Decision-Support for the Seaport Sector
As port decision makers face climate impact, adaptation, and vulnerability (CIAV) 2 decisions, climate change vulnerability assessments (CCVA), including risk and resilience assessments support those decisions by addressing the "adapt to what" question (IPCC 2014a). The process enables a dialog among stakeholders and practitioners on planning and implementation of adaptation measures to enhance resilience. The Intergovernmental Panel on Climate Change (IPCC) describes vulnerability and risk assessment as "the first step for risk reduction, prevention, and transfer, as well as climate adaptation in the context of extremes." [p. 90] (IPCC 2012) The U.S. NCA considers vulnerability and risk assessment an "especially important" [p.
137] (Melillo, Richmond, and Yohe 2014) area in consideration of adaptation strategies in the transportation sector. Such assessments can be made at the single-port scale or at the multi-port scale, with each approach having benefits for different types of decision makers.

Single-Port Scale
Among climate change vulnerability, resilience, and risk assessment methods applied to seaports, most efforts to date have been limited in scope to exposure-only assessments , or limited in scale to a single port; either as case studies (Koppe, Schmidt, and Strotmann 2012, Cox, Panayotou, and Cornwell 2013, USDOT 2014, Chhetri et al. 2014 or as selfassessment tools , Morris and Sempier 2016. While single-port scale CCVA inform CIAV decisions within the domain of one port (e.g., Which specific adaptations are recommended for my port?), a CCVA approach that objectively compares the relative vulnerabilities of multiple ports in a region could support CIAV decisions at the multi-port scale (e.g., Which ports in a region are the most vulnerable and urgently in need of adaptation?). The hitherto focus on individual port scale assessments presents a challenge for how to describe the distribution of climate-vulnerabilities across multiple ports.

Multi-Port Scale
At the multi-port scale, an evaluation of relative climate-vulnerabilities or the distribution of those vulnerabilities among a regional or national set of ports requires standard measures (e.g. indicators, or metrics). Directly immeasurable, concepts such as resilience and vulnerability are instead made operational by mapping them to functions of observable variables called indicators. Indicators are measurable, observable quantities that serve as proxies for an aspect of a system that cannot itself be directly, adequately measured . Indicator-based assessment methods, therefore, are generally applied to assess or 'measure' features of a system that are described by theoretical concepts. The indicator-based assessment process of operationalizing immeasurable aspects of a system consists (Hinkel 2011) of two or sometimes three steps: 1) defining the response to be indicated, 2) selecting the indicators, 3) aggregating the indicators (this step is sometimes omitted but necessary to yield a numerical 'score' or create a comparative index). In this section, we investigate examples of indicator-based assessment methods applied to multi-port systems to aid the further development of such methods for the port sector, which can yield benefits including the ability to not only 'measure' immeasurable concepts like vulnerability and resilience, but also to index and compare them across entities.

Factors Considered in Port Resilience Evaluation
The US National Oceanic and Atmospheric Administration (NOAA) Office for Coastal Management (OCM) along with the federal interagency Committee on the Marine Transportation System (CMTS) produced a port resilience planning web-based tool (NOAA OCM 2015), tailored towards communities undergoing a port expansion or reconstruction, that assembles resilience indicators and their datasets. This web-based prototype tool came online in 2015 with the stated purpose of assisting transportation planners, port infrastructure planners, community planners, and hazard planners to explore resilience considerations and options in developing marine transportation projects. Inspired by and aligned with broader resilience objectives called for in the CMTS's strategic action plan (USCMTS 2011), this tool shows port communities what to look for in resilient freight transportation infrastructure. While the Port Tomorrow resilience planning tool assembles seaport resilience indicators, provides links to their potential data sources, and organizes them with categories and subcategories into a framework for assessing port resilience, the tool stops short of providing a method to normalize and aggregate the indicators into a comparative score.

Assessing Global Port City Exposure
One of the few CCVA to comparatively assess multiple ports, the 2010 work by Hanson, Nichols, et al. (Hanson et al. 2010) made some of the first progress towards comparative seaport CCVA by focusing on assessing the exposure component of seaport climate-vulnerability. Part of a larger project on Cities and Climate Change that was sponsored by the Organization for Economic Cooperation and Development (OECD), this global screening study assesses the exposure 3 of all 136 international port cities with over one million inhabitants in 2005 to coastal flooding. The analysis considers exposure to present-day extreme water levels (represented by a 100-year flood) as well as six future scenarios (represented by the decade 2070 -2080) that include projected changes in sea level and population. The researchers base the methods used on determining the numbers of people who would be exposed to the water level of interest and then using that number to estimate the potential assets exposed within each city. The researchers then rank the cities by number of people exposed and by 2005 U.S. dollar value of assets exposed. These two response variables, i.e. people and dollar value of assets, are semi-empirical quantities rather than theoretical concepts, and as such, the methods involved in this study are not directly analogous to other indicator-based assessment methods. Instead of using indicators to serve as proxies for some immeasurable concept, this study uses indicators to approximate concrete numbers that, due to scale, are difficult to measure.
This study took the form of a Geographic Information System (GIS) elevation-based analysis, after authors (McGranahan, Balk, and Anderson 2007). The researchers used 100-year historic flood levels taken from the Dynamic Interactive Vulnerability Assessment (DIVA) database as current extreme water levels to be modeled in GIS for each city. For the future water levels, the researchers calculate two different scenarios, one that considers only natural factors (i.e. a calculated "storm enhancement factor," historic subsidence rates, and sea level rise (SLR)), and another that adds to those factors one representing anthropogenic subsidence.
For current population, the study takes the ambient population distribution estimates from LandScan 2002 (Bright and Coleman 2003) for each city, delimited by city extents from post code data. The postcodes are taken from geocoding data and, for cities in the USA, from Metropolitan Statistical Areas (MSAs) from Census data. The authors resample the 1km LandScan 2002 data to 30m for all cities in the US and UK and resampled to 100m for the remaining cities. To determine population distribution by elevation, the authors use 90m resolution topographic data from the Shuttle Radar Topography Mission (SRTM) for most cities, 30m SRTM data for the US, and a 10m Digital Elevation Model (DEM) provided by Infoterra for the UK. The authors then overlay each LandScan population distribution over the relevant Digital Terrain Model (DTM), yielding for each city a map of geographical cells with defined population and elevation. From these maps, the authors are able to isolate total population within 1m vertical bands of elevation. To represent future population, the authors start with baseline population projections from the OECD ENV-Linkages model, which itself is based on United Nations (UN) medium variant projections to 2050. To bring these projections to 2070, the authors extrapolate them forward using national growth rates and UN projected rates of urbanization.
To indicate the dollar value of assets, the researchers use what they describe as a "widely used assumption in the insurance industry" (Hanson et al. 2010, 92) (p 92) that as urban areas are typically more affluent than rural areas, each person in a city has assets that are 5 times the national Gross Domestic Product (GDP). This simple calculation is based on the national per capita GDP Purchasing Power Parity (PPP) values for 2005 from the International Monetary Fund (IMF) database. To indicate future GDP, the study uses OECD baseline projections to 2075. To find the total value of assets exposed then, the researchers take the number of people exposed (from the GIS maps described above) and multiply that number by a country's GDP PPP times five.
Using the indicators described above, and organized in Table 1, this study is ultimately able to produce rankings of port cities exposed to coastal flooding by number of people and by dollar value of assets exposed to extreme water levels in 2005 and for projected extreme water levels in 2075.  Table 2. The process to determine weights for the indicators followed the analytic network process (ANP) of Jharkharia and Shankar (2007) (Jharkharia and Shankar 2007), and involved constructing an impact matrix via fuzzy cognitive maps (FCMs) developed and evaluated during these participatory meetings.
The impact matrix represents magnitudes of causal effects of each indicator compared to every other indicator.  The data for the indicators come from published statistics, literature, and GIS maps. Table 2 shows the specific data source for each of the 14 indicators. To score a port's vulnerability, the researchers standardize a port's raw indicator data using Table   3, then sum the standardized indicators multiplied by their weights to produce a total vulnerability score. The results for the 4 Taiwanese case study ports are show in Table   4. Table 4 Results of port vulnerability analysis from (Hsieh, Tai, and Lee 2013)  In addition to the vulnerability assessment method herein described, Hsieh et al.
also conducted an interdependency analysis to determine how strongly each indicator affects and is affected by the other indicators of the port system. This analysis uses groups of experts who fill out a matrix form during an iterative Delphi-style process, similar to that used during the first stages of this project.

Assessing Relative Port Performance
At the multi-port, MTS scale, CCVA have been sparse. Indicator-based multiport assessments to date have tended to focus on port performance rather than vulnerabilities or resilience. Here, we investigate some of the methods used to assess relative port performance in an effort to inform new CCVA methods at the multi-port scale.

Port Performance Indicators: Selection and Measurement (PPRISM)
Carried The results of the internal and external stakeholder assessments guided the final choice of 14 indicators that were then tested in a pilot phase. The 42 indicators were narrowed down to 14 (Table 5) through a process of weighing stakeholders' acceptance vs the feasibility of implementation of each indicator.
The pilot consisted of an EU-wide project to test the feasibility of the 14 selected indicators, with the intent to uncover the real-world availability of data and the willingness of port authorities to provide data. For the pilot study, the PPRISM group sent an electronic form to all port authorities associated with ESPO accompanied by an explanatory letter from ESPO Secretary General Patrick Verhoeven and received back a total of 58 forms fully or partially filled out. The pilot revealed problems with data availability, unclear data requests, and port participation. Given that data provision is voluntary, and hence, the number of ports submitting could fluctuate from year to year, the pilot study recommended that, at least for the initial stages of any port performance dashboard, reporting data in the form of trends rather than single values is the best approach. The results of the pilot study are shown in Table 5. facilities, and the authors conclude that more work is needed to capture the concept of port or MTS resilience using standard metrics. Table 6 compares the indicator selection and aggregation methods of the aforementioned indicator-based seaport assessments.

Discussion
To date, there are relatively few examples of multi-port assessments. The approaches discussed in this chapter, and summarized in Table 6, tend to lean heavily on expert judgement in the selection and evaluation for indicators of climate vulnerability or focus exclusively on the "exposure" aspect of vulnerability.
Worth note is the use of indicators to develop a score or rating of climate vulnerability (or resilience). Such assessment may be welcome or rejected, depending on the goals and objectives of the audience. For example, a high "vulnerability" score may help a port petition a funding agent to build a case for needed resilience investments. On the other hand, a high score could also leave a port at a competitive disadvantage if tenants perceive higher levels of storm risk. Thus, while aggregations, scores, and rankings may be desired by regional or national-level decision makers, creating multi-port assessment tools is not without controversy.

United States Army Corps of Engineers
That said, such tools can help inform the decision-making process. And, as demand for climate-critical resources (both funding and materials) increases, the need to better understand relative vulnerability of coastal systems, such as ports, will also increase. Our review of the literature suggests a need for better tools that can be used to gain an objective understanding of various aspects of port vulnerability. Although expert judgement will likely be necessary to a certain extent, due to the inherent difficulty of measuring and quantifying fuzzy concepts such as "adaptive capacity," publicly available data (e.g., historical storm tracks, types of cargo handled, throughput) can also be leveraged to help decision makers gain a better sense of which areas are more vulnerable, in what ways, and how this vulnerability might be reduced. Researchers developed a visual analogue scale (VAS) instrument to elicit expert perception of the magnitude and direction of correlation between candidate indicators and each of the three dimensions of vulnerability that have become standard in the CCVA literature, e.g., exposure, sensitivity, and adaptive capacity.
For candidate indicators selected from currently available open data sources, portexpert respondents found notably stronger correlation with the exposure and sensitivity of a port than with the adaptive capacity. Results suggest that more open reporting and sharing of port-specific data within the maritime transportation sector

Indicator-Based Assessments
Indicators are measurable, observable quantities that serve as proxies for an aspect of a system that cannot itself be directly, adequately measured . Indicatorbased assessment methods are generally applied to assess or 'measure' features of a system that are described by theoretical concepts. Directly immeasurable, concepts such as resilience and vulnerability are instead made operational by mapping them to functions of observable variables called indicators (McIntosh and Becker 2017). When comparing vulnerabilities of multiple disparate systems, indicator-based vulnerability assessment (IBVA) methods can yield standardized metrics, allowing for high-level analysis to identify areas or systems of concern. To advance IBVA for the seaport sector, researchers investigated the suitability of publicly available open-data, generally collected for other purposes, to serve as indicators of climate and extremeweather vulnerability for 23 major seaports in the North East United States, addressing the question: How sufficient is the current state of data reporting for and about the seaport sector to develop expert-supported vulnerability indicators for a regional sample of ports?
The indicator-based assessment process of operationalizing immeasurable aspects of a system  consists of two or sometimes three steps: 1) defining the response to be The indicator development process described in this work combines a deductive approach with a normative one. To develop indicators using an inductive argument would require a response variable (e.g., drop in revenue, port downtime, loss in throughput), that could allow for building statistical models to test for correlation with candidate indicators. Inductive arguments are generally only available when systems can be defined using only a few variables and sufficient data is available to serve as a response, or dependent variable, and this is rarely the case for the development of indicators of climate change vulnerability ). Hinkel argues that deductive arguments are only available for selecting indicators, not for aggregating them, and notes that deductive arguments are generally applied as a first step in indicator development.
Accordingly, the approach described in this paper begins with the application of a deductive argument to selecting indicators that is grounded in the framework established in the third assessment report of the IPCC (IPCC 2001), which defined climate change vulnerability in terms of three components: exposure, sensitivity, and adaptive capacity. In this research, an initial deductive approach to identifying candidate indicators is then followed by a normative one, where expert-elicitation is applied to seek expert consensus on the value judgements required to determine perceived correlation between the candidate indicators and the components of vulnerability taken from the deductive framework.
Expert-elicitation has become a common approach to applying a normative argument to of comparative CCVA, e.g., the elevation-based, exposure-only assessment of global port cities of , or have focused on assessing other concepts, e.g., (ESPO 2012) which aimed to measure port performance. While understanding how a port or a port-city's elevation affects its exposure to climate-impacts like SLR, it is only one piece of the puzzle that describes how a port is or is not vulnerable to climate and extreme weather impacts. By assessing the sensitivity and adaptive capacity of a port along with its exposure to a wide array of impacts in addition to SLR, a more complete picture of the mechanisms and drivers of seaport climate vulnerability may be better understood.

Why Seaports?
Seaports sit on the front lines of the climate-change challenge. Critical to national economies, global trade and national security, yet restricted to the hazardous land-sea interface, seaports face impacts from today's weather extremes as well as impacts from projected changes in temperature, precipitation, wind, storm frequency and intensity, mean sea level, wave height, salinity and acidity, tidal regime, and sedimentation rates (Koppe, Schmidt, and Strotmann 2012).
Among climate change vulnerability, resilience, and risk assessment methods applied to seaports, most efforts to date have been limited in scope to exposure-only assessments , Klein, Nicholls, and Thomalla 2003, or limited in scale to a single port (either as case studies (Koppe 2012, Cox, Panayotou, and Cornwell 2013, USDOT 2014, Chhetri et al. 2014  difficult. Climate impact, adaptation, and vulnerability (CIAV) 8 decisions at the multi-port (regional or national) scale may be supported by information products that allow decision makers to compare mechanisms and drivers of climate change among ports.
To advance the ability of seaport decision makers to compare levels of vulnerability among ports, and to further the development of IBVA for the seaport sector, this research investigates the suitability of publicly available open-data 9 to serve as indicators of climate and extreme-weather vulnerability for 23 major seaports in the North East United States ( Figure 3). This investigation seeks to examine the suitability of the current state of data reporting for and about the seaport sector to determine how sufficient it may or may not be to develop expert-supported vulnerability indicators for a regional sample of ports.

Vulnerability, Risk, and Resilience
This section describes several of the terms and concepts that are often used in discussions of the concepts of vulnerability, resilience, and risk. In the context of projected changes and current variability 10 in the earth's climate system, the meaning of the term vulnerability continues to evolve in the research literature Klein 2006, Smit and. In the third assessment report of the IPCC (IPCC 2001), vulnerability is defined in terms of susceptibility: Vulnerability is the degree to which a system is susceptible to, and unable to cope with, adverse effects of climate change, including climate variability and extremes.
Vulnerability is a function of the character, magnitude, and rate of climate change and variation to which a system is exposed, its sensitivity, and its adaptive capacity.  According to this definition, a system's vulnerability to climate change consists of external and internal dimensions. The external dimensions of vulnerability, i.e., the character, magnitude and 9 Open-data refers to publicly available data structured in a way that enables the data to be fully discoverable and usable by end users without having to pay fees or be unfairly restricted in its use. 10 Whereas climate change encompasses long-term (decades or longer) continuous changes to average weather conditions or to the range of weather, climate variability refers to yearly fluctuations above or below a long-term average.
rate of climate change, are commonly represented in the CCVA literature collectively as the exposure of the system in question, while the internal dimensions of vulnerability are represented by the system's sensitivity and adaptive capacity. (Clark andParson 2000, Turner et al. 2003). In its 2014 fifth assessment report, the IPCC simplified its definition of vulnerability to, "the propensity or predisposition to be adversely affected," [p. 5] (IPCC 2014a) however, the three components of vulnerability ( Figure 1) remain relevant. In a 2012 report on seaports and climate change, the International Association of Ports and Harbors 11 (IAPH) defines seaport vulnerability using the same three components, i.e., exposure, sensitivity, and adaptation capacity (Koppe 2012). By measuring vulnerability, then, this work aims to inform the measurement of the magnitude of a risk, but not it's probability.

Figure 2 Vulnerability as a component of risk
Resilience, another closely related term with a more positive connotation than vulnerability, is defined by the IPCC as "the capacity of social, economic and environmental systems to cope with a hazardous event or trend or disturbance, responding or reorganizing in ways vulnerability probability risk that maintain their essential function, identity and structure, while also maintaining the capacity for adaptation, learning and transformation" (IPCC 2014b). The National Academy of Sciences (The National Academies 2012) and the President of the United States (Obama 2013) define critical infrastructure resilience as, "the ability to prepare, resist, recover, and more successfully adapt to the impacts of adverse events." With resilience defined in terms of ability, and vulnerability defined in terms of susceptibility, it is tempting to consider them polar opposites , however, resilience can also be considered a broader concept than vulnerability.
Most working definitions of resilience involve a process that begins before a hazardous impact, but also includes temporal periods during and after the impact. Resilience, like vulnerability, can also encompass coping with adverse effects from a multitude of hazards in addition to climate change. While this research will further the development of indicators of seaport climate vulnerability, the objective is that by increasing our understanding of the regional distribution of seaport climate and extreme weather vulnerability, the overall resilience of the marine transportation system 12 (MTS) may be enhanced. 12 The Marine Transportation System, or MTS, consists of waterways, ports, and inter-modal land-side connections that allow the various modes of transportation to move people and goods to, from, and on the water. (MARAD 2016)

Methodology
To refine a set of high-level Rather than taking a purely theoretical approach to developing indicators, e.g., that used in the development of the Social Vulnerability Index (SoVI) (Cutter, Boruff, and Shirley 2003), this work takes a stakeholder-driven approach to indicator development by including port-experts in the selection, evaluation, and weighting of the indicators, as this has been shown to increase the creditability of the indicators as tools (Barnett, Lambert, andFry 2008, Sagar andNajam 1998). By including stakeholders in the design-stage of decision-support tool or boundary-object development, the 13 In visual analogue scale (VAS), respondents measure their level of agreement by indicating a position along a continuous line segment For evaluating candidate indicators of seaport vulnerability, this research was designed to take a holistic approach to vulnerability assessment by considering impacts that extend beyond the borders of the port property. To that end, this research in both the identification and evaluation of candidate indicators considered potential multimodal vulnerabilities at the port location as well as impacts to a port's surrounding community and economy (socio-economic systems) and ecological and environmental surroundings (environmental systems).
A VAS is a measurement instrument that tries to measure a characteristic or attitude that is believed to range across a continuum of values and cannot easily be directly measured. A VAS is usually a horizontal line, 100 mm in length, anchored by word descriptors at each end, as illustrated in Figure 6. The respondent marks on the line the point that they feel represents their perception of their current state. The VAS score is determined by measuring in millimeters from the left-hand end of the line to the point that the respondent marks. As a continuous, or analogue scale, the VAS is differentiated from discrete scales such as the Likert scale by the fact that a VAS contains a real distance measure, and as such, a wider range of statistical methods can be applied to the measurement.
The selection and evaluation of indicators involved four steps which will be described in the following sections: Step 1. Literature review to compile candidate indicators Step 2. Vetting for data availability Step 3. Mind mapping exercise Step 4.

VAS survey instrument
This research focuses on the thirteen medium-use 14 and nine high-use 15 ports found in the United States Army Corp of Engineers (USACE) North Atlantic Division 16 (CENAD) as the sample population for which to develop indicators ( Figure 3). The U.S. Army Engineer Research and Development Center (ERDC) has expressed  an interest in piloting port resilience and vulnerability assessment methods with high use ports, and by adding medium use ports and restricting the selection to the Northeast region researchers were able to create a manageable sample of 23 ports. Though this assessment was tailored to the US NE region, the framework was developed with the intent that it could be applicable (with modifications) to other regions.
Step 1 Step 2: Vetting for Data Availability reproducibility, and can enhance reliability when using standardized data sources (Janssen, Charalabidis, andZuiderwijk 2012, CMTS 2015). Only those indicators with data available for at least 16 of the study's sample of 23 ports were considered further. Table 9 shows the 108 candidate indicators of seaport climate-exposure, sensitivity, and adaptive capacity that were uncovered during this first step, as well as each indicator's preliminary categorization and its open data source. These candidate indicators include a mix of those that measure vulnerability of place at the county scale, à la the hazards-of-place model of vulnerability , e.g., population inside floodplain, and those that measure vulnerability via a characteristic of the port itself, e.g., containership capacity. For a comprehensive review of the data sources used, see (Mclean et al. 2017a). Of the 108 candidate indicators originally compiled, 48 (24 place-based and 24 port-specific) were found to have sufficient data available for the 23 sample ports.

Step 3: Mind Mapping Exercise to Refine the Set of Candidate Indicators
After compiling the 48 candidate indicators that were deemed to have sufficient data availability, researchers mapped them to the components of seaport climate vulnerability using the On the mind maps, each of the 48 candidate indicators with available data was hierarchically mapped to one of the three components of vulnerability, and for each indicator, the research team provided its description, data source, and units ( Figure 4).

Figure 4 Mind map legend showing how each indicator was hierarchically mapped to a component of vulnerability. The mind map also listed a description, data source, and units for each indicator.
During the mind mapping exercise, for each candidate indicator, experts from the USCMTS RIAT denoted with a plus or a minus whether an increase in that indicator correlates to an increase or decrease in the component of vulnerability it was mapped to, or with a zero if no correlation could be determined. In addition to evaluating the 48 candidate indicators with sufficient data availability, participants were also asked to brainstorm other potential data sources for those indicators without sufficient data and to add additional indicators that may have been overlooked.
The mind mapping exercise concluded with 14 candidate indicators marked as having no correlation to vulnerability, 25 marked as having positive correlation, and 9 candidate indicators marked as having negative correlation ( indicators and 20 place-based indicators. Table 7 lists the 34 selected candidate indicators alphabetically, along with their descriptions, units, and data sources. For a more comprehensive description of each of the 34 indicators, see (Mclean et al. 2017b). The RIAT participants suggested one additional candidate indicator, "age of infrastructure," however, they and the research team were unable to identify a data source that contains data on the age of infrastructure for the sample ports.

Pier.Depth
The greatest depth at chart datum alongside the respective wharf/pier. If there is more than one wharf/pier, then the one which has greatest usable depth is shown.

Shelter.Afforded
The shelter afforded from wind, sea, and swell, refers to the area where normal port operations are conducted, usually the wharf area.

Selection of Experts for VAS Survey
Because expert elicitation relies on expert knowledge rather than a statistical sample, the selection of qualified experts is considered one of most crucial steps in the process for insuring the internal validity of the research , Keeney, Hasson, and McKenna 2006. Candidates for the port-expert group were selected according to recommended best practices in expert selection developed by  and expanded by . Researchers first prepared a knowledge resource nomination worksheet (KRNW) (Table 10) modified from  to help categorize the experts prior to identifying them and to help avoid overlooking any important class of expert.
The KRNW was then populated with names, beginning with the professional network of the research team and that of the RIAT and identifying other candidate experts via a review of the relevant literature. This initial group of candidate experts was then contacted, provided a brief description of the study, queried for basic biographical information (e.g., number of papers published, length of practice, or number of years of tenure in government or NGO positions), and asked to nominate other candidate experts for inclusion on the list. Experts were asked to nominate peers with expertise in the fields of seaport operations, planning, policy, seaport data, and/or the vulnerability of the Northeast U.S. Marine Transportation System to climate and extreme weather impacts. This first round of contacts did not include invitations, but was aimed at extending the KRNW to ensure that it included as many experts as could be accessed. Upon completion of snowball sampling, researchers identified a total of 154 candidate experts to invite for participation in the VAS survey.
For this survey, 154 experts were invited and 64 participated, for a response rate of 42%.

Figure 5 Count of respondents' self-identified affiliations. Total n=64
Step 4: Expert-Elicitation VAS Survey The objective of this survey was to measure port-expert perceptions of the suitability of available data to serve as indicators of seaport vulnerabilities to climate and extreme weather impacts. The survey consisted of 34 candidate indicators to evaluate for correlation with the components of seaport vulnerability. For each candidate indicator, respondents were given the indicator's description, units, data source, and example values, and respondents were asked to determine whether the candidate indicator could be correlated with the exposure, sensitivity, and/or the adaptive capacity of ports in the study area. In evaluating candidate indicators, respondents were instructed to consider port vulnerability holistically, inclusive of the port's surrounding socioeconomic and environmental systems. Respondents indicated the magnitude and direction of correlation by dragging a slider along a VAS line segment ( Figure 6). To indicate "no correlation," respondents were to leave the slider in the center of the line. Dragging the slider to the left indicated a negative correlation and dragging the slider to the right indicated a positive correlation ( Figure 6). The distance measure of how far the slider was moved was indicative of the magnitude of perceived correlation. As a second check on the comprehensiveness of the set of candidate indicators, experts were also asked to suggest additional candidate indicators and data sources. While the initial search for candidate indicators was guided by the components (exposure, sensitivity, adaptive capacity) of vulnerability and subsequent sub-categories of those components specific to seaports, the VAS survey did not limit the candidate indicators to a single category or component of vulnerability. On the VAS survey, candidate indicators were presented with their metadata, but without assignment to a single component of vulnerability; instead, respondents denoted each indicator's correlation (or lack of correlation) with each of the three components of vulnerability ( Figure 6). This prevented respondents from inheriting the researchers' notions of correlation between candidate indicator and component of vulnerability. This feature also resulted in some indicators scoring high in correlation with more than one component of vulnerability.

Results
For each of the 34 candidate indicators evaluated, Figure 7 shows the median expertperceived magnitude of correlation with each of the three components of vulnerability, stacked, in descending order of correlation. To reduce the effect of outliers on the measure of central tendency, this work considers the median rather than the mean of responses when aggregating scores for each candidate indicator. Interestingly, respondents reserved their highest levels of aggregate perceived correlation for place-based indicators; though 14 of the 34 candidate indicators were port-specific, the top 12 candidate indicators ranked by total correlation were all place-based ( Figure 7). Also of note in Figure 7 is the low level of perceived correlation with adaptive capacity (pink) compared to exposure (green) and sensitivity (blue). The indicator with the highest median expert-perceived correlation was the same for all three components of vulnerability, i.e., population inside floodplain. The indicator, sea level trend also scored high, rated second highest in median correlation with exposure and sensitivity, and fourth highest with adaptive capacity. In Figure 7, the highest scoring port-specific indicator (bold) was tide range, followed by shelter afforded, both metrics available from the World Port Index (NGIA 2015).
The following three figures illustrate the median expert-percieved magnitude of correlation seperately for each component of vulnerability, revealing expert preferences for the most suitable candidate indicators to represent each concept for the sample set of CENAD ports. Figure 8, Figure   9, and Figure 10 show the top 15 scoring indicators in descending order for correlation with exposure, sensitivity, and adaptive capacity, respectively.
In Figure 8, the ten indicators with the highest median perceived correlation with port exposure were all place based. The port-specific indicator rated highest perceived correlation with exposure was tide range, ranked 11/34, followed by harbor size, ranked 14/34. In Figure 9, the top 13 indicators with the highest median perceived correlation with port sensitivity were all place based. As was the case with exposure, the two highest scoring indicators for correlation with sensitivity were also population inside floodplain, and sea level trend, respectively. The port-specific indicator rated highest perceived correlation with sensitivity was also the same as that for exposure, i.e., tide range, ranked 14/34, followed by containership capacity, ranked 15/34. While the top ten scoring indicators for correlation with exposure and sensitivity were all place-based, the same was not true for adaptive capacity. For correlation with adaptive capacity ( Figure 10), port-specific indicators scored relatively high. The port-specific indicator rated highest perceived correlation with adaptive capacity was shelter afforded, ranked 3/34, followed by entrance restrictions, ranked 8/34, harbor size, ranked 9/34, tide range, ranked 10/34, marine transportation GDP, ranked 12/34, and channel depth, ranked 13/34.
Although the distance measure of the VAS sliders is unitless, the results indicate an overall low level of expert-perceived correlation between candidate indicators and seaports' adaptive capacity ( Figure 10), significantly lower than that for exposure ( Figure 8) and sensitivity ( Figure   9). The highest scoring candidate indicator for adaptive capacity, population inside floodplain, only scored 23 on the unitless VAS, which is lower than 16 th place for exposure and lower than 17 th place for sensitivity. Interestingly, although candidate indicators scored generally low with adaptive capacity, port-specific indicators fared much better with adaptive capacity than with the other two components of vulnerability, with 4 of the top ten indicators in Figure 10 representing port-specific indicators. Figure 10 Top 15 candidate indicators for adaptive capacity, sorted by median expert-perceived magnitude of correlation with seaport adaptive capacity to climate and extreme weather impacts. Port-specific candidate indicators in bold. Overall, experts found significantly lower correlation with adaptive capacity than with the other two components of vulnerability.
Because the VAS expert group was disproportionately represented by those with Federal affiliations (Figure 5), the median aggregate group response considered in the previous four figures is necessarily dominated by those experts. Further insights may be gained by filtering results by expert type, revealing differences in the perceptions of the differently affiliated experts. For example, academically affiliated experts ( Figure 12) found more and higher levels of correlation with adaptive capacity than did other types of expert (Figures 11, 13-15). This may be due to academically affiliated experts having more familiarity with the concept of adaptive capacity than other types of expert, as adaptive capacity has become a more common subject in the academic literature.
Asked to suggest additional candidate indicators, respondent experts suggested seven indicators ( Table 8) that may warrant further development but were not sufficiently supported by data for our study area ports to be included in this study. As this study aimed to evaluate the current state of openly-available data, candidate indicators required an identifiable open data source with data coverage for greater than 75% of the ports in the CENAD sample to be immediately applicable to this work. Some of the suggested indicators that currently lack sufficient data coverage could potentially be synthesized from a combination of other available data sources, derived via geographic information systems (GIS), or compiled via additional computation for evaluation in future studies. For example, robustness of transportation infrastructure, measured in terms of the number of back-up routes, may be determinable via GIS analysis of each ports' multimodal connections' elevations, however, such indicators will be highly sensitive to the value-judgement of how to delimit each port. Port interdependencies also present potential for inclusion in indicator development, e.g., the suggested indicator distance to nearest alternative seaport, which would capture the availability of backup ports available to handle a port's primary cargo should that port experience downtime. Though not presently identifiable in openly available data sources, such an indicator could be synthesized from data records of port cargo types, with a similar caveat that it will also require the value judgement of what qualifies as an "alternative" port in terms of ability to handle similar cargo. short of comparative CCVA, e.g., the elevation-based, exposure-only assessment of global port cities of , or have focused on assessing other concepts, e.g., (ESPO 2012) which aimed to measure port performance. This research builds upon this body of literature by contributing a set of 34 expert-evaluated indicators of seaport climate vulnerability that can be monitored to assess relative vulnerabilities across ports. and extreme weather impacts. It also suggests that the concept of adaptive capacity is considered by port-experts to be more difficult to represent with quantitative data than the concepts of exposure or sensitivity.

Expert Preference for Place-Based Indicators
Results of the VAS survey also indicate that respondents reserve their highest levels of aggregate perceived correlation for place-based indicators; though 14 of the 34 candidate indicators were port-specific, the top 12 candidate indicators ranked by total correlation were all place-based. While port-specific indicators scored low overall, they fared better with adaptive capacity than with exposure or sensitivity, which suggests that more or different port-specific data reporting may lead to improvements in the ability to measure a port's relative adaptive capacity.
While the 34 candidate indicators encompassed a combination of 14 port-specific indicators (i.e., those that capture a specific aspect of the port) and 20 place-based indicators (i.e., those that capture the hazards-of-place at the county scale), respondents found higher levels of correlation with the components of vulnerability for place-based indicators than for portspecific ones. For both correlation with exposure ( Figure 8) and with sensitivity (Figure 9), the ten highest rated candidate indicators were all place-based. For correlation with adaptive capacity, however, while noticeably lower in magnitude, four of the top ten indicators were portspecific, and a port-specific indicator scored second highest overall ( Figure 10). This suggests that of the 34 candidate indicators evaluated, respondents generally preferred the place-based indicators for representing the exposure and sensitivity of a seaport but preferred a mixture of place-based and port-specific indicators for representing a port's adaptive capacity.
This finding suggests that while adaptive capacity is considered by port experts the most difficult component of seaport climate vulnerability to quantify, if expert-supported indicators of seaport adaptive capacity are to be developed, they will most likely be developed from portspecific data, rather than place-based data. As the current selection of port-specific data openly available for the CENAD sample of ports was found to have little expert-perceived correlation with the components of seaport climate vulnerability, efforts will have to be made to identify and share additional port-specific data that can better capture these concepts, and adaptive capacity in particular.

Variation of Results for Different Expert-Affiliation Groups
Filtering responses by expert affiliation revealed differences in the perceptions of the different types of expert. Academically affiliated experts were more willing to indicate correlation with adaptive capacity than other types of expert, while federally affiliated experts indicated the least amount of correlation with adaptive capacity. This discrepancy may reveal a higher familiarity with adaptive capacity as an abstract concept in the academic sphere than in other portexpert professions. This finding highlights the importance of a diverse expert group when using expert-elicitation methods.

Limitations and Next Steps
As the population of experts with the requisite knowledge of the climate vulnerabilities of N.E. U.S. seaports is limited, this study was limited by the sample size of respondent experts. While the total response rate was satisfactory, the total number of experts was not evenly distributed among the seven expert-affiliation categories ( Figure 5). Accordingly, comparisons of responses by expert-affiliation suffer from this small sample size. These expertrelated limitations are a function of applying a stakeholder-driven approach, as opposed to a purely data-drive approach, e.g., SoVI (Cutter, Boruff, and Shirley 2003). Instead of the purely theoretical approach described by the SoVI, this work takes a stakeholder-driven approach by including portexperts in the development and weighting of the indicators, as this has been shown to increase the creditability of the index as a tool (Barnett, Lambert, andFry 2008, Sagar andNajam 1998).
An additional limitation stems from the difficulty of achieving true comprehensiveness in the process of seeking and compiling the candidate indicators for experts to evaluate. To lessen the risk of excluding potential candidate indicators, researchers asked experts, at both the mind map stage and the VAS survey stage, to suggest additional or better indicators. At neither stage were experts able to suggest an indicator with a known data source with sufficient data availability for the sample of ports, suggesting that our search for open-data candidate indicators was suitably comprehensive. Next steps for future studies may involve furthering the development of those candidate indicators suggested by respondents in Table 8, exploring non-open or proprietary sources of data for those indicators identified in Table 9 but lacking available open data sources, or synthesizing novel indicators from combinations of available data.

Conclusion
Seaports are critical to global trade and national security yet sit on the front-line for extreme coastal weather and climate impacts, and such impacts are projected to worsen globally. As port decision-makers wrestle with the myriad of climate adaptation options (including the option of making no adaptations at all), their CIAV decisions can and should be supported with data. For CIAV decision-support, the first step often involves assessing vulnerabilities. For an individual seaport, this process tends to take the shape of CCVA, either as a participatory self-assessment, or as a site-specific case study. For multiple port systems, however, CCVA often rely on indicators.

Abstract
This paper describes a method of weighting indicators for assessing the exposure and sensitivity of seaports to climate and extreme weather impacts. Researchers employed the Analytic Hierarchy Method (AHP) to generate weights for a subset of expert-selected indicators of seaport exposure and sensitivity to climate and extreme weather. The indicators were selected from the results of a previous survey of port-experts who ranked candidate indicators by magnitude of perceived correlation with the three components of vulnerability; exposure, sensitivity, and adaptive capacity. As those port-expert respondents found significantly stronger correlation between candidate indicators and the exposure and sensitivity of a port than with a port's adaptive capacity, this AHP exercise did not include indicators of adaptive capacity. The weighted indicators were then aggregated to generate composite indices of seaport exposure and sensitivity to climate and extreme weather for 23 major ports in the North East United States. Rank order generated by AHPweighted aggregation was compared to a subjective expert-ranking of ports by perceived vulnerability to climate and extreme weather. For the sample of 23 ports, the AHP-generated ranking matched three of the top four most vulnerable ports as assessed subjectively by port-experts.
These results suggest that a composite index based on open-data may eventually prove useful as a data-driven tool for identifying outliers in terms of relative seaport vulnerabilities, however, improvements in the standardized reporting and sharing of port data will be required before such an indicator-based assessment method can prove decision-relevant.

Key Findings:
• Experts weighted adaptive capacity higher than sensitivity and nearly equal with exposure in terms of importance to seaport climate vulnerability, yet, adaptive capacity lacks expertsupported indicators.
• Prototype composite-index matched the most vulnerable port and three of the top four most vulnerable ports as subjectively ranked by port experts.
• An indicator-based composite-index approach, weighted via AHP shows promise for identifying relative outliers among a sample of ports in terms of vulnerability.

Seaport Vulnerability to Climate and Extreme Weather
Seaports sit on the frontlines of our shores, consigned to battle the elements at the hazardous intersection of land and sea. Ports face projected increases in the frequency and severity of impacts driven by changes in water-related parameters like mean sea level, wave height, salinity and acidity, tidal regime, and sedimentation rates, and port functions are expected to be increasingly affected directly by changes in temperature, precipitation, wind, and storm frequency and intensity (Koppe, Schmidt, andStrotmann 2012, Becker et al. 2013). At the same time, ports are often located in environmentally sensitive ecosystems such as estuaries and river mouths, which provide important nursery habitat for juvenile marine organisms (Beck et al. 2001).
As infrastructure assets, ports are critical to both the public and the private good, playing a key role in the network of both intranational and international supply-chains.
Ports serve as catalysts of economic growth locally and regionally, as they create jobs and promote the expansion of nearby industries and cities (Asariotis, Benamara, and Mohos-Naray 2017).
Port decision-makers have a responsibility to manage a multitude of risks and enhance port resilience to achieve the minimum downtime safely possible in any given circumstance. When regional systems of ports are considered, responsible decisionmakers may wish to prioritize limited resources, or to identify outliers among a set of ports in terms of vulnerability to certain hazards. At the single-port scale, port decisionmakers (e.g., a local port authority) may be questioning which specific adaptation actions to take, or where to start with climate-adaptation. At the multi-port scale, port decision-makers (e.g., the U.S. Army Corps of Engineers) may be questioning which ports in a certain jurisdiction are the most vulnerable and hence the most in need of urgent attention. As climate adaptation decisions often involve conflicting priorities (e.g., politics, national priorities, local priorities), providing a data-driven, standard metric can help bring objectivity into the process.

Indicator-Based Composite Indices
Indicators are measurable, observable quantities that serve as proxies for an aspect of a system that cannot itself be directly, adequately measured ). Indicator-based assessment methods are generally applied to assess or 'measure' features of a system that are described by theoretical concepts. Directly weights. For the research described in this paper, the SoVI recipe was considered, but deemed to be unsuitable for ports as the small sample size and the sparseness of available data (compared to Census data) led to difficulty in identifying and naming the principal components. Instead of the purely theoretical approach described by the SoVI, this work takes a stakeholder-driven approach by including port-experts in the development and weighting of the indicators, as this has been shown to increase the creditability of the index as a tool (Barnett, Lambert, andFry 2008, Sagar andNajam 1998). By including stakeholders in the design-stage of decision-support tool or boundary-object development, the stakeholders' perceptions of the credibility, salience, and legitimacy of the tool can be increased (White et al. 2010).
Indicator-based assessments and indices have provoked debate in the literature, and some researchers (Barnett, Lambert, and Fry 2008, Eriksen and Kelly 2007, Klein 2009, Gudmundsson 2003 have criticized attempts to assess theoretical concepts with them as lacking scientific rigor or lacking consistency. Nonetheless, policymakers are increasingly calling for the development of methods to measure relative risk, vulnerability, and resilience , and developing better indicators and expert-driven weighting schemes through participatory processes like AHP may lead to improvements in this field. Despite these criticisms of indicator-based vulnerability assessments (IBVA) and indicator-based composite indices in particular, such decision-support tools can play an important role in bringing objective data into the complex decision-making process. The use of such indicator-based decision-support products can provide guidance in identifying areas of concern, but they should always be supplemented with additional expertise as they lack the high-resolution found in more detailed case-study assessment approaches.
Whereas low-level, high-resolution analyses are better served by more comprehensive case-study approaches, e.g., (Hallegatte et al. 2011, McLaughlin, Murrell, and DesRoches 2011, USDOT 2014, indicator-based composite indices are well suited to provide high-level overviews of relative outliers among a sample.
Indicator-based assessments and indices, then, are simply one tool among a suite of tools that decision-makers should have at their disposal.
Port decision-makers faced with climate impact, adaptation and vulnerability (CIAV) 19 decisions involving multiple ports can benefit from information products that allow them to compare the mechanisms and drivers of vulnerability among ports.
Indicator-based assessments provide an example of such a product that can quantify complex issues and bring a standardized data-driven approach to measuring theoretical concepts, with the caveat that the decision-relevance of their results hinges on the quality of data available to serve as indicators.

Selection of Indicators
This paper describes the process of deriving weights for previously North Atlantic Division 23 (CENAD). The steps involved in compiling and evaluating this set of candidate indicators is also illustrated in Figure 18, below. 20 The degree to which a system is susceptible to, and unable to cope with, adverse effects of climate change, including climate variability and extremes. Vulnerability is a function of the character, magnitude, and rate of climate change and variation to which a system is exposed, its sensitivity, and its adaptive capacity. (IPCC 2001) 21 Medium use here refers to ports with annual throughput > 1M tons 22 High use here refers to ports with annual throughput > 10M tons 23 The North Atlantic Division is one of nine USACE divisions and encompasses the U.S. Eastern Seaboard from Virginia to Maine (USACE 2014). and example values, and respondents were asked to determine whether the candidate indicator could be correlated with the exposure 27 , sensitivity 28 , and/or the adaptive capacity 29 of ports in the study area. Respondents indicated the magnitude and direction of correlation by dragging a slider along a VAS line segment ( Figure 6). In addition to evaluating 34 indicators of seaport vulnerability, respondents of the VAS survey also subjectively ranked the CENAD ports by magnitude of perceived vulnerability to climate and extreme weather impacts.

Figure 19 VAS slider for indicating expert-perceived correlation between a candidate indicator and each of the components of vulnerability.
For the 34 candidate indicators that were evaluated, none scored a median rating higher than 23 on the unitless VAS scale of correlation with adaptive capacity, compared to a high of 62 with exposure and 52 with sensitivity. This low level of perceived correlation with adaptive capacity suggests a dearth of open-data 30 sources suitable for representing the adaptive capacity of seaports to climate and extreme weather impacts. It also suggests that the concept of adaptive capacity is considered by port-experts to be more difficult to represent with quantitative data than the concepts of 27 The presence of people, livelihoods, species or ecosystems, environmental functions, services, and resources, infrastructure, or economic, social, or cultural assets in places and settings that could be adversely affected (IPCC 2014b) 28 The degree to which a system is affected, either adversely or beneficially, by climate-related stimuli (IPCC 2001) 29 The ability of systems, institutions, humans and other organisms to adjust to potential damage, to take advantage of opportunities, or to respond to consequences (IPCC 2014b) 30 Open-data refers to publicly available data structured in a way that enables the data to be fully discoverable and usable by end users without having to pay fees or be unfairly restricted in its use. exposure or sensitivity. For these reasons, this AHP exercise did not include indicators of adaptive capacity but focused instead on generating weights for indicators of exposure and sensitivity.
As AHP best-practice recommends each category should have at least 4, but not more than 7 to 10 sub-categories (Goepel 2013), researchers selected the 6 highest scoring indicators for exposure and the 6 highest scoring indicators for sensitivity for inclusion in the AHP exercise (Table 11) described in the following section.

Analytic Hierarchy Process
The Analytic Hierarchy Process (AHP) is a method to support multi-criteria decision-making first described by Thomas Saaty (Saaty 1977) that is based on the solution of an eigenvalue problem. Participants make pairwise comparisons, the results of which are arranged in a matrix where the dominant normalized right eigenvector gives the ratio scale (weighting) and the eigenvalue determines the consistency ratio (Goepel 2013, Saaty 1977, 1990b. AHP has become well established for group decisions based on the aggregation of individual judgements (Ramanathan and Ganesh 1994, Dedeke 2013, Goepel 2013. Psychologists have noted that respondents have an easier time making judgements on a pair of alternatives at a time than simultaneously on all the alternatives (Ishizaka and Labib 2011). AHP also allows consistency cross checking between the pairwise comparisons. Additionally, AHP uses a ratio scale, which, unlike methods using interval scales, does not require units in the comparison (Kainulainen et al. 2009, Hovanov, Kolari, andSokolov 2008).
AHP has also proven useful as a standardized method for generating the weights of indicators in composite indices in a variety of different fields, e.g., environmental performance index (EPI) (Dedeke 2013), disaster-resilience index (Orencio and Fujii 2013), composite indicator of agricultural sustainability (Gómez-Limón and Riesgo 2009), and the urban public transport system quality (Pticina and Yatskiv 2015). While these studies assessed different theoretical concepts from performance, to disasterresilience, to agricultural sustainability, they all employed AHP as a means of quantifying expert-preferences for weighting the relative importance of the indicators used. AHP simplifies the process of quantifying subjective weight preferences based on multiple criteria by using pairwise comparisons. Participants are given two items at a time and asked which is more important with respect to the given category. Using pairwise comparisons not only helps discover and correct logical inconsistencies (Goepel 2013), it also allows for translating subjective opinions into numeric relations, helping make group decisions more rational, transparent, and understandable (Goepel 2013, Saaty 2008.

Expert Selection
Researchers  For the first level of the AHP, respondents weighted the three components of seaport vulnerability via pairwise comparisons. Respondents were given two components at a time and asked, "With respect to seaport climate vulnerability, which criterion is more important, and how much more on a scale 1 to 9," where '1' represents equal importance (Figure 22).

Figure 21 AHP hierarchy showing equal weighting prior to pairwise comparisons. Each column represents a level of the AHP, and each red rectangle indicates a node (for which a priority vector will be calculated).
The second level of the AHP involved two nodes; weighting six indicators of exposure, and weighting six indicators of sensitivity. For the former, respondents were given two indicators at a time and asked, "With respect to seaport climate exposure, which criterion is more important, and how much more on a scale 1 to 9." For calculating the number of pairwise comparisons required, Equation 1 is used where n is the number of components or indicators (Saaty 1977, 1990a, Orencio and Fujii 2013.
Where aij is the preference for indicator Ii over Ij when both were compared pairwise, for i, j = 1, 2, … n. If a respondent decided that indicator i was equally important to another indicator j, a comparison of aij = aji = 1 was recorded. If a respondent considered indicator i extremely more important than indicator j, the preference-matrix score was based on aij = 9 and its reciprocal given as aji = 1/9, where aij > 0.
After compiling a preference matrix for each expert for each node of the AHP, the dominant eigenvector of each matrix was then calculated using the power method (Larson 2016, Goepel 2013

Equation 4 Consolidated preference matrix based on the geometric mean of individual judgements
To measure the consensus for the aggregated group result, the AHP software used Shannon entropy and its partitioning in two independent components (alpha and beta diversity) to derive an AHP consensus indicator based on relative homogeneity S (Goepel 2013). The consensus of the complete hierarchy was calculated as the weighted arithmetic mean of the consensus of all hierarchy nodes. This similarity measure, S, is zero when the priorities of all pwc are completely distinct and S=1, when the priorities of all pwc are identical (Goepel 2013).

Aggregating Weighted Indicators
After generating the indicator and component weights via AHP, the next step was to create a composite index of seaport vulnerability based on the weightings. Due to the lack of expert-supported indicators of adaptive capacity, the AHP-based composite index was limited to the aggregation of two of the three components of vulnerability: exposure and sensitivity, yielding a composite score that may be

AHP-Generated Weights
The aggregation of judgements from the first level of the AHP, which weighted the three components of seaport vulnerability to climate and extreme weather, resulted in exposure ranked most important, with a ratio scale (weight) of .394 (Table 12).
Adaptive capacity was ranked a close second, with a weight of .390, which is noteworthy since the component of adaptive capacity lacks expert-supported indicators.
Sensitivity was ranked least important of the three components, with a weight of .216.
For this node, the maximum consistency ratio, CR, was 0.1% (highly consistent) and the group consensus, S, was 50.1% (low) 32 . The second level of the AHP consisted of two nodes, the first evaluated six indicators for relative importance in terms of seaport exposure to climate and weather extremes, and the second node evaluated six indicators in terms of seaport sensitivity.
The first node resulted in the indicator "number of disasters," ranked most important for the component of exposure with a weight of .200, and resulted in weights for the remaining indicators of exposure as shown in Table 13. For this node, the maximum consistency ratio, CR, was 0.3% (highly consistent) and the group consensus, S, was 53.6% (low). The second node of the second AHP level resulted in the indicator "population inside floodplain," ranked most important for the component of sensitivity with a weight of .229, and resulted in the remaining indicators of sensitivity weighted as shown in Table  14. For this node, the maximum consistency ratio, CR, was 0.5% (highly consistent) and the group consensus, S, was 61.1% (low).

Composite Index of CENAD Ports
To test the degree to which a ranking of ports by level of vulnerability to climate and extreme weather, created by a WSM using AHP-generated weights, would or would not resemble an a priori ranking generated 33 subjectively by the same participating experts, researchers compiled a composite index for the CENAD sample of ports. Applying the AHP-generated indicator weights to the z-score-standardized input variables for 23 CENAD ports, and aggregating them in a WSM yielded the following ranking (Table 15) where a larger number corresponds to a higher degree of 33 As part of the VAS survey described in the second chapter of this dissertation, port-experts were asked to rank the top ten most vulnerable ports out of the sample of 22 CENAD ports. The rank distribution (Table 16) was generated from a sum of weighted values, which were weighted as the inverse of the number of ports the respondent chose to rank.
vulnerability. In Table 15, a score of zero represents the mean, a negative number represents a vulnerability score below the mean, and a positive number represents a vulnerability score above the mean. Interestingly, the most vulnerable port according to the model-generated port vulnerability rankings matches the most vulnerable port as subjectively ranked by experts in the VAS survey (Table 16). While the second most vulnerable port according to the subjective expert-ranking, the Port of New York and New Jersey, was second to least vulnerable according to the model rank, the model did capture three out of four of the most vulnerable ports consistent with the experts' rankings. One benefit of indicator-based composite indices is their ability to synthesize multiple variables into a single, measurable concept while still retaining the ability to explore the disaggregated substructure behind the composite construct. As such, their users are able to ask, "Why does a particular entity score high or low according to this index?" Figure 23 shows the disaggregated substructure behind the composite 'vulnerability scores' of the three highest scoring ports from the composite index, in which the relative performance of a port can be explored in terms of the individual indicators. Similarly, Figure 24 shows the disaggregated substructure for the three lowest scoring ports of the composite index.    "number of critical habitat areas," "hundred year high water," and "number of cyclones," it scored near the bottom of the sample for "number of disasters," "number of storm events," and "environmental sensitivity index ESI," and did not score higher than average for any indicator.

Discussion
The method of generating indicator weights based on aggregated expertpreferences using AHP described in this paper has shown both promise and limitations.
Port rankings generated by a composite index based on a WSM using the AHP-derived weights, was compared to an a priori subjective ranking generated by port experts.
Though the model lacked indicators of adaptive capacity, it matched (Table 15) the experts' ranking for the most vulnerable port, and also matched three of the four ports ranked most vulnerable by the experts (Table 16).
Whereas previous work on assessing the climate vulnerability of seaports has tended to focus on the single port scale, either as case studies (Koppe, Schmidt, and Strotmann 2012, Cox, Panayotou, and Cornwell 2013, USDOT 2014, Chhetri et al. 2014  To the observed problem (i.e., the current difficulty of comparing relative vulnerability across ports), this work contributes a prototype composite-index (and a method to replicate such an index for other sectors) that allows rudimentary quantitative comparisons of exposure and sensitivity levels across ports. This prototype index was able to capture relative outliers in the sample of ports (i.e., the main objective of composite-indices) and shows the promise of an indicator-based approach to address this problem.

Adaptive Capacity Considered Highly Important
Adaptive capacity is defined in the glossary of the IPCC Fifth Assessment Report (IPCC 2014b) as ''The ability of systems, institutions, humans and other organisms to adjust to potential damage, to take advantage of opportunities, or to respond to consequences." As noted by Siders (Siders 2016), this definition bears some resemblance to generally accepted definitions of resilience, i.e., the ability to bounce back from an impact (McIntosh and Becker 2017). As such, Siders recommends that adaptive capacity can be distinguished from resilience by ascribing the latter to maintaining stability by "bouncing back" to pre-shock conditions, and by taking adaptive capacity, to refer to the broader ability of a system to self-organize, learn, and embrace change to limit future harms (Klein, Nicholls, andThomalla 2003, Siders 2016).
It may be significant that the AHP resulted in adaptive capacity ranked a close second to exposure in terms of importance with respect to seaport climate and extreme weather vulnerability (Table 12). This suggests that port-experts consider adaptive capacity to be more important than sensitivity and practically equal in importance to exposure with respect to seaport vulnerability. Though experts place a high degree of importance on adaptive capacity as a component of vulnerability, a previous study (see the second manuscript of this dissertation) found that adaptive capacity may be the most difficult of the three components of seaport vulnerability to represent with quantitative data. While this discrepancy may point to a need to improve the data collection and sharing of metrics that can capture the concept of adaptive capacity for ports, it also suggests that the concept of adaptive capacity may be better captured by other, less quantitative assessment methods. This finding also suggests a disconnect between what experts perceive as an important component to understanding seaport vulnerability to meteorological and climatological threats and the types of data that are currently being reported and available to represent that component.
As noted by Brooks et al. (Brooks, Adger, and Kelly 2005), adaptive capacity is a component of vulnerability primarily associated with governance. Hence, next-step efforts to assess relative levels of seaport adaptive capacity should start by examining ports' governance structures to find measurable metrics to assess and compare the ports' ability to adjust, take advantage, or respond to climate and weather impacts.

Limitations
A limitation of this AHP method can be the difficulty of achieving high levels of group consensus. For each of the three nodes of this AHP, the consensus indicator, S, was low (50.1%, 53.6%, 61.1%), suggesting low relative homogeneity of expert preferences. Improvements in group consensus may be achieved by using iterative approaches such as the Delphi 34 method, in which participants are shown descriptive 34 The Delphi method is a structured communication technique designed to obtain opinion consensus of a group of experts by subjecting them to a series of questionnaires interspersed with feedback in the form of a statistical representation of the group response. The goal of employing the Delphi method is to reduce the range of responses and arrive at something closer to expert consensus. statistics of the group responses and given the opportunity to revise their answers during subsequent iterations of the AHP, as was employed in (Orencio and Fujii 2013). A drawback of this iterative approach, however, is the additional time required to complete the process. For this study, researchers held 20 different webinars with a total of 34 experts to complete the AHP, lasting approximately 30 minutes to one hour each webinar. Experts may be more reluctant to participate the longer the process proposes to take. As the number of pairwise comparisons increases quickly due to Equation 1, even a single-round AHP can become a considerable imposition on the time constraints of busy professional experts.
Though the aggregation of weighted indicators into a composite index was performed mainly as a means to validate the AHP-generated weights by comparing the port-rankings they produced via a WSM to a subjective port-ranking, the process also yielded insight into the benefits and limitations of such methods. As a means to identify relative outliers among a sample, this method showed promise by successfully matching the most vulnerable port and three of the four most vulnerable ports as ranked subjectively by port-experts. While partially successful at identifying the relative outliers among our sample of ports, the composite index also ranked several ports (e.g., Providence, New York and New Jersey) near the bottom of the sample that experts had subjectively ranked near the top. Some of this discrepancy may be due to the sensitivity of indicator-based composite indices to differences in the interpretation of data used for the indicators. For example, an indicator for an entity that spans multiple counties, like the port of New York and New Jersey, could be represented by a measure of central tendency of the data for the collection of counties, by the data from the county with most extreme value, or by a single representative county. In this experiment, the single county of New York was taken to represent the port of New York and New Jersey for the purposes of compiling the indicator data, which may have resulted in lower than expected values for that port in some of the indicators. Additionally, indicator-based assessments are always limited by the quality of data available to incorporate into them.
Although the AHP weighted all three components of vulnerability, including adaptive capacity, and the composite index incorporated the weights for the components of exposure and sensitivity into the WSM, it should be noted that this composite index of seaport vulnerability to climate and extreme-weather did not include indicators of adaptive capacity. As such, the composite index is more accurately described as a weighted measure of seaport exposure and sensitivity to climate and weather extremes.
This may have also contributed to some of the discrepancy between model results and the subjective ranking of ports (see the second manuscript of this dissertation) which was based on a definition of vulnerability that included all three components (e.g., exposure, sensitivity, adaptive capacity).
Additionally, indicator-based methods are inherently limited by the availability of data. The second manuscript of this dissertation, which describes the identification, development, and evaluation of candidate indicators of seaport climate vulnerability, illustrates these data availability limitations in more detail. For example, the lack of openly available data to serve as indicators of adaptive capacity resulted in the reduction of the composite index described here from an assessment of holistic vulnerability to one of exposure and sensitivity only.

Conclusion
To further the development of indicator-based assessment methods for the port sector, this study performed an AHP with 37 port-experts that developed weights for the three components of vulnerability (i.e., exposure, sensitivity, and adaptive capacity), and for a selection of 12 indicators of seaport exposure and sensitivity to climate and extreme weather impacts. The AHP resulted in adaptive capacity weighted higher than sensitivity and nearly equal to exposure in importance with respect to seaport climate and extreme weather vulnerability. This finding suggests a disconnect between what experts believe is an important component to understanding seaport vulnerability to meteorological and climatological threats and the types of data that are currently being reported and available to represent that component. An opportunity for future research may exist to develop an answer to what types of data, if any, experts would accept as more representative of the concept of seaport adaptive capacity than what data is currently available.
To validate the results of the AHP, the AHP-generated weighting scheme was applied using a WSM to create a composite index for 23 CENAD ports that was compared to a subjective ranking of the ports by the same experts. This comparison revealed that while the model showed promise in fulfilling the main objective of composite indices (i.e., identification of relative outliers among a sample) by matching the top port and three out of the top four ports subjectively chosen as most vulnerable by the experts, there were considerable discrepancies between the model rank and the subjective, expert rank that point to some of the limitations of this method. Those limitations include the potential for low group consensus during the AHP, for which the remedy, Delphi-style iterations, contains its own limitation of increased time-cost.
Indicator-based methods are also limited by their sensitivity to small changes in the methods used to compile the individual indicators. Variations in spatial scale of available data can require subjective choices regarding the compilation of indicator data, e.g., how to compile indicator data for ports that span multiple counties. Additionally, the process of compiling indicators introduces other subjective decisions that affect model sensitivity, such as whether to use the max value or a measure of central tendency of a concept as an indicator. Because of both the sensitivity and subjectivity of these decisions, researchers recommend a stakeholder-based approach for the early stages of indicator development such as the expert-elicitation methods applied in (Mcleod et al. 2015, Teck et al. 2010. While this research has furthered the development of indicatorbased assessment methods for the port sector by constructing and trialing a prototype composite-index of seaport climate vulnerability, it should be noted that further work exploring the sensitivity of results to data compilation methods and developing a measure of adaptive capacity will be needed before such methods are robust enough for use in critical decision-making. Finally, the main caveat of these methods is that they are always limited by the quality of the data that they incorporate.

Comprehensive Conclusion
This work began with a call to develop a method to assess the relative vulnerabilities of seaports to climate and extreme-weather impacts. In the first of three manuscripts, this research identified an opportunity to contribute to the CCVA literature for the seaport sector by piloting a multi-port vulnerability assessment representing the adaptive capacity of seaports in the sample. This finding also suggests that port-experts consider the concept of adaptive capacity to be less amenable to representation with quantitative data than the remaining two components of vulnerability, i.e., exposure and sensitivity.
Results of the VAS survey also indicate that respondents reserve their highest levels of aggregate perceived correlation for place-based indicators; though 14 of the 34 candidate indicators were port-specific, the top 12 candidate indicators ranked by total correlation were all place-based. While port-specific indicators scored low overall, they fared better with adaptive capacity than with exposure or sensitivity, which suggests that more or different port-specific data reporting may lead to improvements in the ability to measure a port's relative adaptive capacity.
After evaluating candidate indicators, researchers then constructed and trialed a prototype composite-index of seaport climate vulnerability using the highest scoring indicators from the VAS survey. The objective of this experiment was to investigate the ability of a data-driven composite-index approach to measure relative climate vulnerability for a sample of ports. Interestingly, during the AHP part of the index construction, respondents weighted adaptive capacity higher than sensitivity and nearly equal with exposure in terms of importance to seaport climate vulnerability.
This finding is noteworthy because the previous VAS survey found a lack of expert-support for candidate indicators of adaptive capacity. This suggests a disconnect between those concepts experts consider important to capture when measuring vulnerability and what data is available to measure those concepts.
Finally, results of the prototype index were compared to experts' subjective port rankings to evaluate how well the model captured The results of the prototype composite-index are highly sensitive to value-judgements such as how to delimit each port (e.g., Where should the boundary be? Which terminal to include?) or how to compile indicator data (e.g., Use max value or average value? Take the value for the highest county or the average of counties when ports span multiple counties?) Additionally, the reproducibility of the expert-elicitation processes will necessarily be limited by expert subjectivity. A further limitation of the prototype composite-index stems from its lack of indicators of adaptive capacity.
To the observed problem (i.e., the current difficulty of comparing relative vulnerability across ports), this body of research contributes a set of 34 expertevaluated indicators that can be monitored to assess relative vulnerabilities across ports. This work also contributes a prototype composite-index (and a method to replicate such an index for other sectors) that allows rudimentary quantitative comparisons of exposure and sensitivity levels across ports. This prototype index was able to capture relative outliers in the sample of ports (i.e., the main objective of composite-indices) and shows the promise of an indicator-based approach to address this problem. Step 4: Rank experts

Results
• Create four sub-lists, one for each discipline • Categorize experts according to appropriate list • Rank experts within each list based on their qualifications Step 5: Invite experts • Invite experts for each panel, with the panels corresponding to each discipline • Invite experts in the order of their ranking within their discipline sub-list • Target size is 10-18 • Stop soliciting experts when each panel size is reached Electronic Consent: Please select a choice below. Clicking on the "Agree" button indicates that You have read the above information You voluntarily agree to participate * ( ) Agree -Enter Survey ( ) Disagree -Exit Affiliation Please select the category that best describes your professional affiliation:* ( ) Consultant ( ) Academic ( ) (Port / Marine Transportation System) Practitioner ( ) Federal Government ( ) State Government ( ) Non-governmental Organization ( ) Other -Please Specify: _________________________________________________* Instructions Please consider whether this candidate indicator, (Measurable, observable quantity that serves as a proxy for an aspect of a system that cannot itself be directly, adequately measured [page("title")]), could be correlated (The condition of being interdependent; a mutual relation of two or more things such that a change in the value of one is associated with a change in the value or the expectation of the others) with one or more of the three components of climate vulnerability (The degree to which a system is susceptible to, and unable to cope with, adverse effects of climate change, including climate variability and extremes. Vulnerability is a function of the character, magnitude, and rate of climate change and variation to which a system is exposed, its sensitivity, and its adaptive capacity): Exposure: The presence of people, livelihoods, species or ecosystems, environmental functions, services, and resources, infrastructure, or economic, social, or cultural assets in places and settings that could be adversely affected Example: a port on the US East coast has a higher exposure to hurricanes than a port on the US West Coast; independent of the ports' sensitivity to damage Sensitivity: The degree to which a system is affected, either adversely or beneficially, by climate-related stimuli Example: a port with a storm surge barrier may be less sensitive to storm driven flooding impacts than a similar port without a storm surge barrier; independent of the ports' exposure and/or the Adaptive Capacity: The ability of systems, institutions, humans and other organisms to adjust to potential damage, to take advantage of opportunities, or to respond to consequences Example: a port with a robust master plan that considers climate resilience and has a high degree of operational flexibility may have a higher adaptive capacity than a port with minimal planning and low redundancy; independent of the ports' exposure and sensitivity of a port, including the port's surrounding socioeconomic and environmental systems.
. Large, Medium, Small, Very Small

Description
The classification of harbor size is based on several applicable factors, including: area, facilities, and wharf space. It is not based on area alone or on any other single factor.

Data Source
The National Geospatial-Intelligence Agency (NGA) World Port Index (Pub 150) contains the location and physical characteristics of, and the facilities and services offered by major ports and terminals world-wide (approximately 3700 entries

Description
1% annual exceedance probability high water level which corresponds to the level that would be exceeded one time per century, for the nearest NOAA tide station to the port

Data Source
NOAA Extreme Water Levels Extremely high or low water levels at coastal locations are an important public concern and a factor in coastal hazard assessment, navigational safety, and ecosystem management. Exceedance probability, the likelihood that water levels will exceed a given elevation, is based on a statistical analysis of historic values.

Description
1% annual exceedance probability low water level for the nearest NOAA tide station to the port, which corresponds to the level that would be exceeded one time per century.

Data Source
NOAA Extreme Water Levels Extremely high or low water levels at coastal locations are an important public concern and a factor in coastal hazard assessment, navigational safety, and ecosystem management. Exceedance probability, the likelihood that water levels will exceed a given elevation, is based on a statistical analysis of historic values.

NOAA Tides and Currents-Sea Level Trends
The Center for Operational Oceanographic Products and Services has been measuring sea level for over 150 years, with tide stations of the National Water Level Observation Network operating on all U.S. coasts. Changes in Mean Sea Level (MSL), either a sea level rise or sea level fall, have been computed at 142 long-term water level stations using a minimum span of 30 years of observations at each location. These measurements have been averaged by month to remove the effect of higher frequency phenomena in order to compute an accurate linear sea level trend.
Tide stations measure Local Sea Level, which refers to the height of the water as measured along the coast relative to a specific point on land. Water level measurements at tide stations are referenced to stable vertical points (or bench marks) on the land and a known relationship is established. However, the measurements at any given tide station include both global sea level rise and vertical land motion, such as subsidence, glacial rebound, or large-scale tectonic motion. Because the heights of both the land and the water are changing, the land-water interface can vary spatially and temporally and must be defined over time. Depending on the rates of vertical land motion relative to changes in sea level, observed local sea level trends may differ greatly from the average rate of global sea level rise, and vary widely from one location to the next.
Relative Sea Level Trends reflect changes in local sea level over time and are typically the most critical sea level trend for many coastal applications, including coastal mapping, marine boundary delineation, coastal zone management, coastal engineering, sustainable habitat restoration design, and the general public enjoying their favorite beach. This website focuses on relative sea level trends, computed from monthly averages of hourly water levels observed at specific tide stations, called monthly mean sea level.

Description
The percent change from observed baseline of the average number of days per year above baseline "Extremely Hot" temperature projected for the end-of-century, downscaled to 12km resolution for the port location.

Description
The percent change from observed baseline of the average number of "Extremely Heavy" Precipitation Events projected for the end-ofcentury, downscaled to 12km resolution for the port location. For most pollutants in the index, the concentration is converted into index values between 0 and 500, "normalized" so that an index value of 100 represents the short-term, health-based standard for that pollutant as established by EPA (U.S. EPA, 1999). Each person in physical possession of a hazardous material at the time that any of the following incidents occurs during transportation (including loading, unloading, and temporary storage) must submit a Hazardous Materials Incident Report on DOT Form F 5800.1 (01/2004) within 30 days of discovery of the incident: An unintentional release of a hazardous material or the discharge of any quantity of hazardous waste; A specification cargo tank with a capacity of 1,000 gallons or greater containing any hazardous material suffers structural damage to the lading retention system or damage that requires repair to a system intended to protect the lading retention system, even if there is no release of hazardous material; An undeclared hazardous material is discovered; or A fire, violent rupture, explosion or dangerous evolution of heat (i.e., an amount of heat sufficient to be dangerous to packaging or personal safety to include charring of packaging, melting of packaging, scorching of packaging, or other evidence) occurs as a direct result of a battery or battery-powered device.

"Extremely Heavy" Precipitation Events
Hazardous materials in various forms can cause death, serious injury, long-lasting health effects and damage to buildings, homes and other property. Many products containing hazardous chemicals are used and stored in homes routinely. These products are also shipped daily on the nation's highways, railroads, waterways and pipelines. "Structurally deficient" means that the condition of the bridge includes a significant defect, which often means that speed or weight limits must be put on the bridge to ensure safety; a structural evaluation of 4 or lower qualifies a bridge as "structurally deficient". The designation can also apply if the approaches flood regularly.
"Functionally obsolete" means that the design of a bridge is not suitable for its current use, such as lack of safety shoulders or the inability to handle current traffic volume, speed, size, or weight. Excellent (5), Good (4), Fair (3), Poor (2), None (1) Description Shelter afforded from wind, sea, and swell

Data Source
The National Geospatial-Intelligence Agency (NGA) World Port Index (Pub 150) contains the location and physical characteristics of, and the facilities and services offered by major ports and terminals world-wide (approximately 3700 entries).

Description
Entrance Restrictions are natural factors restricting the entrance of vessels, such as ice, heavy swell, etc.

Data Source
The National Geospatial-Intelligence Agency (NGA) World Port Index (Pub 150) contains the location and physical characteristics of, and the facilities and services offered by major ports and terminals world-wide (approximately 3700 entries).

Description
Overhead Limitations: indicates that bridge and overhead power cables exist.

Data Source
The National Geospatial-Intelligence Agency (NGA) World Port Index (Pub 150) contains the location and physical characteristics of, and the facilities and services offered by major ports and terminals world-wide (approximately 3700 entries). A (over 76 ft) to Q (0 -5 ft) in 5-foot increments

Description
The controlling depth of the principal or deepest channel at chart datum

Data Source
The National Geospatial-Intelligence Agency (NGA) World Port Index (Pub 150) contains the location and physical characteristics of, and the facilities and services offered by major ports and terminals world-wide (approximately 3700 entries).
Depth information is generalized into 5-foot units, with the equivalents in meters, for the main channel, the main anchorage, and the principal cargo pier and/or oil terminal.
Depths refer to chart datum. Depths are given in increments of 5 feet (1.5 meters) in order to lessen the number of changes when a small change in depth occurs.
A depth of 31 feet (9.5 meters) would use letter "K," a depth of 36 feet (11.0 meters) would use "J," etc. The letter "K" means a least depth of 31 feet (9.5 meters) or greater, but not as great as 36 feet (11.0 meters).
CHANNEL (controlling)-The controlling depth of the principal or deepest channel at chart datum is given. The channel selected should lead up to the anchorage if within the harbor or to the wharf/pier. If the channel depth decreases from the anchorage to the wharf/pier and cargo can be worked at the anchorage, then the depth leading to the anchorage is taken.
Large ports may have sub-ports (smaller) which have their own number and entry in the World Port Index. The controlling depth of the channel should refer to a smaller channel (if present) leading from the main channel into the sub-port facilities and anchorages.
Note.-The depth of small shoals is not a controlling depth unless it limits the passage of vessels. For example, if a channel is charted as having a depth of 39 feet (11.9 meters), but there are small shoals noted or charted with depths of 30 feet (9.1 meters), then the controlling depth is still 39 feet (11.9 meters) unless a ship with a draft of 39 feet (12 meters

Description
The greatest depth at chart datum alongside the respective wharf/pier. If there is more than one wharf/pier, then the one which has greatest usable depth is shown.

Data Source
The National Geospatial-Intelligence Agency (NGA) World Port Index (Pub 150) contains the location and physical characteristics of, and the facilities and services offered by major ports and terminals world-wide (approximately 3700 entries).
Depth information is generalized into 5-foot units, with the equivalents in meters, for the main channel, the main anchorage, and the principal cargo pier and/or oil terminal.
Depths refer to chart datum. Depths are given in increments of 5 feet (1.5 meters) in order to lessen the number of changes when a small change in depth occurs.
A depth of 31 feet (9.5 meters) would use letter "K," a depth of 36 feet (11.0 meters) would use "J," etc. The letter "K" means a least depth of 31 feet (9.5 meters) or greater, but not as great as 36 feet (11.0 meters).
CARGO PIER/WHARF-The greatest depth at chart datum alongside the respective wharf/pier is given. If there is more than one wharf/pier, then the one which has greatest usable depth is shown. For example, if there are three cargo/container piers with depths of 23 feet (7.0 meters), 33 feet ( The mean tidal range at the port

Data Source
The National Geospatial-Intelligence Agency (NGA) World Port Index (Pub 150) contains the location and physical characteristics of, and the facilities and services offered by major ports and terminals world-wide (approximately 3700 entries).

Data Source
The NOAA Office for Coastal Management: Economics: National Ocean Watch (ENOW) ENOW Explorer contains annual time-series data for over 400 coastal counties, 30 coastal states, 8 regions, and the nation, derived from the Bureau of Labor Statistics and the Bureau of Economic Analysis. It describes six economic sectors that depend on the oceans and Great Lakes and measures four economic indicators: Establishments, Employment, Wages, and Gross Domestic Product (GDP).

MARINE TRANSPORTATION
Includes deep sea freight, marine passenger transportation, pipeline transportation, marine transportation services, search and navigation equipment, and warehousing.

Data Source
The NOAA Office for Coastal Management: Quick Report Tool for Socioeconomic Data provides easy access to economic and demographic data for multiple coastal jurisdictions.
Information is derived from several key socioeconomic sources, including the U.S. Census Bureau, Bureau of Economic Analysis, Bureau of Labor Statistics, and Federal Emergency Management Agency's Hazus database.
In 2010, 123.3 million people, or 39 percent of the nation's population lived in Coastal Shoreline Counties. Population growth in these counties occurred at a lower rate than the nation as a whole from 1970 to 2010. The population in Coastal Shoreline Counties increased by 34.8 million people, a 39 percent increase, while the nation's entire population increased by 52 percent over the same time period.
Within the limited space of the nation's coast, population density far exceeds the nation as a whole, and this trend will continue into the future. This situation presents coastal managers with the challenge of protecting both coastal ecosystems from a growing population and protecting a growing population from coastal hazards.
The concentration of people impacts the integrity of coastal ecosystems, and at the same time, the lives and livelihoods of some of these residents and visitors can be at risk from natural processes at the coastsuch as hurricanes, erosion, and sea level rise.

People + Floodplains = Not Good
The more homes and people located in a floodplain, the greater the potential for harm from flooding. Impacts are likely to be even greater when additional risk factors (age, income, capabilities) are involved, since people at greatest flood risk may have difficulty evacuating or taking action to reduce potential damage.

Units
The SoVI® Social Vulnerability score is classified using standard deviations. Social vulnerability scores that are greater than 2 standard deviations above the mean are considered the most socially vulnerable, and scores below 2 standard deviations less than the mean are the least vulnerable. Description The SoVI® Social Vulnerability score of the port county

Social Vulnerability
The hazards-of-place model  combines the biophysical vulnerability (physical characteristics of hazards and environment) and social vulnerability to determine an overall place vulnerability. Social vulnerability is represented as the social, economic, demographic, and housing characteristics that influence a community's ability to respond to, cope with, recover from, and adapt to environmental hazards.

The Social Vulnerability Index (SoVI®)
County-level socioeconomic and demographic data were used to construct an index of social vulnerability to environmental hazards, called the Social Vulnerability Index (SoVI®) for the United States After obtaining the relevant data, a principle components analysis is used to reduce the data into set of components. Slight adjustments are made to the components to ensure that the sign of the component loadings coincide with each individual population characteristic's influence on vulnerability. All components are added together to determine a numerical value that represents the social vulnerability for each county. The U.S. DOT Maritime Administration: Vessel Calls in U.S. Ports, Selected Terminals and Lightering Areas is a report containing a calculation of vessel calls for privately-owned, oceangoing merchant vessels of all flags of registries over 1,000 gross tons (GT) calling at ports and selected ports/terminals within the contiguous United States, Hawaii, Alaska, Guam and Puerto Rico.
Calls are calculated by how many times a vessel arrived at a port, facility or terminal. This number may include berth shifts, movement to and from an anchorage while awaiting cargo and may also include other activities related to vessel, port or terminal operations. Calls do not include vessels arriving at a designated anchorage area. In addition, vessels calling on a port may not necessary be engaged in onloading/offloading of cargoes.
Capacity is calculated as the sum of vessel calls weighted by vessel deadweight (DWT). DWT is defined as the total weight (metric tons) of cargo, fuel, fresh water, stores and crew which a ship can carry when immersed to its load line.

Data Source
The U.S. DOT Maritime Administration: Vessel Calls in U.S. Ports, Selected Terminals and Lightering Areas is a report containing a calculation of vessel calls for privately-owned, oceangoing merchant vessels of all flags of registries over 1,000 gross tons (GT) calling at ports and selected ports/terminals within the contiguous United States, Hawaii, Alaska, Guam and Puerto Rico.
Calls are calculated by how many times a vessel arrived at a port, facility or terminal. This number may include berth shifts, movement to and from an anchorage while awaiting cargo and may also include other activities related to vessel, port or terminal operations. Calls do not include vessels arriving at a designated anchorage area. In addition, vessels calling on a port may not necessary be engaged in onloading/offloading of cargoes.
Capacity is calculated as the sum of vessel calls weighted by vessel deadweight (DWT). DWT is defined as the total weight (metric tons) of cargo, fuel, fresh water, stores and crew which a ship can carry when immersed to its load line. Calls are calculated by how many times a vessel arrived at a port, facility or terminal. This number may include berth shifts, movement to and from an anchorage while awaiting cargo and may also include other activities related to vessel, port or terminal operations. Calls do not include vessels arriving at a designated anchorage area. In addition, vessels calling on a port may not necessary be engaged in onloading/offloading of cargoes.
Capacity is calculated as the sum of vessel calls weighted by vessel deadweight (DWT). DWT is defined as the total weight (metric tons) of cargo, fuel, fresh water, stores and crew which a ship can carry when immersed to its load line.

Description
Annual containership capacity at the port Containership -Container Ship and Passenger/Container Ships

Data Source
The U.S. DOT Maritime Administration: Vessel Calls in U.S. Ports, Selected Terminals and Lightering Areas is a report containing a calculation of vessel calls for privately-owned, oceangoing merchant vessels of all flags of registries over 1,000 gross tons (GT) calling at ports and selected ports/terminals within the contiguous United States, Hawaii, Alaska, Guam and Puerto Rico.
Calls are calculated by how many times a vessel arrived at a port, facility or terminal. This number may include berth shifts, movement to and from an anchorage while awaiting cargo and may also include other activities related to vessel, port or terminal operations. Calls do not include vessels arriving at a designated anchorage area. In addition, vessels calling on a port may not necessary be engaged in onloading/offloading of cargoes.
Capacity is calculated as the sum of vessel calls weighted by vessel deadweight (DWT). DWT is defined as the total weight (metric tons) of cargo, fuel, fresh water, stores and crew which a ship can carry when immersed to its load line. Comments (Please also explain any extreme views):: climate change, including climate variability and extremes. Vulnerability is a function of the character, magnitude, and rate of climate change and variation to which a system is exposed, its sensitivity, and its adaptive capacity. Where are the lowest levels of climate vulnerabilityThe degree to which a system is susceptible to, and unable to cope with, adverse effects of climate change, including climate variability and extremes. Vulnerability is a function of the character, magnitude, and rate of climate change and variation to which a system is exposed, its sensitivity, and its adaptive capacity among the principal ports of the USACEUnited States Army Corps of Engineers North Atlantic Division?
Based on your present knowledge and opinion, Please select from the following list and arrange the 5 LEAST VULNERABLE ports in ascending order from lowest to highest level of relative climate vulnerabilityThe degree to which a system is susceptible to, and unable to cope with, adverse effects of climate change, including climate variability and extremes. Vulnerability is a function of the character, magnitude, and rate of climate change and variation to which a system is exposed, its sensitivity, and its adaptive capacity. Can you suggest additional candidate indicators (Measurable, observable quantities that serve as proxies for an aspect of a system that cannot itself be directly, adequately measured) of seaport climate vulnerability (The degree to which a system is susceptible to, and unable to cope with, adverse effects of climate change, including climate variability and extremes. Vulnerability is a function of the character, magnitude, and rate of climate change and variation to which a system is exposed, its sensitivity, and its adaptive capacity)?* ( ) Yes, I have additional candidate indicators to suggest. ( ) No, I have no indicators to suggest.