Previous Page Table of Contents Next Page

Interface--

Interfaces

Archaeology and Technology

Integration of Global Positioning Systems into Archaeological Field Research: A Case Study from North Kohala, Hawai'i Island

Thegn L. Ladefoged, Michael W. Graves, Blaze V. O'Connor, and Robin Chapin


Contents


Recent developments in global positioning system (GPS) technology now enable archaeologists to survey landscapes quickly and efficiently and make detailed plan maps of archaeological sites and features. Many GPS receivers can now establish positions to within 20 to 70 cm in a matter of seconds. The speed and accuracy of GPS receivers make them excellent tools for archaeologists. The results of a recent survey in Pãhinahina ahupua'a, Hawai'i Island, demonstrate the utility of this technology for archaeological research.

Global Positioning Systems

Global Positioning Systems (GPSs) is a positioning system operated by the U.S. Department of Defense (DoD) based on a constellation of 24 satellites orbiting the earth at an altitude of more than 20,000 km. The following discussion is based on several publications of Trimble Navigation, Sunnyvale, Calif., and one in particular: Mapping Systems: General Reference, published in 1996.

Satellites act as reference points for receivers on the ground to trilaterate their position. By measuring the travel time of radio signals transmitted from the satellites, a GPS receiver on the ground can calculate its distance from a satellite. The distance from a satellite to a receiver can be conceptualized as forming a sphere of possible receiver locations around the satellite. The actual location of the receiver on the earth is determined through the trilateration of four or more satellites, that is, the location is defined as the intersection of four or more satellite spheroids.

In the past, the use of GPS receivers by archaeologists has been limited by several error sources. The positions of the satellites and their atomic clocks are monitored and are supposed to be highly accurate. In reality, however, the satellites drift slightly from their predicted orbits, and their on-board atomic clocks are not completely synchronized. Furthermore, the satellite's transmission is disrupted and slowed as it travels through the earth's troposphere and ionosphere. In addition, as the signal arrives at the receiver, it can be reflected off local obstructions, causing "multipathing," a process that interferes with the true straight line signal. All of these factors introduce errors into the calculation of the distance between a GPS receiver and a satellite. By far the largest source of error, however, is "selective availability." This is the intentional degradation of satellite signals by the DoD and is achieved by introducing errors into the clock and orbit data transmitted by the satellites. Taken together, these sources of error cause most GPS receivers to have horizontal accuracies of within 100 m and vertical accuracies of within 173 m 95 percent of the time.

To offset these errors the process of differential correction is often used. This involves establishing a "base station" receiver at a known location, and collecting GPS positions at unknown locations with "rover" receivers. The data collected at the base station are compared with the known coordinates of the base station to determine the ever-changing errors in the satellite data. The error information from the base station is then applied to the data collected by the rovers to establish accurate rover positions. Rover receivers must be within approximately 150 km of base stations to obtain optimal differential correction. Base station data can be obtained through government organizations, in our case, the National Park Service, or through local base stations set up by a user. The process of differential correction can occur in either real time (i.e., in the field), or through postprocessing once the base and rover data have been collected (i.e., at a field station). With differential correction, the horizontal accuracy of many GPS receivers is on the order of 20 to 70 cm.

Two other recent developments have enhanced the utility of GPS for archaeologists. The first concerns the type of satellite transmissions being received. Whereas older and less expensive receivers have relied on coarse/acquisition (C/A) pseudo-random code signals, newer models incorporate carrier-phase receivers and have become much more affordable. The difference between the two involves how continuous transmissions are divided and measured. A clear explanation of the process can be found on the Trimble web page (http://www.trimble.com). The second development concerns the number of channels that a GPS receiver can monitor. Older receivers monitor a limited number of channels and must sequence through all of the visible satellites to obtain information. In newer, multichannel GPS units, multiple channels track several satellites simultaneously, allowing the receiver to monitor carrier phase signals, but more importantly for archaeologists, it allows the instrument to calculate accurate positions at a much faster rate. Whereas three years ago it would take a good receiver four or five minutes to calculate an accurate position, modern receivers establish positions in one or two seconds. The combination of increased accuracy and speed now make it feasible to use GPS to map both the morphology and distribution of archaeological features.

There are many brands of GPS receivers on the market. Some of the most affordable and accurate are manufactured by Trimble Navigation. We used two Trimble Pathfinder GPS 8-channel Pro XR receivers attached to TDC1 dataloggers. The units are capable of receiving C/A code with carrier-phase filtering and have instantaneous full wavelength carrier-phase measurement. The units were connected to compact dome antennas and were held along with rechargeable batteries in ergonomically designed fanny packs. The individual dataloggers were attached to the GPS receiver by a short cable, and information was displayed on a small 6-x-4-cm LCD screen. The interface of the datalogger consisted of an easy-to-use menu-driven program. Users input data through a keyboard, numeric pad, directional arrows, enter keys, and a range of function keys. Customized database "libraries" can be loaded into the datalogger to make data acquisition easier, but we found the "generic library" to be so simple, flexible, and easy to use that it was unnecessary to modify it. The GPS unit is easily operated by one person in the field, although two people are shown acquiring data in Figure 1. We used the Pathfinder Office differential correction software on a 75 MHZ Pentium laptop computer. We acquired two types of base station data. The first was base station data collected at Kaloko National Park on Hawai'i Island by the National Park Service, which we downloaded via a modem. As we were interested in how the GPS system would work in situations where institutional base station data were unavailable (as would be the case on some isolated Pacific islands), we also configured one of the receivers to act as a local base station and used the other receiver as a rover. The GPS rover data were downloaded from the datalogger daily and corrected with one or both of the base station data sets. We printed out maps each night on a Hewlett-Packard DeskJet 340 printer and field checked them the following day. This collection of equipment was used to rapidly complete the survey and mapping of surface archaeological features in Pãhinahina ahupua'a of North Kohala District.

Figure 1
Figure 1: The collection of data with the Trimble GPS unit.

Return to top of page

North Kohala Archaeology and Environment

North Kohala is located on the northernmost tip of the island of Hawai'i and encompasses both windward (eastern) and leeward (western) regions of the island. Archaeological research on the leeward side of the district has focused on the Kohala agricultural field system with significant contributions by S. Newman (1970, Hawaiian Fishing and Farming on the Island of Hawaii in A.D. 1778, Division of State Parks, Department of Land and Natural Resources, Honolulu, Hawai'i), P. H. Rosendahl (1994, Aboriginal Hawaiian Structural Remains and Settlement Patterns in the Upland Agricultural Zone at Lapakahi, Island of Hawaii. Hawaiian Archaeology 3:14-70), and H. D. Tuggle and P. B. Griffin (editors, 1973, Lapakahi, Hawaii: Archaeological Studies. Social Science Research Institute, University of Hawai'i, Honolulu). Recently, T. N. Ladefoged, M. W. Graves, and R. P. Jennings (1996, Dryland Agricultural Expansion and Intensification in Kohala, Hawai'i Island, Antiquity 70:81-880) have developed a GIS for the entire Kohala field system. This reticulated and essentially contiguous field system is estimated to be more than 40 km2, extending along the slopes of North Kohala for more than 19 km. Our published work has emphasized the environmental and social factors implicated at the end point of expansion and intensification of the field system, although we are now involved in modeling the variable process of agricultural change during the last 300 to 400 years. While the field system is increasingly well known and documented via aerial photographs and GIS analysis, the associated coastal settlements have only been documented at Lapakahi and a few other localities during recent contract projects. We have begun to focus on these coastal settlements to understand the variable role that gender-and elite- based social organization and subsistence practices may have played in the expansion and/or intensification of upland agriculture.

Return to top of page

1997 Fieldwork in North Kohala

A joint research project involving the University of Auckland and the University of Hawai'i at Mãnoa conducted preliminary fieldwork during summer 1997. The season's objectives were to understand the coastal settlements of the southern ahupua'a in North Kohala and evaluate the potential of GPS receivers for effectively gathering archaeological data. The coastal settlement of a single ahupua'a known as Pãhinahina was surveyed, mapped, and recorded (Figure 2). We also did less intensive work in two nearby ahupua'a (Kahuã 2 and Mãkiloa), one of which had been previously surveyed. The work in Pãhinahina represented our initial effort to integrate GPS with the ongoing GIS we have developed for North Kohala. No previous systematic archaeological research had been done at Pãhinahina; only four sites had been recorded, but their nature and locations were only partly described.

Figure 2
Figure 2: The figure displays an aerial photograph of Pahinahina ahupua'a that was georeferenced and photogrammetrically corrected using ground control points collected with the GPS unit. The ahupua'a boundaries are shown as straight black lines. The archaeological features shown as white lines and the dirt roads shown as black lines were also recorded with the GPS recorder.

The fieldwork in Pãhinahina took place over 10 days, with a crew varying between two and seven individuals, and a total of ca. 240 person hours spent on the survey. An 11.6 ha parcel--defined by the coastline, the Mahukona-Kawaihae highway, and the north and south boundaries of the ahupua'a--was surveyed, identifying 97 separate architectural features. The feature density of 8.4/ha is comparable to the density recorded in the adjacent ahupua'a Kahuã 1 and Kahuã 2, which had been previously surveyed.

Field procedures involved initial survey of the area, flagging archaeological features, assigning inventory numbers, and making basic morphological descriptions. Each feature was later revisited to establish its location and morphology with a GPS and to complete architectural recording forms. Usually, the rate at which features could be located and mapped with the GPS exceeded that of the completion of forms. The GPS data were downloaded daily to a laptop computer installed with the Pathfinder software. Maps of the feature distributions were created and printed after the data were differentially corrected, allowing us to identify potential problems in feature location and to derive solutions before the next day's fieldwork.

As this was the first time to use the equipment, a number of operator errors were identified each night. One error was in the establishment of local base stations. Initially, every time we set up the local base station, we let the receiver determine its position in real time without differentially correcting the data. Further complicating matters, we subsequently established additional base stations throughout the survey area for use in later mapping sessions. The result was that the rover positions corrected with the local base station data for each session were accurate in relative terms to each other, but skewed in relation to any other session of base station data collection. Fortunately, we had the National Park Service base station data to fall back on and use for differential correction. However, in contexts where institutional base station data are not available, the solution is to establish the approximate location of a local base station using undifferentially corrected data, and continue to use those coordinates for subsequent mapping sessions and the establishment of other local base stations.

Another operator error occurred when the number of satellites visible to the local base station dropped to a level where it was difficult to establish three-dimensional positions. We thought this could be overcome by selecting the two-dimensional option on the local base station and rover receivers, requiring one less satellite than the 3-D option. Unfortunately, this option requires an accurate elevation to be determined and entered into the local base station receiver. The error was apparent once we differentially corrected the data and noticed that the points we collected that day were significantly offset in comparison to the data from previous days. These and other instances of operator error were resolved by re-reading portions of the 10 lengthy manuals that came with the equipment. Alternatively, we could have attended the training sessions offered by the manufacturers. Once our initial problems were resolved, the equipment proved easy to use.

A major advantage of GPS survey is having the ability to make plan maps of archaeological sites and features with the instrument. This capability is not generally appreciated, but advancements in GPS technology now make it possible to create feature maps with comparable accuracy to detailed tape and compass maps. In most cases the level of accuracy provided by the GPS receiver (20 to 70 cm) is the same or greater than the discrepancy introduced by the value judgment made by the archaeologist as to where to place the GPS antenna to record a point. In Hawai'i, where architectural features dominate the surface archaeological record, our survey emphasized their location and mapping. Typically, the process of using the GPS to map smaller features simply involved walking around the feature's perimeter. For larger, more complex features, the process involved mapping associated internal or external components, such as walls, terraces, cupboards, and pits. In either case, the GPS data was used to print out a plan map of the feature each night, to be used in the field as a template for drawing in detailed architectural components. An example of mapping a relatively simple residential enclosure is shown in Figures 3 through 5. Figure 3 is an unaltered printout of the differentially corrected GPS data. This feature was mapped by walking clockwise around it, beginning at the exterior of the southwest wall and returning to the starting point. The start and end points of the exterior line match up almost perfectly, demonstrating the accuracy of the GPS unit. The interior of the feature was then mapped, with additional lines defining the doorway, interior cupboard, and facings. These procedures were completed in approximately 20 minutes. A plan map was produced and printed and taken into the field to add the details shown in Figure 4. Figure 5 was then drafted from the field data.

Figure 3
Figure 3: A printout of the differentially corrected data from a residential enclosure.

Figure 4
Figure 4: The field sketch of the residential enclosure shown in Figure 3.

Figure 5
Figure 5: A plan map of the residential enclosure shown in Figure 3.

An example of a more complicated feature, probably a religious temple or heiau, is shown in Figures 6 through 8. Again the unaltered printout of the differentially corrected data shows the location of the exterior and interior facings, pavings, and depressions (Figure 6). These components were used as reference points to sketch in the details shown in Figure 7; the final map is shown in Figure 8. It took approximately 25 minutes to complete the initial GPS survey of the feature, and another 30 to 35 minutes to draw in its details--far less time than it would have taken with tape and compass or with plane table and alidade.

Figure 6
Figure 6: A printout of the differentially corrected data from a religious heiau.

Figure 7
Figure 7: The field sketch of the religious heiau shown in Figure 6.

Figure 8
Figure 8: A plan map of the religious heiau shown in Figure 6. The entire structure is rock filled, and only the paved areas have been indicted.

One problem encountered in the mapping of features was the GPS receiver's inability to obtain data under thick vegetation. For the most part, the coastal section of Pãhinahina is grassland with intermittent 1- to 2-m-high kiawe trees and bushes. However, in certain areas the kiawe trees have grown to a height of 4 or 5 m and have trunks and branches that are over 90 cm in diameter. The GPS receiver was consistently able to record data under trees with branches up to 6 cm in diameter, but when the branches were up to 12 cm in diameter, the GPS unit would loose track of the satellites and would take from 5 to 30 seconds to reorient itself and obtain a position. It was virtually impossible to map features under trees with branches over 12 cm in diameter. The GPS unit could not properly track the satellites, and we would move the antenna away from the branches until the unit could be reoriented, delaying the process for about one minute. Even when the unit was able to obtain a position under thick vegetation, the effect of signal multipathing degraded the accuracy of the position, rendering it unusable.

In one instance, the recorded positions deviated by 5 to 7 m. In this case, we resorted to making a plane table and alidade map of the feature, which took approximately four hours. Without the vegetation, a comparable map using the GPS could have been completed in one hour and would have had the additional advantage of plotting the location of the feature in relation to all the other features in the area. It should be noted that Trimble has recently incorporated new "Everest technology" into their latest receivers that should significantly reduce the problems associated with multipathing.

In addition to mapping the morphology and distribution of archaeological features, we used the GPS receiver to gather control points to photogrammetrically correct an aerial photograph of the area. We established the accurate coordinate positions of nine control points that were clearly visible on the aerial photograph. The aerial photograph was scanned using Adobe Photoshop and then rectified with the GPS control points using Arc/Info NT 7.11. We also used the GPS to survey the jeep trails on the property, the major drainages, and the boundary markers for the ahupua'a. On returning to Auckland, the Trimble Pathfinder GPS data for Pãhinahina was easily exported to Arc/Info NT 7.11 and ArcView 3.0a GIS format, and a map of the area, showing archaeological features, coastline, drainages, boundaries, and highway, was created (see Figure 2).

Return to top of page

Conclusions

During this short period of fieldwork, we collected locational, spatial, and contextual data on 97 features in the ahupua'a of Pãhinahina. We estimate that it would have taken at least four times as long to collect comparable data using traditional archaeological survey and mapping techniques. A number of factors enabled us to collect the data in Pãhinahina so quickly and efficiently: (1) the area is quite dry, generally with relatively sparse vegetation cover; (2) Pãhinahina has relatively gentle topography, and no large mountains or valley walls inhibit the visibility of the satellites; (3) the GPS instruments generally withstood the high summer temperatures of Hawai'i (over 90[[ring]]F), although on one occasion the local base station datalogger overheated and automatically shut down the LCD display (it did, however, continue to record data); (4) the availability of the National Park Service base station data was an asset, although with two GPS units it is quite simple to set up one as a local base station to enable differential correction.

Archaeologists engaged in research and cultural resource management are increasingly using new technologies to speed data capture and develop new types of analyses. As our work shows, it is now possible to use GPS receivers to quickly and easily survey large areas and make detailed plan maps of archaeological sites and features. Distributional analyses (including seriations) of the architectural features at Pãhinahina are now underway, putting us within reach of describing the trajectory of community development, expansion, and change. Future research along the coast of North Kohala is planned, so that eventually the spatially variable role of resources and cultural practices on the organization and diversification of traditional Hawaiian agriculture and polities also will be within our analytical reach.

Acknowledgements

We thank Sarina Pearson, Roger Green, and C. Kehaunani Cachola-Abad for their helpful comments and suggestions. David Matsuda and Kathryn Kewalu assisted in the field. Melia Lane-Kamahele and Tom Fake of the National Park Service provided advice on how to download the base station data from Kaloko National Park. We also thank Jade Moniz, archaeologist at the Pohakuloa Training Area, Hawai'i Island, for her help with data acquisition. Patrick McCoy and Eric Komori of the Hawai'i State Historic Preservation Division provided useful advice and assistance. Permission to conduct the fieldwork in Pãhinahina and Kahua ahupua'a was provided by the state of Hawai'i, Department of Land and Natural Resources, and by Monty Richards and Puna Van Holt. The fieldwork was partially funded by a grant from the University of Auckland Research Committee.

Thegn L. Ladefoged and Blaze O'Connor are at the University of Auckland, New Zealand; Michael W. Graves and Robin Chapin are at the University of Hawai'i at Manoa.

Return to top of page


Previous Page Table of Contents Next Page