GITCHI-GAMI STATE TRAIL USAGE STUDY SUMMER 2015/2016. Prepared for Gitchi-Gami Trail Association (GGTA)

Transcription

1 GITCHI-GAMI STATE TRAIL USAGE STUDY SUMMER 2015/2016 Prepared for Gitchi-Gami Trail Association (GGTA) Prepared by Arrowhead Regional Development Commission (ARDC)

2 1 Gitchi-Gami State Trail 2015/2016 Usage Study Introduction The Gitchi-Gami State Trail is a destination trail connecting users to a variety of outdoor recreation opportunities, including walking and biking, along the North Shore of Lake Superior. Planned to connect the cities of Two Harbors and Grand Marais with 86 miles of paved trail, the current trail stretches 29 miles along the North Shore in six unconnected segments. The idea for the Gitchi-Gami State Trail was originally conceived in the late 1980s and early 1990s, when a group of stakeholders began discussing the possibility of constructing a recreational trail to provide a safe corridor for bicyclists, in-line skaters, joggers, and walkers along Highway 61, paralleling the North Shore of Lake Superior. This vision came to life in 2001, when the first segment of trail was completed. Development of the trail has continued since 2001, and in 2013 a trail plan update identified future segments for construction and renewed the memorandum of understanding for the roles of the Minnesota Department of Transportation (MnDOT), Minnesota Department of Natural Resources (MN DNR), and Gitchi-Gami Trail Association (GGTA). Today, the trail is generally considered popular, but limited data on its actual bicycle and pedestrian traffic volumes and use patterns exist. In turn, the Gitchi-Gami Trail Association, in collaboration with the Arrowhead Regional Development Commission (ARDC), MnDOT, MN DNR, GGTA, Sawtooth Mountain Clinic, and the Center for Prevention at Blue Cross and Blue Shield of Minnesota, launched a project to pilot short-duration continuous monitoring efforts and to collect data on traffic volume and use patterns along the trail in In the summer of 2016 the GGTA was awarded a grant from Minnesota s Lake Superior Coastal Program to continue the project and purchase a new piece of counting equipment for the summer of ARDC served as the technical lead on the project, conducted the study and produced this report. The resulting data and comparison statistics will enable trail stakeholders and partnering agencies to more effectively plan and manage the trail. It will also provide information needed to determine future routine monitoring of the trail. This document explores the methods, costs, and comparison results of the Gitchi-Gami State Trail usage study, discussing the project and making conclusions about the findings at its close. Methods To collect quantitative traffic volume and use data, the ARDC worked with MnDOT staff and University of Minnesota researches to adapt procedures for non-motorized traffic monitoring first published in the Federal Highway Administration s Traffic Monitoring Guide in These

3 2 procedures, which are analogous to the procedures used by MnDOT to estimate annual average daily traffic (AADT) and vehicle miles traveled (VMT) for state highways and county roads, involve establishing permanent reference (or control) monitoring sites and short duration sampling sites. Ratios and statistics from the reference cites are then used to extrapolate short duration samples to estimate AADT. Because the Gitchi-Gami is not plowed and little bicycle (or pedestrian) traffic is expected in the winter, ARDC decided to estimate summertime average daily traffic (SADT) rather than AADT. ARDC team members employed the use of seven automatic infrared trail counters. The counters used in this study, TrailMaster TM1550, utilize invisible infrared beams that count activity when the beam, generated between the transmitter and receiver, is broken by movement. To ensure best collection of passing trail users, the transmitter and receiver windows were set up at a height of 40 inches (just above waist height, the optimal height for counting bicyclists due to potential issues arising from wheels) at 1.0 baud (the rate at which a trail user must break the beam to be registered as an event ). The transmitter and receiver were set up on or next to either side of the trail on stanchions or available structures (i.e. underpass retention walls, bridge railings, etc.); since the TrailMaster TM1550 model has a range of 150 feet, the transmitter and receiver were easily able to register passing trail users. Special consideration was made as to which side of the trail the transmitter was placed, as the counter runs the risk of picking up car activity if the trail parallels a road. In these instances, the transmitter unit was aimed away from the road on the roadside of the trail. Consideration was also made to ensure natural features (i.e. grass, tree branches, etc.) that may wave in the breeze did not interfere with the counter beam, as the counter also runs the risk of registering this kind of activity. If needed, minimal trimming of natural features around the counters was also conducted in order to avoid false counts. Pictures of the trail counters and equipment are shown in Figure 1. The 2016 count process included the addition of one new piece of counting equipment that was used for the Lutsen control site. An Eco Pyro Box counter was purchased using grant funding provided by the Minnesota Lake Superior Coastal Program. This counter works using similar technology as the TrailMaster TM1550 counters by use of an infrared scanner. The Pyro Box also did not require a receiving unit, meaning that there were no misalignment issues when gathering counts. Additionally, the Eco Pyro counter has the ability to transmit data automatically to a cloud based server where data analysis can be performed online. This automatic data transfer and upload capability allowed for consistent monitoring of the Lutsen control site and was less prone to error.

4 3 The tools made available through the Eco-visio online site allowed for a more in-depth comparison analysis of past data and current data for the Lutsen control site. Monitoring took place at 23 sites along the Gitchi-Gami State Trail, with each collection site being approximately 1 to 2 miles apart. The locations of these sites are displayed in Figure 2. Control sites were placed at Iona s Beach Scientific and Natural Area ( Iona s Beach ) and at Lutsen, which were monitored continually throughout the summer. The Lutsen control site was equipped with the Eco Pyro counter. These control site locations were chosen based on their abilities to host seasonal counters (and not fall or be tampered with) as well as their ability to represent respective county segments (by being located centrally and/or in proximity to amenities that represent the Gitchi-Gami State Trail segments, broken down by county). The other sample sites were each monitored for 10 consecutive days. In both 2015 and 2016 the rotating sites were monitored between May 23 and September 8. Due to the availability of five trail counters, five sites were monitored simultaneously throughout each 10-day monitoring increment. Each site was marked with signs to inform trail users of the ongoing count. After each site was counted, ARDC personnel inspected the infrared trail counter for interference, downloaded and stored the data into a database, and moved equipment to the next designated count site. This process began at Lake County trail sites and was repeated northward along the trail until the study concluded at Cook County trail sites. ARDC staff organized and analyzed the count data by exploring collected site usage information as well as by deriving a summer average daily traffic (SADT) count for each site. SADT represents total traffic along the trail, including pedestrians, runners, bicyclists, inline-skaters, and other trail users. Identifying the busiest places and times of use on the Gitchi-Gami State Trail, site usage data and SADT measures were analyzed in adjusted form. Adjusted values were determined by multiplying raw data values by 1.292, a multiplier based on previous trail count studies that corrects for occlusion (when more than one trail user passes the counter simultaneously and is counted as one user). After configuring the data, ARDC personnel broke data down into daily and hourly events, and then calculated SADT counts for each site. To derive SADT, ARDC calculated the percent of summer traffic that occurred at the control site, divided the count at each sample site by the percent of summer traffic at the control site to derive the estimated summer traffic at each site, and divided the estimated summer traffic at each site by the total number of collection days at each site to derive the average daily traffic count for the summer. For example, Site 8 (Split Rock Overlook)

5 4 was counted for nine days (between August 2-10, 2016) totaling 1,146 counts. The control site total for these nine days was 1,009 counts. The percentage of total summertime traffic was calculated by dividing the control site sum for those nine days (1,009) by the total control site sum for the entire summer (7,597) then multiplying by 100, which equaled 13.28%. Then an estimated sample of total summertime traffic was calculated by dividing the Site 8 sum (1,146) by 13.28% which equals 8,628 estimated total counts for the entire summer for Site 8. To calculate the summer average daily traffic count (SADT) for Site 8, the estimated total summertime count value (8,628) was divided by the number of days that the control site was set up (97 days) giving Site 8 an estimated SADT count of 89. This calculation can be represented with the following string of equations, where N C is adjusted trail user counts at the control site throughout summer, N s is adjusted trail user counts at a sample site throughout summer, n c is adjusted trail user counts at the control site during sample dates, n s is adjusted trail user counts at a sample site during sample dates, P nc is the percentage of summer traffic at the control site during the sample dates, and SADT s is the SADT at a sample site: n c N c = P nc n s P nc = N s N s N c = SADT s As staff was interested in using continuous control site counts for analysis purposes, any missing control site collection numbers were estimated on weekday averages generated through SADT numbers. The rationale for this approach is that it is expected that all the sites along the trail vary similarly in response to weather, day of week, and holidays. Thus, for the summer while total traffic at each site may vary because of factors like access, the proportions of traffic within any particular time period are expected to be the same. The idea is that while more or fewer trail users may use particular segments, the users respond similarly to factors like rain or day of week. Hourly graphs and charts shown in the appendices indicate these patterns. Notable findings produced using these data are highlighted in the Results section below. Costs Multiple staff members were involved in this effort. ARDC hired a summer part-time employee to conduct the physical portion of the counts. Their tasks included setting up the

6 5 counters, moving them as needed, and downloading the data for the sections in Lake County and Cook County. ARDC s GIS Specialist assisted with mapping out count sites and analyzed the data for both Lake County s and Cook County s sections. ARDC received significant assistance from other entities that helped minimize costs. MnDOT was able to loan ARDC some of its moveable TrailMaster counters, allowing ARDC to monitor several sites at once. Greg Lindsey of the University of Minnesota guided the setup of initial counters and trained ARDC on counter use and data analysis. Michael Petesch of MnDOT s Bicycle and Pedestrian Section provided data analysis forms and technical assistance with data analysis. Even with this significant assistance, ARDC costs for personnel hours, mileage, and travel expenses for conducting the counts in 2016 totaled to just over $10,000. Results & Comparison The 2015 and 2016 SADT results are shown geographically in Table 1 as well as in Figure 3. Similar trends were observed between the 2015 and 2016 count studies. The highest SADT value for both 2015 and 2016 was derived at Silver Creek Overlook (ID#0), which generated a SADT of 201 in 2015 and 435 in The lowest SADT value for both years was derived at the Silver Bay Railroad Underpass (ID#15), which generated a SADT of 36 in 2015 and 27 in Overall, these renderings demonstrate SADT counts with a range of 165 in 2015 and 408 in 2016; and an average of 109 in 2015 and 103 in 2016 daily trail users per site. Aggregate data from all 2015 trail count sites are plotted in Figures 8 through 42, with data from control sites shown in Figures 8 through 18 and data from sample sites shown in Figures 19 through 42. Aggregate data from all 2016 trail count site are plotted in Figures 43 through 71, with data from the Iona s Beach control site shown in Figures 43 through 47 and data from sample sites shown in Figures 48 through 71. Individual collection site numbers were generated using control site count numbers collected throughout the summer. The 2016 portion of this study included the use of EcoVisio software. Summaries for each of the two control sites for 2015 and 2016 were generated for comparison and can be found in Figures 72 through 75. The 2015 adjusted daily count numbers for Iona s Beach (ID#C1) and Lutsen (ID#C2) are shown in Figures 4 and 5. For the Iona s Beach control site, the 2015 peak day of traffic was registered on August 15, 2015 with a count of 545 trail users; July 4, 2015 was the second highest day of traffic with 345 trail users. Figures 6 and 7 show the 2016 daily count numbers for the Iona s Beach and Lutsen control sites. In 2016, Iona s Beach collection site s peak day of traffic was August 5, 2016 with 198 users; May 29, 2016 was the second highest day of traffic with 171 users. The

7 6 lowest day of traffic in 2015 for Iona s Beach was registered on June 3, 2015 with a count of three trail users. In 2016 the lowest day of traffic for Iona s Beach had a count of one trail user on June 29, For the Lutsen control site, the peak day of traffic in 2015 was registered on August 10, 2015 with a count of 164 trail users; July 4, 2015 was also the second highest day of traffic with 162 users. In 2016 the highest day of traffic was July 3, 2016 with 317 users, and the second highest was August 14, 2016 with 269 users. The lowest day of traffic for Lutsen in 2015 was registered on August 7, 2015 with a count of nine trail users. In 2016 the lowest traffic count was on June 15 with six users. When looking for common comparison trends for both control sites between 2015 and 2016, the most common high count dates occur in August and the busiest day of the week is Saturday. Interpretations of the results are presented in the Discussion section below. Discussion Results from this study suggest popularity of certain segments or nodes along the Gitchi- Gami State Trail. The popularity of segments may be connected to a number of different factors, such as proximity to popular wayside locations, trailheads, population centers, and regional amenities (i.e. state parks and other bicycle trails) as well as a count location on a longer stretch of continuous trail and access to a view of Lake Superior. The collection site with the highest SADT for both years is the Silver Creek Overlook (ID#0). A popular destination attached to a short, unconnected segment of the Gitchi-Gami State Trail, the Silver Creek Overlook is the location of Silver Creek Cliff Wayside Park, an 18-stall state wayside near the Silver Creek Cliff tunnel with sweeping views of Lake Superior. Here, travelers who want to stretch their legs venture down the trail, which would explain high count numbers at the site. In contrast, the collection site with the lowest SADT for both years is the Silver Bay Railroad Underpass (ID#15). Located on a short, yet-unconnected segment of the Gitchi-Gami State Trail, the underpass site winds through the forest and does not offer views of Lake Superior. The closest collection site is Silver Bay City (ID#16), where the current trail terminates near the City s sports complex; incidentally, this site offers the second lowest SADT for both years. These trends show to be consistent between 2015 and 2016 data. Other sites with relatively high SADT numbers include North Gooseberry Bridge (ID#1), Driveway #3396 (ID#3), Blueberry Hill Beach (ID#4), South of Beaver Bay (ID#13), and Tofte Forest Service (ID#19). One reason for these high SADT numbers may be the sites close proximities

8 7 to state parks. North Gooseberry Bridge is located within Gooseberry Falls State Park, and Driveway #3396 and Blueberry Hill Beach are located nearby; and Schroeder/Tofte FS Road and Tofte Forest Service are located near Temperance River State Park. Even if these collection sites were not actually located within these parks, they were connected and within accessible distance (up to one mile) to be used by park visitors. The presence of other recreational facilities nearby may offer another reason why SADT numbers at certain sites were higher than others. This would explain higher counts at Schroeder/Tofte FS Road and Tofte Forest Service sites, where area bike facilities include the network of adventure trails in Cook County, such as the Heartbreak Hill Loop and the Britton Peak Trails in Tofte. Additional facilities like these, which create a recreational destination area along the Gitchi-Gami State Trail corridor, could also contribute to relatively high SADT numbers. Results from this study also suggest peak usage times on the Gitchi-Gami State Trail. The popularity of the trail at certain times may be connected to a number of different factors, including trail events, holidays, popular vacation times, and work schedules. Understandably, count data largely identified weekends, in contrast to weekdays, as the busiest time for the Gitchi-Gami State Trail for both 2015 and Common highest and lowest recorded trail usage days support this (see appendices). Identified through continuous control site numbers, days with the highest trail user counts throughout the season were mostly Saturdays in July and August. While supporting the idea that use of the trail is highest during off-work hours, these high trail usage dates also verify the idea that trail events and programming, such as August 15 th s GGTA Annual Ride, encourage use of walking/biking facilities. Some common days with the lowest trail user counts at control sites throughout the season were August 19, July 6, and June 29 in 2015 and June 29, July 11, and August 11, These numbers may be attributed to a number of different things, but what they all have in common is their status as a weekday. While this similarity does not necessarily mean that weekday usage is lower than weekend usage along the trail, a general conclusion that weekday trail counts are lower than weekend trail counts would make sense given discussion about work schedules and trail use patterns above. Weather is also a main factor that can play a large part in observing low trail counts. According to the reports from the KBFW weather station in Beaver Bay, MN, rainfall, fog, and thunderstorms were reported on all six of the lowest observed count days in 2015 and While most trail data came back normal, it should be noted that abnormally high counts were registered at Gold Rock Point (ID#9) on one count day, measuring 191 users during the summer of Since the TrailMaster counters cannot differentiate a trail user from, for example,

9 8 a child running back and forth, this outlier number was cause for concern in terms of calculating the site averages. In this case, the day s count data was omitted from the data analysis as to not disrupt any findings. Strengths and Limitations The main strength of the study was the use of best practices for trail counting, given available resources. Data collection methods were held constant throughout the study window for both 2015 and 2016, including the length of time counters were set at each location and how the counters were set up to receive usage data. The study also utilized control points that collected information throughout the entire data collection window. This allowed comparison of data collected during different time frames. Despite these strengths, the research also had a number of weaknesses. Two weaknesses came from the counters themselves. Although reasonably reliable, the infrared TrailMaster counters do not differentiate between modes of transportation along the trail (bicyclists, walkers, runners, etc.). Therefore, in order to obtain data on the type of usage along the trail, a device differentiating user types or a person taking notes manually (or collecting and reviewing trail video) would have to count the trail. Another weakness with the counters was their inability to count two or more trail users who pass the counter simultaneously. Known as occlusion, this limitation can only be overcome by validation, or a person counting trail users manually and comparing the manual results to users registered by an automated counter. A third weakness was derived from different time windows of collection at sample locations. Ideally, the data would all be collected simultaneously within the same time window at every collection point along the trail. But with access to only seven counters (2 stationary and 5 roaming), this was impossible, and it hindered the clarity of the final study results. The final weakness came from things going wrong with the trail counters due to outside forces and unforeseen technical problems. Throughout the data collection period, the trail counters were tampered with, likely by trail users, weather, or insects. During the 2016 summer count process, one of the TrailMaster rotating counters ran into an error that caused the counter to record readings of 8,000 plus counts at some point during the 10 day counting period. This error caused the data to become invalid and re-counts were conducted at those sites. Also, at times counters could not register passing trail users due to dying batteries. This occurred on three occasions during 2015: at the control site at Iona s Beach (ID#C1) from June 11th to June 22nd and from July 10 to July 19 and at the control site at Lutsen (ID#C2) from July 15 to August 5. One location during

10 had a battery error: Iona s Beach (ID#C1) from June 28 to July 22. In the case of sample points where this happened, the sites were recounted; in the case of the control sites, data for analysis was derived from control site averages. Conclusion In the end, this project is important because it offers a chance to explore what trail users value and find accessible on the Gitchi-Gami State Trail based on where and when they choose to walk, run, ride, skate, etc. The data collected during the project will allow trail stakeholders and partnering agencies to more effectively plan and manage the trail in the future. Specifically, this includes using the data collected to be used as evidence in support of funding for trail improvements, as level-of-service measures to determine priority areas or the need for trail improvements or expansions, as reference for when scheduled maintenance or trail activities would disrupt the fewest number of trail users, and as fodder for trail marketing and communications efforts. The findings of this project suggest that sections of the trail with direct connections to unique and, in turn, popular destinations (i.e. waysides, state parks, and walking/biking facilities) experience higher usage than those that do not have such ties. This idea is supported by the highest SADT counts being located at collection points in or within one mile of areas such as popular state parks or areas with numerous recreational opportunities. Findings also suggest that peak usage of the trail occurs on weekends and days where programming takes place on the trail and the weather is favorable. This idea is supported by control site counts being highest on weekends in July and on a day when a trail event occurred; these ideas are further supported by control site counts being typically lower on weekdays for both 2015 and Also an important output of this project, ARDC staff acquired and learned how to use continuous trail counters, a skill that can be utilized for future monitoring efforts along the Gitchi- Gami State Trail and other regional trails. Based on this experience, future monitoring activities will include continuing trail monitoring efforts to analyze cross-year data and changes in trail use patterns between years, expanding upon monitoring efforts in different segments of the trail as it continues to be developed, creating validation methods to ensure acceptable counter performance and considering investment in permanent monitoring devices for constant trail use data collection through the year. ARDC is seeking to conduct the counts again in the summer of 2017 at a region-wide level. This would include a continuation of control site counting for the Gitchi-Gami State Trail, and the

11 10 addition of rotating count sites for the Munger Trail, St. Louis Trail, and other various paved state trails throughout the Arrowhead Region. As part of the count, ARDC will purchase more Eco Counters to ensure that data is accurately transmitted. While the new counters do add an expense, ARDC also feels that the experience gained in 2015 and 2016 will create significant efficiencies in 2017, with less training and guidance needed and less travel costs for data collection. Therefore, it is expected that a counting process next summer would cost an estimated $10,000. In conclusion, the aim of this project was to collect usage information for the Gitchi-Gami State Trail as well as learn how to best routinely monitor trail usage for the future. Based on the experiences gained and findings presented above, the lessons learned through this experience will greatly enhance future trail projects and monitoring efforts associated with their trails.

12 11 Appendix Figure 1. Equipment used for the Gitchi-Gami State Trail count project.

13 Figure 2. Gitchi-Gami State Trail Collection Sites. 12

14 Summer Average Daily Traffic Counts 13 Table 1. Gitchi-Gami State Trail: 2015 Summer Average Daily Traffic Counts Site ID # Site Description 2015 Summer Average Daily Traffic (SADT) 2016 Summer Average Daily Traffic (SADT) 0 Silver Creek Overlook North Gooseberry Bridge Driveway # Driveway # Blueberry Hill Beach C1 Control Site Iona s Beach North of Trail Bridge South of Split Rock River Split Rock Trail on Rock Creek Split Rock Overlook Site Gold Rock Point Mile Post Pine Bay Loop Site Mile Post South of Beaver Bay Beaver Bay City Silver Bay Railroad Underpass Silver Bay City Temperance Traders Drive Schroeder/Tofte FS Road 146 No Data 19 Tofte Forest Service Tofte East 121 No Data 21 Rollins Creek 113 No Data C2 Control Site Lutsen C C2 Site Number 2015 SADT 2016 SADT

15 Figure 3. Gitchi-Gami State Trail: 2015 Summer Average Daily Traffic. 14

16 Figures 4 and 5. Adjusted Daily Counts at Iona s Beach and Lutsen Control Sites

17 Figures 6 and 7. Adjusted Daily Counts at Iona s Beach and Lutsen Control Sites

18 Figures 8 through 18. Average weekday/weekend Counts at Iona s Beach and Lutsen Control Sites for

19 18

20 Figures 19 through 42. Weekday and Weekend SADT Counts at Sample Sites for

21 20

22 21

23 22

24 23 Figures 43 through 46. Average weekday/weekend Counts at Iona s Beach Control Site for Figures 47 through 71. Weekday and Weekend SADT Counts at Sample Sites for 2016.

25 24

26 25

27 26

28 27

29 Figures 72 and 73. Eco Counter Data Summaries for Iona s Beach and Lutsen controls in

30 29

31 Figures 74 and 75. Eco Counter Data Summaries for Iona s Beach and Lutsen controls in

32 31

33 32 GITCHI-GAMI STATE TRAIL 2015 USAGE STUDY Prepared for Gitchi-Gami Trail Association (GGTA) Prepared by Arrowhead Regional Development Commission (ARDC) Summer 2015, Summer 2016 This report was funded in part by the Coastal Zone Management Act of 1972, as amended, by NOAA s Office for Coastal Management in cooperation with the Minnesota Department of Natural Resources Minnesota s Lake Superior Coastal Program. Funding was also provided in part by the Center for Prevention at Blue Cross and Blue Shield of Minnesota.