Mining Cell Phone Data to Create a Multimodal Transportation Plan for Lake Tahoe
by Cynthia Albright, AICP-CUD, GISP
Lake Tahoe is an awe-inspiring destination that straddles the border of California and Nevada. It is the second deepest--and one of the largest--fresh water lakes in the U.S. A 72-mile, mostly two-lane roadway encircles the lake connecting visitors to popular beaches, mountain hiking trails, boat ramps, golf courses, live entertainment venues, casinos, restaurants, and ski resorts. Millions travel to the Lake Tahoe Basin annually, with more coming each year. Tracking visitors historically relied in large part on transient occupancy tax collections (TOT), license place surveys, and state-implemented traffic counts. These methods of data collection involve a great deal of time, financial resources, and extrapolation.
To create the Linking Tahoe: Multimodal Transportation Management Plan on behalf of the Tahoe Transportation District (TTD), Stantec analyzed cell phone data from an enormously large geographic area. We gained unprecedented insights into the magnitude of visitors and their travel patterns. This article highlights a few of our findings and some lessons learned. Cell phone data enabled us to develop an innovative transportation management plan using the newest information technology and sophisticated ArcGIS tools to visualize the results.
Like many popular resort destinations, the steady increase in visitors worsens traffic congestion--compromising the quality of experience and impacting air quality. Lake Tahoe agencies and non-profits have collectively spent $1Billion to date to improve Lake Tahoe’s clarity. Countless installations reduced sediment loading from construction activities and stormwater runoff; bike lanes striped, and shared use paths paved. However, transit remains disconnected with long headways and few rider amenities. The purpose of our multimodal transportation plan was to fully understand the travel patterns of the various travel groups and integrate those findings with: transit routes, stops and boardings; locations of sidewalks, bike lanes and shared use paths; development patterns and zoning; popular Basin destinations, and accident locations. The outcome was a bold, yet achievable, action plan based on a far more comprehensive dataset than ever collected before now.
How We Did It
Third party vendors, in partnership with wireless carriers, aggregate and anonymize the locations of cell phones. These companies process tens of billions of data points each day and turn them into meaningful information. Vendors aggregate the data points by a desired geographic area, time period, and selected variables or attributes. The size of geography, number of data fields, and duration of data collection determines the cost. Stantec acquired a vendor’s complete menu of variables for the entire Tahoe Basin for under $100,000.
The data acquisition included the months of February and July to analyze the variance between winter and summer travel patterns. We also obtained the month of August to confirm the peak summer month in Lake Tahoe. To capture population movements at the five locations of ingress/egress into the Basin, Stantec created five external polygons encompassing all area highways. Due to the size of the Lake Tahoe Basin and the unique qualities of specific areas, we also divided the Basin in six subareas. We provided the vendor with a geospatial file that included the five external zones and over 300 traffic analysis zones (TAZs) for data aggregation.
This seemingly straightforward data acquisition process required careful and strategic thinking to fully comprehend the questions we needed the data to answer before engaging the vendor. If you do not think it through beforehand and ask for additional data after the vendor has aggregated it to your study area, the cost is potentially double. To avoid this possible outcome, we engaged the entire team to discuss what answers we could glean from the menu of data fields available. The vendor’s menu included: origin and destination by TAZ, start and end dates collapsed to a week day or weekend day trip, time of day period, trip purpose, and residence class. Time of day period can be as small as hourly; however, we selected five increments beginning at midnight given the nature of casino operations.
Trip purpose includes home based work, home based other, non-home based. The most important variable: residence class, which we refer to as “travel group”. The residence class attribute identifies if the cell phone belongs to a short-term visitor (1-2 days), a long-term visitor (3+ days), inbound or outbound commuter, resident worker, or home based worker. Our lesson learned: if the study area encompasses a resort destination request the number of visitor trips counted for under a 24-hour period.[1] Our provider was unable to break short term visitors into day versus overnight visitors. Had we thought about that beforehand, we may have chosen another vendor. Unfortunately, it did not occur to any of us the need to identify the number of visitors who enter and exit the Basin in under 24 hours. However, we had access to previous research documenting 43 percentof all visitor trips are less than a 24-hour period.
The data deliverables included:
- trip matrices in a .csv file format—February, July, and August aggregated to 300+ TAZs and five external zones
- geospatial point density files representing the average weekday and weekend destinations for February and July
- an excel spreadsheet with various tabs that identified specific visitor activities including home locations by county and state, arrival and departure dates, length of stay, and visitor arrivals at area airports
The vendor extrapolates the unique cell phones seen in the study area and time-period to represent the entire population based upon the census tract population of the home locations.
What Did We Do With all this Information?
With this much information about the population movement patterns, it was difficult to stop conducting query-analysis-ArcGIS mapping exercises. The data is gloriously robust! We analyzed traffic counts available from Caltrans and NDOT to identify any irregularities or inaccuracies in the wireless device data. Our traffic engineer annualized the wireless device data using the state DOT’s cordon counts from the five points of entry. We calculated that annual visitors entering the Lake Tahoe Basin total 24 million, or 2 ½ times the number of annual visitors to the busiest national park: Great Smoky Mountains, that coincidentally also straddles two states. The home location data identified California as the source for over 60 percent of all visitors. Even small growth forecasts in a state of 37 million people have huge implications for the increasing number of visitors to Lake Tahoe. Figure 1 highlights the magnitude of visitor trips compared to residents entering the Basin.
Stantec used the ArcGIS Hot Spot Optimization Tool and the geospatial point files for February and July to pinpoint those destinations that received the greatest number of visitor trips. With tens of thousands of points, this tool enabled us to statistically identify with a 90-99 percent confidence interval the hot spots, in red, and the cold spots in blue. Cold spots indicate secondary frequented locations. Of note, is the absence of popular Sand Harbor from either a hot or cold spot designation. Further investigation of the hot spots revealed many of these locations offer both popular visitor destinations and vacation homes. Figure 2 illustrates the July 2014 Hot Spot Destinations.
The Tahoe Transportation District established a goal of achieving “20 in 20” – 20 percent transit mode share in 20 years. Transit ridership totaled 1.1 million in 2015. Using the cell phone data, we estimated the total number of internal trips made throughout the Lake Tahoe basin annually exceeded 79 million. Since only 1.4 percent of all trips were made on transit, the District has a long way to go to reach its goal. Using ArcGIS and the wireless device data, the team correlated transit ridership by route and converted cell phone or “person” trips to vehicle trips to key destinations. Figure 3 shows that 1.6 million vehicle trips were counted at the popular Emerald Bay State Park/Eagle Falls areas, while the seasonal transit shuttle carried just 7,500 people. This finding, coupled with the hot spot destination findings, allow for better alignment of transit services and mobility infrastructure with population movements and locations.
The Linking Tahoe: Multimodal Transportation Management Plan included detailed recommendations for transit services and route, prioritized gaps in bikeways, proposed sidewalks; and most importantly, identified numerous mobility hub locations throughout the Lake Tahoe Basin and at key interception points. Mobility Hubs and not park-n-ride lots. Instead, they offer additional services such as bike rentals, bike lockers, car parking, and a transit shelter. The intent is to intercept visitors and shift the transportation form from passive to active and single car to transit use. An overall summary of the transit recommendations is shown in Figure 4.
Converting vehicle trips to transit rides will require significant human and financial resources. After several high-traffic summers and a debilitating winter, the basin agencies are motivated. Lake Tahoe successfully demonstrated that when all five counties, two state agencies, and three federal land management agencies come together to pool resources they achieve tremendously successful outcomes.
The Linking Tahoe: Multimodal Transportation Management Plan could not have been possible without the financial contribution and support of the Tahoe Metropolitan Planning Organization (TMPO). The analysis and findings from cell phone data will enable the TTD and TMPO to continue mining population movements. Instead of a pure “vehicle count”, each person trip counted is associated with a travel group, time of day segment and weekday or weekend movement. The possibilities of exploring trips between the 300+ TAZs with data this robust are endless for transportation and multimodal enhancement planning.
About Stantec
The Stantec community unites approximately 22,000 employees working in over 400 locations across 6 continents. Our work—engineering, architecture, urban design, planning, landscape architecture, surveying, environmental sciences, construction services, and project economics, from initial project concept and planning through to design, construction, commissioning, maintenance, decommissioning, and remediation—begins at the intersection of community, creativity, and client relationships. Our local strength, knowledge, and relationships, coupled with our world-class expertise, have allowed us to go anywhere to meet our clients' needs in more personalized ways. Visit us at stantec.com or find us on social media.
Endnotes
- Fortunately, we had access to several years of Nustats data that identified the proportion of single day visitors as 43%.
PHOTO BY JOHNSTONE STUDIOS.
Cynthia Albright, AICP-CUD, GISP is a certified urban planner and urban designer with 28 years of experience. As a principal with Stantec based in Reno, Nevada, Ms. Albright works on a wide variety of master planning and development projects for both the private and public sectors. Apple Inc.’s expansive data center campus and downtown warehouse, the city of Reno’s sanitary sewer capacity analysis, and the Linking Tahoe: Transit Master Plan, Ferry Oriented Development (FOD) Plan, and Corridor Connection Plan keep her busy. Cynthia presented on this topic in April 2017 at the Ontario Public Transportation Conference in Toronto. Her article was published in the American Planning Association, Transportation Planning Division Newsletter July 2017.
Published in the August 2017 issue