"To make a discovery is the dream of most [sports divers]. A virgin wreck is a high-class trophy. It is also the first and last chance to record the scene in a pristine state."
John R. Halsey
In 1991, the National Park Service identified a critical weakness in carrying out its mission: although responsible for preserving archeological sites and artifacts, park managers didn't know where most of them were. Only about 2 percent of the more than 360 Park Service areas had been systematically surveyed for archeological materials; more than 80 percent had not been surveyed at all. The situation was worse for submerged land: considering the two-and-a-quarter-million acres managed by NPS in 65 park areas, there had been one systematic survey published, for Drakes Bay at California's Point Reyes National Seashore. Many sites had been documented, but systematic inventory projects were not feasible.
The situation changed with the initiation of the Systemwide Archeological Inventory Program, an ongoing survey funded by Congress since 1992. The program's goal is to locate, evaluate, and document archeological sites and artifacts both on land and underwater. An important aspect of SAIP is to recommend strategies for preserving archeological materials in place, in addition to managing and interpreting them in ways parks have traditionally done. Each NPS region was to develop its own plan to implement SAIP. The submerged cultural resources unit—the Park Service underwater archeological team, established in 1980—was asked to create a model submerged site survey method adaptable to different regional environments.
Submerged archeological sites are less known than those on land—they really are "out of sight, out of mind." Underwater archeology is a new discipline, getting its start in the late 1950s. Because few systematic surveys had been conducted anywhere for submerged remains, there was no previous work to use as a guide. So SCRU decided to select a single park to develop the method, and then cross-test it in other parks.
Site managers at Florida's Dry Tortugas National Park enthusiastically agreed to host the pilot project. In the Gulf of Mexico about 68 miles from Key West, the park contains seven small sand islands surrounded by shallow coral reefs, encompassing 100 square miles of submerged lands. The Dry Tortugas, discovered and named by Ponce de Leon in 1513, lie on the western edge of the 75-mile-wide Florida Straits, through which passes the north-flowing Gulf Stream and the major shipping lane connecting the 1,200-mile Gulf coastline with the northeast United States and Europe. The straits have served as a corridor of commerce since the days of Spanish conquest, and as their western terminus, the Dry Tortugas have claimed many a vessel that ventured too close (the park is what archeologists sometimes call a "ship trap"). In the 19th century, the construction of Fort Jefferson—the largest of the masonry harbor forts built prior to the Civil War—demonstrated the strategic importance of these islands to the young nation. With its half-mile perimeter of fifty-foot-high, eight-foot-thick walls, the fort established America's presence in the Caribbean and denied access to the Tortugas' safe anchorages for any enemy fleet attempting to blockade U.S. ports.
Commercial development and military activity have left numerous wrecks here, as have the competition among nations in the New World, the construction of the fort, fishing, and other activities. Consequently, the park contains a remarkable collection of more than 250 documented wrecks over a long period, representing an important international maritime heritage about which little is known. The isolation has minimized looting damage; this, plus the fact that summers bring predictably favorable weather, made a compelling argument for developing the SAIP model here.
Laying the Groundwork
The first step was to assemble what was known about the area. A comprehensive study examined the environmental and historical context, reviewed past work, located wreck and maritime casualty reports, and also presented some recent evaluations of sites known to the park. Dry Tortugas National Park Submerged Cultural Resources Assessment, published in 1993, defined the research framework for the survey, which began the same year.
The survey would be regional in scope, systematic, comprehensive, and cumulative. It would also employ minimum-impact evaluation techniques. Most archeologists have approached shipwrecks as isolated sites disconnected from a larger context. A single wreck is usually evaluated in terms of its preservation and research potential, then excavated to recover all the artifacts. Academic investigations are normally descriptive accounts of methodology, recovered cargo, and other material; it is a rare report that goes beyond description to discuss a wider context.
Generally, studies done to assist in managing public lands, even though most are not excavations, parallel the academic approach: questions of "what" and "where" are usually paramount. NPS managers, on the other hand, are not just concerned with the what and the where, but also with historical significance, with how and why an event occurred. Thus the Dry Tortugas project approached wrecks and other sites such as groundings (where ships ran aground but were later released by the tide) as the result of many complex environmental and cultural factors, not simply as random accidents. The primary assumption was that a maritime site reflects cultural patterns in its unique features as well as in the features it shares with similar sites throughout a region. The concentration of casualty sites was approached not as a haphazard conglomeration of unlucky vessels, but as a representative sample of all maritime activity in the area.
The numerous Tortugas wrecks represent a cross-section of multi-cultural interaction throughout the Gulf and north Caribbean. From this perspective, site interpretation is ultimately directed toward understanding sociocultural patterns responsible for the wrecks. Investigators analyze processes of culture change, commercial expansion, and conflict and competition among maritime nations operating in the region. The region's development, in turn, is looked at as an antecedent to today's economy, which is largely a product of maritime activity. The idea is that the shipwrecks in this centuries-old crossroads can provide important clues to the evolution of the global economic system we live in. This approach is way beyond merely making additions to the historical record.
Forward in Reverse
The survey was done in reverse of most projects; that is, the products were designed before the method. Both park managers and researchers told us that an integrated database that was comprehensive, cumulative, and computer accessible would be the most useful. From the outset, the survey was planned to produce data in a Geographic Information System format, which allows integration of disparate, spatially related data sets. In 1992, when the survey was planned, GIS was not widely used. Recent advances in computers allow GIS to be used on PC-based hardware instead of dedicated workstations that require a specialist, which makes it accessible to small parks. A singular strength of GIS is that multiple data sets can be readily compared, with additions to the database easily made.
Because no off-the-shelf survey system was available, SCRU designed one specifically for this project (in fact, this was one of the first underwater remote sensing surveys to be designed specifically for GIS). Dry Tortugas, like many places where the unit works, is quite remote, with little or no technical support. In remote coastal areas, often there are no survey control monuments (ground control points established by professional surveyors) that allow positions to be tied to geographic coordinates. As a result, the positioning and navigation system has to be capable of geodetic survey as well as real-time accuracy. The system has to be simple, portable, hardened for a rough environment, and completely self-sufficient, utilizing readily available equipment to minimize development and procurement costs. It has to accommodate various suites of remote sensing instrumentation, including a magnetometer (to locate ferrous material representing archeological sites), a survey fathometer (to measure depth), a sub-bottom profiler (to record stratigraphy below the seabed), a RoxAnn bottom classification device (to characterize the sea floor), and—in some applications—side-scan sonar (to render seabed topography and materials on and above it). This variety of sensors, which can be deployed as required for both natural and cultural resource surveys, provides a cost-effective solution that works in a range of park areas. Finally, the hardware has to install quickly and easily in small boats. Typically, SCRU operates out of NPS crafts of opportunity, which are generally less than 30 feet long.
Accurate, real-time positioning is a basic requirement for seaborne survey projects. In shifting waters, submerged site surveyors must know their exact location from moment to moment to ensure systematic coverage of an area. Unlike terrestrial surveys, hydrographic surveys have no landmarks; simply put, it is very difficult to occupy and then reoccupy the same point and to know exactly where you are at all times.
By combining GIS with the satellite-based Global Positioning System, we solved this problem . . . almost. GPS can provide relatively accurate positions at one-second intervals. But due to the Department of Defense's policy of "Selective Availability"—along with other factors natural and manmade—accuracy is reduced to about 100 meters. In order to obtain the 2-meter accuracy that we require for archeological survey, we had to "differentially correct" the raw positions to remove inaccuracies.
A base station, set up on land at a geodetic GPS control point (which has an accuracy equivalent to a professionally surveyed monument), delivers the highest precision. The base station receives the satellite signal, corrects it, and re-transmits to the survey vessel. Corrected base station signals can be picked up from U.S. Coast Guard navigational beacons and some commercial sources. Due to the remoteness of the Dry Tortugas, however, SCRU had to set up its own station.
The Maiden Voyage
For its maiden voyage, the Archeological Data Acquisition Platform was deployed on a 25-foot fiberglass outboard. To ensure systematic coverage, vessels must use GPS positions to follow pre-plotted transects, or lanes, typically within a rectangular area several kilometers long. Survey data are collected on 30-meter lane spacing with each 1-2-second data point coordinating the GPS positions with the time, depth, heading, speed of the vessel, and various instrument readouts. We've found that 30-meter lanes provide cost-effective magnetometer coverage with very high probability of locating most materials related to colonial-period shipwrecks—the most difficult to find because of their small size and few large iron objects.
The onboard computer, following the preplotted course, tells the vessel pilot where he is relative to the desired transect. Position and course information, generated by the onboard GPS receiver, is sent to a steering display on the helm showing cross-track error, speed, course, distance to end-of-line, and bearing to end-of-line. Data from towed sensors are transmitted to the computer with the GPS information and logged to disk for later GIS analysis.
This system produces impressive, comprehensive coverage for submerged lands. Few terrestrial methods compare. However, its real utility lies in being able to manipulate the data with the GIS software. In this case, SCRU developed procedures to produce archival-quality information in the most general format available, ASCII, since no protocol existed at the time. ASCII files can be imported into most surface modelling modules and CAD software. From there, bathymetric and magnetic contours are easily produced, manipulated, and analyzed. The magnetic contours, in turn, can pinpoint concentrations of cultural materials.
Natural resource data for benthic survey was produced by the RoxAnn Bottom Classification Device, which indicates the sediment type of the seabed—sand, coral, or rock. This data, presented as a map, can become part of the park's GIS database. Aerial photographs, bathymetery readings, and other GIS layers can then be overlaid on top. The result is a picture of the environmental context, for both archeological sites and natural resources. The RoxAnn device considerably enhances cost-effectiveness; natural and cultural data sets are collected simultaneously, with no additional costs beyond that of the instrument. A recent multiple-resource survey of Yellowstone demonstrated the success of this approach.
Once the entire suite of survey data is collected, sites can be easily located by examining the GIS layers. When a promising area is selected, usually based on the magnetic data, a site can be located by loading its computer-generated coordinates into a portable GPS receiver and following the range and bearing to its position in the water. Archeologists arrive at the site with accurate environmental information in hand. Once there, they can visually inspect it for cultural materials. The data they ultimately collect varies from a determination of "nothing found" to a large volume of information, including maps, photographs, video, and photogrammetric documentation of sites and features. All data are collected in a manner retrievable through the GIS.
In the Dry Tortugas project, we scanned in aerial photographs and historical maps (some dating from the 1700s) and overlaid them on the GIS data. The aerial overlays help spot "ship tracks," scars left on the bottom by vessels that ran aground, dumped ballast, and then continued on when refloated by high tide. Most of these sites are non-magnetic and thus easily overlooked by researchers relying solely on magnetometers. Because the Dry Tortugas are sand islands perched on a Pleistocene reef, they have moved, reshaped, disappeared, and reappeared over the centuries. Many have had different names at different times. Laying old maps over current ones shows changes in both topography and terminology. Static cartography becomes a dynamic.
Getting the Big Picture
We operated the system in the summers of 1993, 1994, and 1995, completing more than 1,500 linear miles of survey lanes to cover about 27 square miles of the park. Most areas we surveyed were in depths of less than 30 feet, where wrecks are likely to occur (the rest of the park will be randomly sampled).
In some places, the density of magnetic anomalies surpassed 200 per kilometer, indicating many artifacts and much intermingling of sites. This has altered some of our notions about shipwrecks; here, they are rarely discrete. Rather, there are submerged areas of greater and lesser concentrations of material, reflecting multiple events over time. In one area, for example, there are at least four shipwrecks from the 1870s or so—some showing evidence of salvage—overlaying an area where cannons were dropped perhaps half a century earlier, all close to where another vessel ran aground and offloaded two piles of ballast stones to release itself at high tide. We are treating spots like this as maritime casualty and activity areas. Thus we are learning to seek different evidence about the past than if we had dealt with each wreck separately. GIS is basic to this approach because it can track complex relationships over wide areas.
GIS has already proven to be a very powerful tool on land; now, with SCRU's adaptation, it can inform those whose knowledge ends at water's edge. It is difficult to overstate the importance of cumulative databases whose variables can be easily manipulated; they have revolutionized the management of land in the National Park Service. Seaborne GIS inventory programs promise the same for submerged sites.
For more information, contact Larry E. Murphy, Archeologist, NPS Intermountain Field Area, Submerged Cultural Resources Unit, P.O. Box 728, Santa Fe, NM 87505, (505) 988-6750, fax (505) 988-6876.