【正文】 System (GPS) technology incorporated into an onboard instrumentation system can provide position and velocity data as a potential solution to the fieldofview , recording data frequently enough to provide good results produces a very large volume of data. With several thousand data records produced daily on each piece of instrumented equipment, the issue bees managing the data and identifying the relatively few key records that mark the beginning and end of the activities being studied. Regardless of the tool implemented, the process is performed in four phases: data capture, preliminary review, data reduction, and data analysis. The tools used in each phase for various systems can be seen in Table 1. A field observer identifies the key times in real time with instantaneous decisions of when one activity stops and the nextstarts based on enormous volumes of visual information. This work developed a methodology for making equivalent decisions based on GPS data, specifically to answer the question, How can the large volume of shortinterval GPS data collected be automatically reduced to the small volume of key discrete points that mark the start and stop of activities? This paper briefly describes the hardware used to record thenecessary GPS data and presents procedures developed to identify the key records necessary to calculate activity durations. A case study is used to show the application of the system to an earthmoving operation. This paper also postulates how the information from the data can be used in discrete event simulations. Descriptions of discrete event simulation applied to construction operations can be found in Martinez and Ioannou (1995), Ioannou and Martinez (1996a), Martinez (1998), Sangarayakul (1998), Cor and Martinez (1999), and Ioannou (1999). Previous Work An early use of GPS to collect productivity data was a pilot study designed to provide current information regarding earthwork operations (Ackroyd 1998). Position information was transmitted via radio and recorded by a remote PC at 2min intervals and at the closure of proximity switches on levers of articulated haulers and motor scrapers. In the initial phase, analysis of the data focused on measuring cycle times, but no real analysis was performed of positions between events. In a second phase, strain gauges were used to estimate payload and thereby provide productivity data. This work depicts the potential of GPS for recording data from earthmoving operations but focuses on the hardware issues rather than the data reduction issues. Ackroyd concludes that the information may add value to monitoring and scheduling tasks but does not address use of the data for analyzing earthmoving operations.Field Data ReductionOglesby et al. (1989) point out that the purpose of a time study is to record the times of various activities that make up an simplest time studies—stopwatch studies—rely on the observer to decide the point in time at which activities start and et al. (1989) note that placing this responsibility on the observer limits the usefulness of the data. Observers can have differences in opinion, data recorded by a single observer may vary over time, available information is strictly limited to that recorded and is subject to the physical limitations of the , the information recorded is strictly limited to the field of view of the observer.Bjornsson and Sagert (1994) presented PAVIC+ as a puterbased system to extract data from video recording of construction operations, including activity durations. Each activity instance is called a segment and defined by a start and stop time. Three tools are available for registering segments: startstop time, cycle time, and consecutive time. The startstop time tool operates like a stopwatch: the user indicates through keystrokes both the start and stop time. PAVIC+ reads the time of the keystroke from a time stamp placed on the audio track and records cycle time tool is used when the same work sequence is repeated. The keystroke used to indicate the stop of an activity also indicates the start of the successive activity, which eliminates the need to press two keys simultaneously. The consecutive time tool is appropriate for a group of correlated activities, such as a crane serving multiple crews. Each activity is assigned a unique key that is struck when the activity ends. This keystroke indicates the stop of the associated activity and the start of the successive activity. Frank (2001) describes a very similar system used to extract information from digital presents and discusses the Digital Video Analysis Tool (DVAT) and notes that PAVIC+ was the basis on which DVAT was developed. DVAT is a software package specifically developed for stripping information from digital video. The extracted information can be used to generate crew balance charts or to develop input for puter simulation models. These traditional data collection methods have relied upon manual tools limited to the field of view of the observer. Researchers have turned to technology for tools that are both automated and not restricted to field of view. Kannan (1999) describes the use of onboard instrumentation systems for recording data from earthmoving operations. He used mechanical parameters recorded by the Vital Information Management System (VIMS) and the Total Payload Management System (TPMS) produced by Caterpillar to obtain operational data. Such systems place a virtual observer in the equipment, and thus the equipment is always in the virtual field of view.Kannan (1999) states that when using onboard instrumentation, the sensors must be able to detect a change in the status of the truck. This change in status indicates the start or stop of an activity. Kannan and Vorster (2000) state that mechanical parameters can be translated to production measures through the use of surrogate measures and protocol rules. Kannan (1998) describes the use of suspension s