Keywords

1 Introduction

Due to the recently expanding aging population in the U.S., providing transportation accessibility for aging people has become a growing concern. As such, aging people’s transportation accessibility to critical emergency facilities such as hurricane shelters represents a serious challenge. This challenge is exacerbated by their cognitive, physical, mental, emotional and health limitations that can affect their driving and route finding skills, especially in the aftermath of extreme events such as hurricanes. This problem becomes even more complex for the State of Florida, which is ranked as second highest nationally for having 17.34 % of the population being 65 years or older [1]. One strategy to address these challenges involves the use of GIS-based tools. These tools can be used to assess the performance of the transportation network in conjunction with the spatial distributions of aging people (65+) and critical emergency facilities.

Transportation accessibility has been studied with a focus on different facilities in the literature; from urban and commercial facilities such as grocery stores and parks, to hospitals and assisted living facilities. Among these studies, accessibility to urban parks for the general populations has been studied in [2, 3] whereas sports centers, schools and older population-oriented markets were investigated in order to assess their accessibility to the public in [4]. GIS has also been used extensively in order to measure the accessibility of universities [5], health facilities [6], multi-modal facilities [7], and supermarkets [8, 9]. Several researchers have recognized the need of studying accessibility with respect to aging populations. In [10], the accessibility of libraries, parks and hospitals was assessed for both pedestrians and drivers in Florida based on the spatial structure of social, professional, and business opportunities. In another study, the accessibility of nursing facilities was studied in the aftermath of the Los Angeles Northridge Earthquake in [11]. Several previous studies analyzed the mobility and transportation needs of the aging population [12–16]. Please see [7] for a more detailed review on the transportation accessibility of aging populations to the multi-modal facilities, and see [17] for a more detailed discussion on the emergency accessibility and mobility needs of aging populations. To the authors’ knowledge, there are no studies that focus on the transportation accessibility of emergency shelters with a focus on aging populations, and that use dynamic travel times to estimate potential accessibility.

This study utilizes a Geographical Information Systems (GIS)-based methodology, first developed in [7], to measure and assess the transportation accessibility of critical emergency facilities such as shelters through a diverse set of case study applications in the State of Florida. Within this evaluation, spatially detailed county-based accessibility scores are calculated with respect to designated hurricane shelters (regular shelters as well as special needs shelters –SPNS- are included) using both static and dynamic travel times between population block groups and critical facilities. This is followed by a small scale case study application presented for the Panama City area, using Bay County of Florida. This application focuses on the accessibility of critical facilities such as airports, bus stations and shelters that can be critical in the aftermath of an extreme event such as a hurricane. Because aging of the Baby Boom generation is expected to produce a 79 % increase in the number of people over the age of 65 in the next two decades, the proposed methodology can inform transportation agencies’ efforts to develop efficient aging-focused transportation and accessibility strategies.

2 Methodology

This sections describes the steps of our GIS-based methodology (Please refer to [7] for more detail). The first step focuses on the spatial distributions of aging populations living in the State of Florida using available resources such as U.S. Census [1] (Fig. 1). As of 2010, the State of Florida consists of 3,259,602 people over 65 years (approximately 17.33 % of the total population), which makes about one in every six Floridian 65 years and older. Following this evaluation, the following data components are determined (Fig. 2): (a) origins (the geometric centroids of the U.S. Census population block groups), (b) destinations (the critical emergency facility locations including the American Red Cross (ARC)-approved regular and special needs (SPNS) shelters), and (c) the roadway network of Florida, with 127,343 arcs and 53,333 nodes, based on the Florida Standard Urban Transportation Model Structure (FSUTMS) of the Florida Department of Transportation (FDOT) [18]. Locations of the ARC-approved regular and SPNS shelters are obtained from the 2014 Florida Statewide Emergency Shelter Plan [19].

Fig. 1.
figure 1

Population demographics of Florida (a) 65+ population, (b) 65+ population percentage, (c) Population block group centroids (origins)

Full size image
Fig. 2.
figure 2

Destinations (a) American Red Cross (ARC)-approved regular shelters, (b) American Red Cross (ARC)-approved SPNS shelters

Full size image

Between each population block group centroid and the closest facility of interest, identified by ArcGIS, the least cost path is determined. That is, each origin-destination (O-D) pair is assigned a travel cost in terms of different values: (a) roadway distance in miles, (b) free flow travel time in minutes, which represents the uncongested traffic conditions, (c) static congested travel time in minutes, and (d) dynamic congested travel time in minutes, which incorporates the effects of queues and delays (this is only performed for the second case study). Using the first three cost figures, emergency shelter transportation accessibility scores for each population block group are calculated for the whole State of Florida. Using these scores, aging population-weighted average accessibility scores of counties to ARC-approved regular and SPNS shelters are calculated. Since it would be almost impossible to obtain the dynamic travel times for the whole state due to computational requirements, a smaller region, namely Panama City, Bay County of Florida, is selected to compare the static and dynamic traffic assignment results for the accessibility of the emergency facilities to aging populations.

3 Case Study Results

In this section, we present results of our case study application in Florida. First, aging-focused accessibility maps are illustrated for each county with respect to different types of shelters (regular and special needs). These maps are used to identify critical counties that are relatively further away from the shelters. The second case study focuses on a small scale application in order to compare the static and dynamic travel times to the critical emergency facilities in the Bay County of Florida.

3.1 Case Study 1: County-Based Accessibility

Population block group based accessibility measures are used to calculate the weighted average travel costs for each county in Florida, as described extensively per the methodology in [7]. Results are used to compare and rank the counties to point how accessible the counties are. Equation (1) is performed for each type of shelter with the different travel cost types. The weighted county accessibility formula, derived in [7], is as follows:

$$ County\,weighted\,average\,travel\,cost = \frac{{\sum\nolimits_{i = 1}^{n} {\left( {{\text{cost}}_{\text{i}} * {\text{Pop}}_{{65 - {\text{i}}}} } \right)} }}{{\sum\nolimits_{i = 1}^{n} {\left( {{\text{Pop}}_{{65 - {\text{i}}}} } \right)} }} $$
(1)

where cost i is the travel cost (distance or time) at population block group “i” to the closest shelter, and Pop 65-i is the 65+ population living in the population block group “i”. Results are presented in Fig. 3 visually where those counties that have the least accessibility are shown as red. Instead of presenting numerical scores for each county, the five most and least accessible counties to the shelters are shown in Table 1. We note that this case study only includes the static congested travel times due to the excessive computational requirements while assigning the traffic dynamically on the whole Florida roadway network. Dynamic traffic assignment results are presented in the second case study with a small scale application. Figure 3 and Table 1 show that Miami-Dade and Escambia counties stand out as the most accessible counties to the ARC-approved regular shelters although the accessibility of Miami-Dade decreases when congestion is experienced on the network. The five least accessible counties to the ARC-approved regular shelters are Monroe, Franklin, Collier, Lee and Gulf counties (Note that these counties are home to more than 50,000 aging people). Upgrading the locally planned shelters with respect to ARC standards will definitely improve the accessibility for these counties. On the other hand, Monroe County and Franklin County, with the overall lowest accessibility scores, are the least accessible counties to ARC-approved SPNS shelters whereas Miami-Dade County, the most urban county in Florida, stands out as the most accessible county. Repurposing regular shelters in order to serve the aging as well as special needs populations could be a viable option to increase the transportation accessibility for Monroe and Franklin Counties.

Fig. 3.
figure 3

Aging population-based county accessibility maps for distance, free flow and static congested travel times: (a, b, c) ARC-approved regular shelters, (d, e, f) ARC-approved special needs shelters

Full size image

3.2 Case Study 2: Bay County of Florida

The focus area considered in this case study includes one of the largest cities of District 3, namely Panama City, which is located in Bay County. In addition to focusing on the accessibility of shelters, this case study also includes the emergency transportation accessibility of the Northwest Florida Beaches International Airport and the Greyhound bus station located in the Downtown Panama City. Based on the evaluation of the available data, Fig. 4 shows the major roadways and the aging population by block group as well as the locations of the airport, Greyhound bus station, the SPNS emergency shelter, and the only regular shelter outside the affected region. Note that the shaded parts of the maps of Fig. 4a and b show the highest concentration of the total aging populations and aging people living independently, respectively. Please note that the only special needs (SPNS) shelter and airport (Northwest Florida Beaches International Airport) available are located to the far north of the Downtown Panama City, which makes it important to assess the emergency accessibility of these facilities with respect to aging populations living in the downtown Panama City. Due to space limitations, only aging people living independently are considered for this case study (the aging living independently population block group centroids are shown as green dots in Fig. 4). For the purposes of this paper, airport and bus station accessibility scenarios are designed based on non-emergency traffic conditions, and do not consider the closure of any roadway and/or bridge. On the other hand, regular shelter and SPNS shelter scenarios focus on the emergency evacuation of the aging population living independently in Panama City, with closures of the north-bound bridges as well as including the background traffic during the evacuation.

Table 1. Aging population-based accessibility of Florida Counties with respect to ARC-approved regular shelters and ARC-approved SPNS shelters.
Full size table
Fig. 4.
figure 4

GIS-based representation of Panama City/Bay County with the destinations (airport, bus station and SPNS shelter) and origins (population block group centroids for aging people living independently) (a) Total aging populations, (b) Aging people living independently

Full size image

Overall results can be seen in Table 2 with different cost measures: distance, free flow travel time, static congested travel time, and dynamic congested travel times. From an emergency management perspective, it is critical to observe the best and worst case scenarios during the entire evacuation period, presented by two different dynamic congested travel times in Table 2 (calculated based on different evacuation start times). Table 2 indicates that accessibility to the regular shelter, airport and SPNS shelter, in terms of travel time, can increase upto 5, 9 and 17 min more for the selected origins, respectively. This is due to the delays and queues experienced on the transportation network, which cannot be measured by static but dynamic traffic assignment. Travel times also increase when the evacuation starts during the peak hours (presented by the worst case dynamic travel times). This may indicate the need to avoid the peak hours for evacuation purposes, since the transportation accessibility gets relatively reduced.

Table 2. Aging population-based accessibility of 6 Panama City origins with respect to airport, Greyhound bus station, ARC-approved regular shelters, and ARC-approved SPNS shelter.
Full size table

4 Conclusions and Future Work

This study measures and assesses the transportation accessibility of critical emergency facilities through a set of spatial population characteristics and detail travel costs. Different travel cost measures (distance, free flow travel time, and static and dynamic congested travel times) are used to evaluate the accessibility of the counties with respect to these critical emergency facilities such as shelters. Worst and best accessible counties are ranked and presented in tabular format as well as visually with GIS-based maps. Following this, a small-scale emergency transportation accessibility analysis is presented via a case study application developed for Bay County of Florida. This analysis demonstrates the applicability of transportation planning and simulation models for assessing the emergency accessibility provided to aging populations due to extreme events such as hurricanes. The results of this study can be implemented by agencies and stakeholders to help increase the mobility of the aging populations by better planning of future critical facility locations, and by improving knowledge of the emergency and non-emergency transportation accessibility of counties in Florida.

This study currently focuses only on the travel time on roadways; however, waiting and dwelling times can also be critical, and should be added to the travel time costs while calculating the accessibility scores. In future work, this study can be extended to other critical facilities such as hospitals and the local shelters that do not meet the standards of the American Red Cross [20] as well as other vulnerable populations including mobile home residents and seasonal populations. In order to fully support emergency evacuation plans and policies, additional hypothetical scenarios can be developed based on changing the modeling parameters such as the demand loading time, the departure window, and considering different transit alternatives.