# Transportation Deployment Casebook/Dallas Area Rapid Transit LRT

## Introduction: DART LRT development history & environmental context

Dallas Area Rapid Transit (DART) currently operates North America's largest light rail transit (LRT) system in terms of track length, which first opened in 1996. DART has pursued a fairly aggressive, and successful in terms of project completion, campaign of expansion of service throughout a wide variety of markets in the Dallas area, but its accessibility and economic success is not so readily determined. As part of a nationwide resurgence of streetcar service and the closely related LRT systems, DART currently operates over 140 kilometers of track and 63 stations in the third largest city in Texas by 2013 population estimates (Speck, 2013; US Census Bureau, 2013). However, what are the extents to which the DART LRT system has actually increased regional accessibility? Has DART LRT increased the total transit mode share regionally, which would serve the often-cited goal of transit projects to reduce car mode share and the associated congestion problems? Has DART LRT fulfilled the common LRT goal of increasing nearby property valuation metrics? Are there other policies and decisions interacting with, and stunting the success of the DART LRT system? The answers to these questions would appear to paint a mixed picture of the system's success, and yield a substantial amount of necessary changes in development patterns.

Total DART system ridership in 2000 according to census data, after the construction of approximately 32 km of track across three LRT lines, was 7.75% below 1990 levels when the entire system operated solely on standard bus service (Cox, 2002). Further, the Dallas area total transit mode share decreased from 3.9% of all trips in 2000 to 1.53% of all trips in 2010 (Freemark, 2010). Transit mode share in the Dallas region has remained steadfastly stagnant, despite significant transit investment from accrual of a 1 percent annual sales tax levied in 1983 to fund the eventual DART LRT system. Additionally, such non-gains in transit mode share occurred during a time when the Texas GDP rose from $613 billion in 1997 to$1.53 trillion in 2013, and the Dallas-Fort Worth metroplex specifically rose from $255.91 billion in 2001 to$447.57 billion in 2013, which may suggest that upward economic mobility does not correlate with choice transit use in the Dallas region (Federal Reserve Bank of St. Louis, 2014; Bureau of Economic Analysis, 2013).

Additional elements of the Dallas transportation-development landscape consist of current transit-oriented development practices (or lack thereof), related Dallas parking policies, and DART LRT headway. Perhaps a shining example of transit-oriented development executed properly is the Mockingbird station on the Red, Blue, and Orange LRT lines - at an initial capital investment of $105 million, development surrounding the station includes loft apartments, office space, retail, and other amenities in a "transit village" style, along with 735 free transit-oriented parking spaces at the station (Dittmar, 2004). However, Mockingbird station is not the norm among the broader DART system, whose stations boast a sum total 19,576 free parking spaces system-wide; only 25 of the system's 62 stations decline to provide free surface or ramp parking, and a full 35 stations (56\%) provide more than 100 spaces, with 3 stations providing over 1,000 spaces (Arapaho Center, Park Lane, and Bush Turnpike) and one station (Parker Road) providing over 2,000 spaces (DART, 2014). Development patterns typical of a DART LRT station do not follow the Mockingbird example - many stations are placed next to major highway systems in less expensive corridors than more expensive, but walkable, corridors, and are not integrated with any semblance of neighborhood urban fabric (Freemark, 2010). As such, walkability to DART LRT stations is reduced, and amenities towards pedestrians within the immediate station vicinities are reduced in favor of large amounts of surface and ramp parking. If in the words of Portland's Charlie Hales, the modern streetcar is the "pedestrian accelerator," then development around LRT stations needs to follow a "pedestrian enabler" paradigm to be successful, and that involves proper transit-oriented development. Parking policies need not exist in zones or areas immediately adjacent to a transit line or station to influence the transit system's ridership. Transit lines typically are designed to shuttle people into a city's Central Business District (CBD) to encourage additional economic activity in the city's center. However, surface and ramp parking in the CBD of Dallas, a city which cut its teeth on large growth in the automobile era, is in significant abundance; it was found that in 2011, an average of more than 7,000 spaces lay vacant during peak weekday hours, and the typical cost of parking is merely$1 per hour (Watkins, 2013). Thus, it seems Dallas parking policy is locked-in. Many cities with such low cost of parking are underpricing parking relative to the actual open-market value of the parking spaces, but not so in Dallas - the supply is sufficiently large that an "underpriced" pricepoint of $1/hour is approximately equal to the market value, yielding an inordinately large consumer surplus (people would gladly pay significantly more than$1/hour for the luxury of driving to work in car-centric Dallas) (Speck, 2013). And when parking is both very available and very cheap, use of the transit lines into the CBD is heavily disincentivized. The prevalence of free parking at DART LRT stops in an attempt to encourage park-and-ride use of the system is very telling of the parking situation downtown - DART must grossly incentivize driving to most of their LRT stations to the detriment of pedestrian accessibility in order to compete with abundant, cheap CBD parking, and yet Dallas transit mode share remains very low. This continues the cycle of congestion problems that plague auto-era cities such as Dallas; in fact, congestion delays increased by almost 200\% in 68 major US metropolitan areas between 1982 and 1997 (Shrank \& Lomax, 1999).

LRT system headway also factors into how likely users are to board a transit system. DART headways on their downtown LRT lines (e.g. The Red Line) are 15 minutes during peak periods, and reduced to 20 minutes off-peak (DART, 2014). A useful comparison regarding headways is the Blue Line of the Minneapolis-St. Paul METRO system; both on-peak and off-peak service between the hours of 6am and 7pm operates at headways of 10 minutes, with reductions before and after (Metro Transit, 2014). The comparison is justified based on similar urban form and metrics of sprawl - both the DART Red Line and METRO Blue Line extend from their corresponding CBDs outward to less-dense, "sprawled" areas of Plano, TX (Parker Road station) and Bloomington, MN (Mall of America station). According to the 2014 Smart Growth America report, composite sprawl metrics for the Dallas-Plano-Irving and Minneapolis-St. Paul-Bloomington statistical metropolitan areas were 86.15 and 88.69, respectively, putting the Dallas region at 152nd least "sprawled" and the Twin Cities at 147th least "sprawled" out of 221 areas above 50,000 population measured. For benchmarking, sprawling mecca Atlanta-Sandy Springs-Marietta scored 40.99, and the densest metropolitan areas in the U.S. of New York-White Plains-Wayne and San-Francisco-San Mateo-Redwood City scored 203.36 and 194.28, respectively (Smart Growth America, 2014). Sutton (2003) utilized nighttime satellite imagery to evaluate metropolitan area sprawl extent by comparing actual population to expected population based on nighttime luminous intensity. He found that Dallas had a -33% population difference between actual and expected for low-threshold luminous intensity, and a -15% difference for high-threshold luminous intensity. So, instead of concentrating more frequent service in closer-in areas adjacent to the CBD in Dallas, DART has built out their LRT network to less-dense areas further out, and reduced service headways accordingly to maintain operational parity.

Using metrics of system status and maturity of ridership levels, track extent distance, number of stations, and number of free parking spots located at DART stations, we aim to explore the life-cycle of the DART LRT system itself; using metrics of total DART system ridership, population density, "urban sprawl," and interactions with the rest of the built environment, we aim to assess the true effectiveness of the DART LRT system and determine whether funding the system initially was wise.

## Methods

Data were collected from the publicly available sources hosted by DART, and included LRT ridership (manual count method), LRT ridership (automatic count method), track kilometers built, number of stations built, and number of free parking spaces built. All cumulative figures (e.g. track kilometers) are defined at the end of a given fiscal year, in order to correspond directly with DART's ridership tracking. Beginning in FY 2013, DART switched its ridership counting methodology to exclusively electronic, whereas in previous fiscal years counting was performed either manually or, for comparison, using both methodologies. It was found that DART's manual counts had underreported LRT boardings by at least 15% annually compared to the more accurate automatic counts (Dart, 2013). As such, our analysis will consider both datasets simultaneously for comparison. Additionally, the final data point in the series of ridership, for FY 2014, is derived not from measurement but from the DART budget plan for FY 2014.

For each metric of system extent (ridership, track kilometers, number of stations, and free parking spaces), a bivariate nonlinear logistic regression model was constructed using MATLAB, to analyze the usefulness of a given metric in predicting end-state system extent. The model formulation was as follows:

${\displaystyle Y={\frac {K}{1+e^{-b(t-t_{0})}}}}$

where ${\displaystyle K}$ is the overall carrying capacity of the system, ${\displaystyle b}$ is a rate modifier coefficient, and ${\displaystyle t_{0}}$ is the time at which the system will be half-built, i.e. where ${\displaystyle Y=1/2K}$.

The parameters to be estimated by regression analysis are ${\displaystyle K}$, ${\displaystyle b}$, and ${\displaystyle t_{0}}$. Such model formulation was not used for calculated trip-per-variable metrics, which included trips per track kilometer, trips per station, and trips per free parking space, but these metrics were calculated to offer some insight into marginal gains in ridership as related to the metrics of extent. Pearson's two-variable correlation coefficients were used to assess the validity of the regression models.

## Results

The DART LRT system opened relatively recently, in 1996, so annual figures are relatively fewer compared to considerably more mature systems, such as Boston's T heavy rail and trolley system. Following is a table of the estimated parameters for each of the five models constructed of DART LRT network extent, as well as the Pearson correlation coefficients for each model.

Parameter Ridership (millions, manual) Ridership (millions, automated) Track (km) Stations Parking spots
${\displaystyle K}$ 33.61 38.65 7670.13 4745.55 23974.93
${\displaystyle b}$ 0.1277 0.1277 0.0899 0.0749 0.1798
${\displaystyle t_{0}}$ 8.256 8.256 61.32 74.57 9.557
${\displaystyle R^{2}}$ 0.9258 0.9258 0.9151 0.9326 0.9021
Table 1: Nonlinear logistic regression parameter estimation.

However, compared to the typical orders of magnitude of the values of ${\displaystyle K}$ for real systems, it is obvious that a few of the models may not accurately describe the evolution of the DART LRT system. To calculate this, we define system feasibility ${\displaystyle F}$ as a binary function:

${\displaystyle F(K)=\left\{{\begin{array}{lr}1:\lfloor {log_{10}{K}}\rfloor =\lfloor {log_{10}{(M_{t})}}\rfloor \vee K\leq {3M_{t}}\\0:K\,{\text{otherwise}}\end{array}}\right.}$

That is, the carrying capacity of the model ${\displaystyle K}$ must be of the same order of magnitude as the latest actual measurement of the metric ${\displaystyle M_{t}}$, or ${\displaystyle K}$ must be less than or equal to 3 times the latest metric value (i.e. we assume that the validity of a model of DART LRT is suspect if the model predicts a final extent of more than three times the current extent). The binary feasibility values for the five models are 1, 1, 0, 0, 1. Note that this is a heuristic calculation to give a rough idea of whether the model would describe the system's actual end-state behavior, and is not a rigorous statistical validation based on similar, existing systems. The ${\displaystyle t_{0}}$ values for both ridership models were 8.256 years, indicating a prediction that DART LRT ridership reached half of its maximum value shortly after fiscal year 2004. The predicted end-values of system ridership were 33.61 million annual trips under the manual count regime, and 38.65 million annual trips under the automatic count regime. The predicted end-value of system parking availability was 23,975 spaces. The ${\displaystyle t_{0}}$ value for the model according to cumulative number of free parking spaces available in the system was 9.557 years, indicating a prediction that DART LRT system extent (as a function of parking availability) reached half of its maximum value around January of 2006.

At the end of the available data window, DART LRT ridership values for manual and automatic counting methods were 26.87 million and 30.90 million, respectively, indicating 79.95% system maturation by the end of FY 2014. The model based on parking space availability predicted 81.65% system maturation.

Figure 1ː Annual DART Ridership (millions, manual tracking)
Figure 2ː Annual DART Ridership (millions, automatic tracking)
Figure 3ː DART track length (km) at end of FY
Figure 4ː Cumulative DART stations (end of FY)
Figure 5ː Cumulative DART parking spots (end of FY)
Figure 6ː Normalized DART unlinked trips per track km (automatic counting)
Figure 7ː Normalized DART unlinked trips per station (automatic counting)
Figure 8ː Normalized DART unlinked trips per parking spot (automatic counting)

Figures 1 and 2 display the source data for DART LRT ridership for manual and automatic counting methods, respectively, as well as the nonlinear logistic regression models. Figures 3, 4, and 5 display similar time series and regression functions for cumulative track length, station numbers, and parking spots. As can be seen in figures 3 and 4, no inflection point exists in the regression functions for track length or station progression, within the region defined by data availability, and the halfway points of development were predicted as 61.32 and 74.57 years after breaking ground, respectively.

Some interesting scaled metrics to investigate, to evaluate the efficiency and marginal gains of the system, are passenger trips per kilometer of track, trips per DART station, and trips per parking space available. For these metrics, we only consider the equivalents of automated trip counting, and adjusted the previous years' data upwards by 15%, the minimum amount of underestimation reported by DART. Figures 6, 7, and 8 report these data - each figure shows the time series of trips per metric, total trips, and cumulative metric, with each series normalized by its respective maximal value.

## Discussion

Prior to the advent of DART LRT service in Dallas, TX, DART transit service consisted entirely of an extensive network of standard bus service, which was very much in line with the type of urban fabric employed in the Dallas area - that is, sprawling, low-density development without nodal neighborhoods, and extensive underpriced surface parking supplies. How cities develop over time significantly influences the range of their possible, and plausible, trajectories of development going forward. Even with the most extensive and well-funded sprawl reversal efforts, the status quo dictates real-world feasibility. The reign of the automobile in Dallas is significant, and the presence of stratified pedestrian systems in a city not often stricken with inclement weather might suggest an implicit hierarchy of modes on the surface level (Speck, 2000). We previously discussed the overabundance of severely underpriced surface parking, even in the Dallas CBD, and its relationship to the effective zoning regulations at many DART LRT stations demanding significant levels of parking availability. It was into this car-dominated market, with no precedent for successful TOD implementation around bus hubs (it can be done), that DART tossed its aggressive LRT vision on the backs of sales tax funding since the early 1980s.

Our analysis of the progression of the DART LRT system suggested that total passenger boardings, and total numbers of parking spaces, were reasonably good proxies for the extent or maturity of the system, and that miles of track and number of stations were not good predictor variables. The less smoothly clustered data are, the less likely will nonlinear regression find a well-fit curve that also makes sense regarding scale of variables involved. Typically, the kilometer-extent of a LRT system only changes in significant amounts or not at all, corresponding to the extension of existing lines or construction of new lines altogether - such "batch" additions of kilometers can serve to grossly overestimate the parameter ${\displaystyle K}$ in logistic regression models, by tending the curve more upwards, even if a system is nearing relatively realistic full extent.

The measures of trips per km of track, station, and parking space available offer some insight into the marginal gains offered by building the system out further. In each case, the total trips and metric approximately follow each other (normalized). Since 2002, the trips per km of track decreased from 0.65 of maximal to 0.55 of maximal; the trips per station fluctuated between 0.6 of maximal and maximal a few times across the entire time window from 1996 to 2014; trips per parking space were near maximal between 1996 and 2000, and held relatively steady at 0.5 of maximal for the duration of the time series. The decrease in marginal trip returns per km of track makes sense, given typical sprawled urban form - each km of track built further out will draw in a lower amount of riders relative to the previous km of track built, thus lowering the overall metric. This is not perfect however, since new lines are started emanating from the CBD in new directions and thus a given, newly constructed km of track may be closer to the CBD than the previous km on a different line, and will thus increase the overall metric. Considering DART's extensive lengths of track, the diminishing returns in ridership are worth noting. Trips per station likely fluctuated for a few reasons, namely that the number of stations is several orders of magnitude lower than the annual ridership figure, and thus ridership per station will be very sensitive to integer changes in the number of stations; additionally, the urban fabric surrounding stations can vary greatly, and comparing a far-flung suburban station with minimal park-and-ride facilities to the "transit village" style Mockingbird station will yield significantly different ridership.

Finally, trips per parking space remained relatively steady throughout the respective changes in both ridership and parking availability, which sheds some light on exactly why DART is so insistent upon maintaining an abundance of free park-and-ride parking. In fact, DART attempted to enforce selective paid parking (targeted only at residents not located in DART's cities of service) at LRT and bus transit stations for two years, but the program was cancelled in 2014 due to lack of revenue seen (Formby, 2013). It is possible that users from outside of the DART service area simply were not using transit in the first place, but it is also possible that even small parking costs at DART stations for users outside of the service area served to dissuade those users from entering the system, and instead incentivized driving into Dallas and taking advantage of abundant surface parking.

So, we know from both our logistic models of ridership and parking availability, and of the reality that long-range DART service extension plans exist, that the system is not currently fully built out, but that we are likely past the 50% mark (a LRT system of 250+ kilometers in sprawling Dallas seems very unlikely). Our model would have us conclude that the system is nearing maturation, with the ridership metric reaching 80% of maximal by end of FY 2014.

Now, we previously discussed the overall decline in transit mode-share for the Dallas region, despite heavy investments in the DART LRT system, which at its face seems to suggest that DART failed. This is not quite so simple - DART built an operational system with extensive coverage across the Dallas region and reasonable (but not ideal) headways, but the real failure in Dallas is the lack of policies discouraging driving and parking in the CBD. Even economic valuation around transit stations has seen gains, inasmuch that residential properties appreciated 32.1% near DART rail stations compared to 19.5% for equivalent properties not near DART stations, and office properties appreciated 24.7% near DART service compared to 11.5% for non-DART properties (Clower, 2002). However, valuation appreciation is typically limited to within 0.4 km (1/4 mile) of LRT stations, and if development density is relatively low and sprawling, a lower total number of businesses and residential areas will see benefits from LRT proximity (Cervero, 2002). Additionally, the urban extent of cities varies nonlinearly with total population (Stewart \& Warntz, 1958). Thus, transit planning agencies need to be very careful about using total population to determine system extent.

Mackett (2003) identified several urban planning policy areas that must be paid attention to when a new LRT system is built within a region, if the system is to be successful. Agencies must adapt policy to take advantage of the new system, transit-oriented development must take place involving mixed-use facilities, public development must occur near stations, and pedestrian amenities must be included to enable and encourage walking to transit. By many of these measures, Dallas-area planners have failed to give back to the DART LRT system in return for the connectivity and accessibility the system affords the region. So even if the system exceeds our models' estimates of carrying capacity, an environment devoid of policy changes and a transportation market that fails to disincentivize parking in the Dallas CBD will fail to capture value from the DART system. Macket & Edwards (1998) explain that even successful LRT systems that meet ridership expectations can fail, in that they lose mode-share over time as automobile ownership increases and urban policy facilitates decentralization. It is not enough to simply build and mature a system without crafting an environment around it.

## Conclusion

In this investigation, we evaluated both the life-cycle status of the DART LRT system and the effectiveness of the system within the broader economic context of Dallas. Using nonlinear logistic regression, we determined that the DART LRT is possibly ~ 80\% mature in terms of ridership and parking availability. However, within the broader context of transit mode-share within the Dallas region, it is clear that policy and development have not followed in the footsteps of track-building. Updating this framework with future years' ridership levels and infrastructure metrics will further adapt the models, and assuming no significant policy shifts in Dallas towards parking availability and development practice, may continue to show stagnation towards LRT goals.

## References

• Bureau of Transportation Statistics. (2000). Transportation Statistics Annual Report 2000. US Department of Transportation, Washington, DC.
• Bureau of Economic Analysis. (2013). Regional Data: GDP & Personal Income. U.S. Department of Commerce, Updated September 16, 2014. Retrieved November 6, 2014.
• Cervero, R. (2001). Walk-and-ride: Factors influencing pedestrian access to transit. Journal of Public Transportation, 3(4):1-23.
• Cervero, R., Duncan, M. (2002). Transit's Value-Added Effects: Light and Commuter Rail Services and Commercial Land Values. Transportation Research Record, 1805:8-15.
• Cervero, R., Murphy, S., Ferrel, C., Goguts, N., Tsai, Y., Arrington, G.B., Boroski, J., Smith-Heimer, J., Golem, R., Peninger, P., Nakajima, E., Chui, E., Dunphy, R., Myers, M., McKay, S., Witenstein, N. (2004). Transit-Oriented Development in the United States: Experiences, Challenges, and Prospects. TCRP Report 102, Transportation Research Board, National Academy Press, Washington, DC.
• Clower, T.L., Weinstein, B.L. (2002). The Impact of Dallas (Texas) Area Rapid Transit Light Rail Stations on Taxable Property Valuations. Australasian Journal of Regional Studies, 8(3): 389-400.
• Cox, W. (2002). DART's billion-dollar boondoggle. Dallas Business Journal, June 16, 2002. Retrieved November 6, 2014.
• Dallas Area Rapid Transit. (2013). In Motion: The Official Newsletter of Dallas Area Rapid Transit. Dallas Area Rapid Transit, Spring 2013. Retrieved November 6, 2014.
• Dallas Area Rapid Transit. About DART. Dallas Area Rapid Transit, Retrieved November 6, 2014.
• Dittmar, H., Ohland, G. (2004). The New Transit Town: Best Practices in Transit-Oriented Development. U.S. Department of Commerce, Updated September 16, 2014. Retrieved November 6, 2014.
• Dunphy, R.T. (1996). New Developments in Light Rail. Urban Land, July 1996:37-41,87-88.
• Formby, B (2013). DART looks to end paid parking at light-rail, bus stations. Dallas News, November 5, 2013. Retrieved November 6, 2014.
• Freemark, Y. (2010). Transit Mode Share Trends Looking Steady; Rail Appears to Encourage Non-Automobile Commutes. The Transport Politic, October 13, 2010. Retrieved November 6, 2014.
• Hensher, D.A. (1999). A bus-based transitway or light rail? Continuing the saga on choice versus blind commitment. Road and Transit Research, 8(3):3-21.
• Kim, S., Ulfarsson, G.F, Hennessy, J.T. (2007). Analysis of light rail rider travel behavior: Impacts of individual, built environment, and crime characteristics on transit access. Transportation Research Part A, 41:511--522.
• Kuby, M., Barranda, A., Upchurch, C. (2004). Factors influencing light-rail station boardings in the United States. Transportation Research Part A, 38:223-247.
• Mackett, R.L., Edwards, M. (1998). The impact of new urban public transport systems: will the expectations be met? Transportation Research Part A, 32(4):231-245.
• Mackett, R., Babalik, E. (2003). New urban rail systems: a policy-based technique to make them more successful. Journal of Transport Geography, 11:151-164.
• MetroTransit. (2004). METRO Blue Line.MetroTransit, Retrieved November 6, 2014, from https://www.metrotransit.org/route/901.
• Parkinson, T., Fisher, I. (1996). Rail Transit Capacity. TCRP Report 13, Transportation Research Board, National Academy Press, Washington, DC.
• Pickrell, D.H. (1992). A desire named streetcar: fantasy and fact in rail transit planning. Journal of American Planning Association, 58(2):158-176.
• Plane, D. (1995). Urban Transportation: Policy Alternatives. The Geography of Urban Transportation, second ed., Guilford Press, New York:435-469.
• Shrank, D., Lomax, T. (1999). The 1999 annual mobility report: information for urban America. Texas Transportation Institute, College Station, TX.
• Smart Growth America, Metrpolitan Research Council. (2014). Measuring Sprawl 2014. Smart Growth America, April, 2014.
• Speck, J., Duany, A., Plater-Zyberk, E. (2000). Suburban Nation: The Rise of Sprawl and the Decline of the American Dream. North Point Press, New York.
• Speck, J. (2012). Walkable City: How Downtown Can Save America, One Step at a Time. North Point Press, New York.
• Stewart, J., Warntz, W. (1958). Physics of population distribution. Journal of Regional Science, 1:99-123.
• Federal Reserve Bank of St. Louis (2014). Total Gross Domestic Product by State for Texas. Federal Reserve Bank of St. Louis, Updated June 6, 2014. Retrieved November 6, 2014.
• Sutton, P.C. (2003). A scale-adjusted measure of "Urban sprawl" using nighttime satellite imagery. Remote Sensing of Environment, 86:353-369.
• United States Census Bureau (2013). City and Town Totals: Vintage 2013. United States Census Bureau, Washington, DC. Retrieved November 6, 2014.
• Watkins, Matthew (2013). As downtown Dallas rebounds, parking is becoming a challenge. The Dallas Morning News, December 31, 2013. Retrieved November 6, 2014.
• Weinstein, B.L., Clower, T.L. (2002). An Assessment of the DART LRT on Taxable Property Valuations and Transit Oriented Development. University of North Texas Center for Economic Development and Research, Report Prepared for Dallas Area Rapid Transit.

## Appendix: Data

FY Ridership (millions, manual) Ridership (millions, automated) Track (km) Stations Parking spots
1996 4.73 5.44 18.0 14 1739
1997 7.97 9.17 32.2 20 4014
1998 10.97 12.58 32.2 20 4014
1999 11.34 13.04 32.2 20 4014
2000 11.43 13.14 32.2 20 4014
2001 11.51 13.24 37.0 22 4502
2002 13.73 15.79 60.7 29 7643
2003 16.97 19.52 71.9 34 12025
2004 16.49 18.96 71.9 34 12025
2005 17.48 20.10 71.9 34 12025
2006 18.58 21.37 71.9 34 12025
2007 17.90 20.59 71.9 34 12025
2008 19.40 22.31 71.9 34 12025
2009 19.00 21.85 76.8 39 12225
2010 17.80 20.47 76.8 39 12225
2011 22.30 25.65 115.4 55 17321
2012 24.09 27.70 124.1 58 18036
2013 25.65 29.50 137.6 61 19576
2014 26.87 30.90 137.6 62 19576
Table 2: Raw DART system data.
FY Trips per track km (thousands) Trips per track km (adjusted, thousands) Trips per station (thousands) Trips per station (adjusted, thousands) Trips per parking spot Trips per parking spot (adjusted)
1996 262.42 301.78 337.86 388.54 2719.95 3127.95
1997 247.62 284.76 398.50 458.28 1985.55 2283.38
1998 339.89 390.87 547.00 629.05 2725.46 3134.28
1999 352.32 405.17 567.00 652.05 2825.11 3248.88
2000 355.11 408.38 571.50 657.23 2847.53 3274.66
2001 310.96 357.60 523.18 601.66 2556.64 2940.14
2002 226.30 260.24 473.45 544.47 1796.42 2065.88
2003 235.390 271.28 499.12 573.99 1411.23 1622.91
2004 229.23 263.61 485.00 557.75 1371.31 1577.01
2005 242.99 279.44 512.12 591.24 1453.64 1671.68
2006 258.28 297.02 546.47 628.44 1545.11 1776.88
2007 248.83 286.15 526.47 605.44 1488.57 1711.85
2008 269.68 310.13 570.59 656.18 1613.31 1855.30
2009 247.51 284.63 487.18 560.26 1554.19 1787.32
2010 231.87 266.66 456.41 514.87 1456.03 1674.44
2011 193.26 222.25 405.45 466.27 1287.45 1480.57
2012 194.12 223.24 415.29 477.59 1335.49 1535.82
2013 186.43 214.39 420.53 483.61 1310.39 1506.95
2014 195.28 224.57 433.38 498.39 1372.58 1578.46
Table 3: Calculated DART trip-per-metric data.