State and local departments of transportation (DOTs) increasingly deploy road detectors, such as inductive loops, to monitor congestion on their road networks. As deployment increases, the operating and maintenance cost associated with these detector systems will become issues for many state DOTs. Agencies will need to decide where to add new detectors and which detectors should continue receiving maintenance, given their resource constraints. For data collected from these sensors to remain meaningful, traffic data quality should not be adversely affected in these decisions. The needed traffic data quality depends on the data’s intended purposes. An empirical study was conducted to address the impact of sensor spacing along freeway corridors on the computation of performance measures such as travel time index. The scenario that has the smallest sensor spac- ing (greater density of sensors) is considered to capture actual traffic con- ditions most closely. If sensor spacing is increased, how will the quality of the traffic data be affected? The results showed that when sensors were deleted relative to the baseline sensor spacing condition, the congestion measure statistic varied. This congestion measure did not become “worse” as more sensors were deleted. Instead, sometimes one spacing pattern overestimated the measure, and at other times another spacing pattern underestimated this measure. The analysis showed that the location of the sensors is important in the estimation of congestion for the corridor. State and local departments of transportation (DOTs) increasingly deploy road detectors, such as inductive loops, to monitor congestion on their road networks. As deployment increases, the operating and maintenance cost associated with these detector systems will become issues for many state DOTs (1). Agencies will need to decide where to add new detectors and which detectors should continue receiving maintenance given their budgetary constraints (2). For data collected from these sensors to remain meaningful, traffic data quality should not be adversely affected in these decisions. The needed traffic data quality depends on the data’s intended purposes. For example, the required quality level of the data used for real-time traffic monitoring will be different from that of data used for planning (3). BACKGROUND An empirical study was conducted to address the question regard- ing the impact of sensor spacing along freeway corridors on the computation of performance measures such as travel time index (TTI). The scenario that has the smallest sensor spacing (greater density of sensors) is considered to capture actual traffic conditions most closely. If sensor spacing is increased, how will the quality of the traffic data be affected? In this study, the TTI is used to measure the congestion on a macroscopic level. Data from the Mobility Monitor- ing Program (MMP) were used to conduct this analysis (4 ). The 2002 MMP data from Cincinnati, Ohio, and 2003 MMP data from Atlanta, Georgia, were used. This document describes the sensor spacing analy- sis along particular freeway corridors in these cities, not including data from ramps and arterials. The detector measurements of spot speeds are used to represent the travel time over the entire zone of influence, basically half the distance upstream and downstream to the next detector. This assumes that the speed is constant over the entire subsegment defined by the zone of influence. As the zone of influence increases in length, in theory, the spot speed from the detector should become less repre- sentative of the travel time over its length. The major purpose of this paper, then, is to determine “how bad” the resulting performance measure becomes as this distance increases. Cincinnati Data from only one corridor were analyzed. There were not enough sensors on this corridor to divide it into smaller segments. The sensor spacing along this corridor is approximately 0.5 mi. Table 1 presents a description of the corridor used in the analysis. Figure 1 is a map of the sensor locations in Cincinnati. The sen- sors are located along the I-75 corridor. The map does not show the location of all the sensors. Atlanta For the city of Atlanta, I-75 was divided into four corridors. The average sensor spacing is 0.3 mi. Table 1 presents the different corridors analyzed. Figure 2 shows the corridors that were included in the analysis for Atlanta. METHOD To generate files of different sensor spacing, the files with the detec- tor information were edited. Table 2 presents a listing of the different sensor spacings that were generated. For the purposes of this study, the baseline sensor spacing condition is represented by an average spacing of 0.5 mi for Cincinnati and 0.3 mi for Atlanta (the sensor spacing that is actually present for these cities). It is assumed that Effect of Sensor Spacing on Performance Measure Calculations Iris Fujito, Rich Margiotta, Weimin Huang, and William A. Perez Transportation Research Record: Journal of the Transportation Research Board, No. 1945, Transportation Research Board of the National Academies, Washington, D.C., 2006, pp. 1–11. I. Fujito and W. A. Perez, Cambridge Systematics, Inc., 4800 Hampden Lane, Suite 800, Bethesda, MD 20814. R. Margiotta, Cambridge Systematics, Inc., 1265 Kensington Drive, Knoxville, TN 37922. W. Huang, Cambridge Systematics, Inc., 2457 Care Drive, Suite 101, Tallahassee, FL 32308. 1