Fieldwork in 1996-7

Summary reports on fieldwork activities undertaken by members of the MCFE project team and our collaborators during 1996 and 1997.

Contents

• Cottam, September 1996
• Mull, May 1997
• Canterbury
• Sweetwaters, Kenya, August 1997
• Quenza, Corsica, August-September 1997
• Seville, September 1997
• Mambila, Cameroon, July-August 1997

Cottam, September 1996

At the end of September, Jason Pascoe and Nick Ryan spent four days at Cottam in the Yorkshire Wolds where Julian Richards (Dept of Archaeology, University of York) was directing his third season of archaeological excavations on a series of multi-period farm enclosures. The main purposes of this visit were to firstly to provide Jason with some background on archaeological fieldwork and, secondly, to gather ideas and requirements for appropriate mobile computer applications.

The visit was successful, and we were able to define some initial requirements and to discuss experience of using GPS in other fieldwork projects. The requirements include: Rapid, low precision, topological and other surveys of areas surrounding an excavation site. In many cases, detailed survey may not be needed, or the purpose may be to confirm or enhance existing topographic data. Recording of finds during fieldwalking. Control of sampling during fieldwalking. As part of this investigation, some simple, and very approximate, mapping of the area surrounding the excavation site was undertaken using available equipment.

Mull, May 1997

We discovered many possibilities for time-saving automation in the data collection aspects of bird censuses. Timing and location information could be automatically extracted from an attached GPS receiver, location being especially useful in areas with poor map resources or homogeneous terrain. Speed was of the essence in many of the data collection activities and often required the collaboration of two people in order to record the data reliably and successfully. For example, one person to watch the bird through binoculars and another to write down dictated behaviour notes. Another helpful feature to novice ecologists or those new to the area would be a set of guide-notes that point out useful information about the terrain and environment.

The importance of the desktop side of the system was also illustrated. After all, data collection is not an end in itself; the data needs to be analysed to develop an understanding of the subject of observation. The data would, at least, need to be downloaded to a PC and exported to other application formats, such as a text file for importing to a spreadsheet. More advanced uses of the desktop computer also occurred to us such as the ability to share observation notes amongst many ecologists.

Finally, our basic concept of a stick-e note was substantially influenced by the experience on this trip. The original simple model of attaching textual notes to a location was evolved to support the idea of a note with many fields, any of which could be considered the anchor for the stick-e note. For example, a stick-e note with location, date, and bird fields, could be considered as having the date and bird elements attached to the location or the date and location elements attached to the bird, etc.

Canterbury

Several final year Computer Science and MSc students have undertaken projects based on related topics ranging from differential postprocessing of GPS data to context-aware campus guides.

Sweetwaters, Kenya, August 1997

In particular, Jason has been working with Kathy Pinkney, an MSc student in the Dept of Biological Sciences at Manchester Metropolitan University. Having seen our prototype fieldwork tools at work in Mull, Kathy became interested in using the technology to assist in her fieldwork in Kenya a few months later. The aim of her project was to study the behaviour of giraffe in the Sweetwaters reserve near Nanyuki, and in particular to study feeding behaviour. From the behaviour data collected in the field she hopes to be able to assess the giraffe's impact on the reserve's vegetation and to calculate a sustainable population level.

The hardware chosen to implement the fieldwork tools was a 3Com PalmPilot connected to a Garmin GPS-45 receiver. The pocketable size, notebook form factor, and easy-to-use handwriting recognition system, all made the PalmPilot a natural replacement for an ecologist's conventional paper notepad. It was small enough to be kept in a trouser pocket when not in use, and it was also tough enough to withstand a variety of harsh conditions when it was in use (such as a very dusty environment, strong direct sunlight, high temperatures, etc.). The uncorrected location data, provided by the attached GPS receiver, proved a fine enough resolution for recording the position of giraffes within the reserve.

The need to record sets of discrete contextual data (e.g. habitat conditions, location, number of giraffe present, current behaviour of a giraffe, etc.) was ideally suited to our stick-e note applications. A general forms-based interface was developed that allowed the ecologists to create form templates that were then used to create new stick-e notes where fields such as location (derived from the attached GPS receiver), date, and time, would be filled-in automatically by the stick-e note system. A set of templates was created for various data collection needs, e.g. focal giraffe studies, vegetation surveys, dung pile recording etc. The stick-e notes generated from these templates could then be viewed on a map display where the field-worker could visualise their own whereabouts in respect to all the stick-e notes that they have created in the vicinity. The user-interface was designed with the goal of minimising the required user-interaction with the PalmPilot, aiming to allow the ecologist to spend as much time as possible focused on the actual subject under observation. Features that were designed for this purpose included automatic completion of fields, automatic time-stamping of behaviour observations, and 'eyes-free' operation of some user-interface features (i.e. operating the device via tactile and audio feedback without the need to actually look at the device). This approach was very successful and enabled Kathy to record more data than would have been possible if she had simply been equipped with pen, paper, map, and stop-watch, and furthermore she could also record the data at faster speeds.

Quenza, Corsica, August-September 1997

A house and bread oven at Vandalaccio

Ground truthing of satellite imagery and aerial photography, a continuing land-use survey and checking of 18th and 19th century cadastral maps, extended earlier collaborative work with Janet Bagg (Dept of Anthropology, UKC). One of the main tasks of this year's fieldwork was to visit and record several seasonal upland pastoral settlements. Many of these are no longer used, but some of the more accessible are still occupied as summer residences by descendents of the former pastoralists, most of whom have their main homes some forty kilometres away on the coastal plain around Porto Vechio.

The system performed well despite being an early prototype with only a restricted subset of the functions intended for later versions. A large number of notes were produced during four days of visits to the upland settlements. These included details of buildings and other features, and the location and subject material of numerous photographs. The spatially located notes and photographs will be incorporated into a GIS based on a spatio-temporal database (described in an earlier paper), and used to provide a rich source of material for teaching and further research.

el Gandul, Seville, September 1997

Nick Ryan and Jason Pascoe joined David Wheatley, Simon Keay and several postgraduate students from the Dept of Archaeology, University of Southampton on an archaeological survey of a Iberian and Roman settlement at el Gandul near Seville in southern Spain. The area has already provided the source material for several MSc student dissertations at Southampton. Heavy winter rains and changes in agricultural practice in response to EC subsidies have lead to serious erosion of many Andalusian sites, particularly those where the soil is exposed. There is a local requirement to conduct rapid, large-scale surface surveys of settlements and small towns in order to gather information to support attempts to protect these sites.

El Gandul is situated some 30 kilometers southeast of Seville on the edge of a low plateau. Today, it is part of a rural estate devoted largely to olive production. At first there is little for the untrained eye to see besides olive trees and bare soil, but loser inspection reveals that much of the surface is covered with building rubble; stone, brick and tile scattered with broken pottery. Occasionally, we encountered fragments of moulded or inscribed stonework, wall plaster, and other artifacts.

Archaeological field survey involves tedious study of small objects on the ground, but it is also essential to view the larger picture. Undulations in the surface may indicate ancient roads or the remnants of buried structures. At el Gandul, two sides are bordered by tree-covered earthen banks as high as 15 meters. This enclosed area yielded more pottery and masonry for these banks are in fact the walls of an ancient town. The many varieties of pottery suggest the town flourished in the Iberian and Roman periods. In common with other sites in the region, the urban center was established by the third century BC and flourished until the late fifth century AD. Simon Keay has been studying the development of urbanism in the region. His fieldwork involves sampling surface finds to refine his understanding of the site's dating and its relationship to other sites in the area. Our task was initially to shadow the conventional recording methods used in these studies, with the longer-term goal of providing archaeologists with tools to improve their efficiency.

Recording archaeological remains and artifacts might appear to present quite different challenges from those of recording animal behavior. Archaeological material is relatively static, it can be observed from close quarters, and all but the most delicate artifacts can be handled. Although these differences affect how researchers approach their work, they have minimal impact on the software requirements for recording. Archaeological observations can be made in a more relaxed way than those of animals, but the volumes of material found in surface scatters presents equal demands for rapid recording. Consequently, we were able to use similar computer systems for these apparently quite different tasks.

The requirements did differ in that archaeologists need much better precision. Repeatability was essential as work at the site is expected to continue for several years. Clearly, this required a differential GPS (DGPS) solution. We were not aware of any free beacon services in the area, so we opted for a do-it-yourself solution. Real-time correction would have required extra equipment such as radio modems, so we chose a postprocessing approach using pseudorange data from a GPS receiver used as our base station.

One of us carried a second receiver with a battery in a jacket pocket. To this, we attached a miniature antenna on an 80x100 millimeter metal plate mounted in a hat. We also tried a more conventional approach with the antenna mounted on a ranging pole. The hat-mounted antenna proved superior, though, as it left a hand free for operating the computer and, apart from occasions when the cable became entwined in low branches, was better suited to scrambling up and down the steep banks. Our first day in the field began with a guided tour of the site getting to know the main visible features and examining the types of pottery found on the surface. Later we left the others collecting samples at predefined locations while we set up and tested our equipment.

The photographic record is an important aspect of any archaeological survey. We spent two days recording and photographing extant features and details of the local topography, including exposed walls, dense concentrations of artifacts, and possible entrance routes into the town.

We spent another day on a more wide-ranging survey of the area, visiting many ancient tombs. Some are of Roman origin but most are much earlier, several dating back to the Chalcolithic, or early copper-using, period (2000-2500 BC). Here, we made notes about each tomb and obtained many photographs, including general views of the town.

Most of our remaining time involved collecting observations from the outer parts of the town beyond the area covered by the sampling grid. Later, we fed these records into the growing collection of background information about the site, which helped complete the story told by the quantitative sampling.

One final task was to visit each of the surface-sampling points. We filled in an electronic version of the paper forms used to record the quantities of pottery and other material found in a 2-meter radius. Apart from shadowing the paper records, this would provide a check on our postprocessing accuracy because the location of each point was known from the survey grid.

On our return to Canterbury, we transferred the field data to our desktop computers for display and analysis. When the postprocessing software was ready, we corrected our mobile locations using our base station data. Later, we checked both the base station and mobile records against IGS RINEX data from the regional data center at the Institute for Applied Geodesy in Leipzig, Germany We used the sample records taken at known points on the grid to double-check our postprocessed results. We were pleased with the results, which revealed that we had consistently achieved better than 10-meter accuracy and 5 meters.

Mambila, Cameroon, July-August 1997

MCFE Home

Fieldwork in 1998