E. B. Banning And Philip Hitchings 2015-09-18 01:39:07
E. B. Banning is Professor of Anthropology at the University of Toronto. Philip Hitchings is a Ph.D. candidate in the Department of Anthropology at the University of Toronto. The use of tablet computers in archaeological research has been gaining momentum as tablet computers become more available and affordable and software applications become more sophisticated and user-friendly. Software for databases, mapping, GPS, and other tasks is widely used on tablet computers and other handheld devices, and many archaeologists have now experimented with paperless recording in the field with various positive and negative experiences (Cascalheira et al. 2014; Fee et al. 2013). The introduction of relatively inexpensive yet powerful tablet computers to the civilian market opened the door to this experimentation. The portable, lightweight design of tablet computers makes them ideally suited to the task of paperless recording, and, to date, archaeologists have taken advantage of these features primarily in excavation contexts. Some of the advantages of tablet computers and paperless recording are readily apparent. Collection of data in a digital format alleviates the need for subsequent digitization, a process that inevitably increases the risk of various sources of error. Additionally, the ability to record and recall information “on the fly” is greatly enhanced, and the same device can be used to collect, store, and present a great variety of data, including, but not limited to, coordinates, photographs, maps, database records, notes, and voice recordings. Data from several devices can be uploaded to a main computer and merged. Later, the merged data can be downloaded to the devices again so that all team members have the whole project record at their fingertips. However, this relatively new method of data recording in the field is not free of its own difficulties. We experimented with paperless recording during fieldwork on the Wadi Quseiba project in northern Jordan, and our experiences shed light on some of these difficulties but also on ways to mitigate them and take the utmost advantage from tablets. Unlike previous case studies involving tablet computers in archaeological excavation (e.g., Fee et al. 2013), we made exclusive use of tablet computers and digital recording in archaeological survey over two field seasons. IPads on the Wadi Quseiba Survey, Northern Jordan A number of factors motivated us to experiment with the exclusive use of tablet computers, in our case, Apple iPads, for documentation of our survey of Wadi Quseiba and two smaller wadis in northern Jordan to search for evidence of late prehistoric occupation and activity. Among these were that tablet computers would enable us to efficiently record multiple types of archaeological data, eliminate the need to digitize information later in the lab, and allow us to rapidly compile and compute data to facilitate an experimental survey method that required us to update a GIS predictive model on a daily basis. These data included, but were not limited to, the lengths of transects, artifact densities along those transects, surveyor sweep widths (discussed in further detail below), survey coverage, and the continually updated probabilities that different spaces in the survey area contained undetected late prehistoric sites. One of the most immediately recognizable advantages of shifting from traditional record keeping and data collecting to digital recording methods was that it eliminated the need for each individual team member to carry many separate pieces of often cumbersome equipment during field-walking (Figure 1). Eliminating the excess weight and unwieldiness of several pieces of equipment while surveying for many hours in the heat of the day was certainly a blessing during the Wadi Quseiba project, considering that temperatures in the Jordan Valley in the middle of August often exceeded 40°C. With the addition of the weight of artifacts and samples that we collected as the day’s survey progressed, any respite from carrying equipment and supplies— forms, clipboards, cameras, GPS units, pens/pencils, a variety of paper maps— was more than welcome. Tablet computers combine the capabilities of many of these things into one portable, lightweight instrument. The iPad’s built-in camera and on-board GPS, which we found to provide both horizontal and vertical accuracy similar to that of a handheld GPS unit in field trials, eliminate two of the largest and perhaps most cumbersome pieces of equipment that an archaeological surveyor must generally deal with. The photo quality achieved with the tablet’s onboard camera proved more than sufficient for documenting both the terrain features under investigation and most of the archaeologically relevant material that team members located within that terrain. As in excavation contexts, wherever we required highly detailed photographic records to document important sites or materials, a high-quality digital camera was also available, carried by one team member only. The majority of the time, however, this simply was not required and the tablet’s onboard camera more than sufficed for the task at hand. It also had the substantial advantage of allowing us to capture images directly into the project database (Figure 2). The use of iPads also made it unnecessary for team members to carry paper maps to display terrain features or target areas within the survey region. Many software applications, including Google Earth and GIS software such as iGIS, Wolf GIS, and others, allow viewing of locations, tracing of paths, and marking of points considered archaeologically significant or valuable on digital, georectified, and often queryable maps. During the Wadi Quseiba survey project, which employed a predictive model created within a GIS environment, maps displaying each day’s targeted survey areas could easily be uploaded to the tablet computers so that any team member could access an up-to-date version of the survey’s “plan of attack” in different geographical locations within the survey universe. As we discovered, one difficulty is that data connections are needed to load and display maps within Google Earth or other mobile GIS applications. Yet in certain parts of our survey region, such as deep valleys, there were no ready connections to a mobile data network, making the display of dynamic maps impossible. In these cases, however, we were able to use static maps from the last logon or even screen-shots taken from a desktop computer in the lab to display targets in the survey universe on team members’ iPads so that they could identify and navigate to and from targeted survey areas. Standard note-taking software, along with note or comment fields in a database application, allowed instant digital input of observations about any aspect of survey without the need for pens, pencils, or clipboards. Above and beyond the simple convenience of eliminating extraneous equipment, displaying maps, or taking quick notes are several important advantages that come with the availability of quick, on-the-fly entry of digital information in a database. This eliminates many potential sources of error commonly encountered during the transfer of information from paper forms to a digital database. Instead, all of the information is input directly into the database as work is carried out in the field. For our purposes, we made use of a well-known piece of database software, FileMaker, which exists in both a desktop and a mobile version (known as FileMakerGo). The mobile version, loaded onto each iPad carried by survey team members, allowed us to record GPS coordinates directly from the iPad’s onboard GPS, while also quickly recording sites, objects, survey transects, and landscape elements of all types encountered during survey. In addition, this database has picture fields, which allowed us quickly and easily to enter photographs of sites, transects, artifacts, or landscape elements directly into the database from the onboard camera (Figure 2). Our survey design required us to collect several specific types of data that needed immediate analysis before we could implement it. The details of the survey methodology for the Wadi Quseiba Project will be published elsewhere, but a brief description is necessary to show why this was the case. In our attempt to maximize the efficiency of locating hard-to-find archaeological materials in the modern landscape, our survey methods relied on an iterative predictive model that, by way of GIS and remotely sensed imagery, accounted for the dissection of the landscape since the Neolithic period and targeted areas in the modern landscape that are likely to be remnants of the landscape as it existed in prehistory. For each of these remnants we made initial estimates of the probability that they contained Neolithic materials. From here, we used a Bayesian allocation algorithm at the end of each day to revise these probabilities in light of our findings to date and to allocate how much additional search effort, if any, should be assigned to those portions of the landscape on subsequent days. We carried out this allocation process every day or two in the field so that our predictive model was constantly updated, and our search effort focused on those places that were most likely to yield positive results. In order to accomplish these tasks, however, certain kinds of data on a day-to-day basis were essential. Recording whether or not we found anything was the easy part, but revising our estimates of coverage for each surveyed space required good estimates of the total length of transects walked, which we could measure from the beginning- and end-coordinates of each transect segment and from our sweep widths, which we estimated with “calibration runs” over fields seeded with artifacts in known locations, using methods similar to those of Banning et al. (2011) (Figure 3). Even with iPads, this process was somewhat involved, as we needed to upload data from each iPad at the end of each day, merge these data into a master database, calculate proportions of survey areas covered that day by dividing the coverage (product of effective sweep width and total length of transects) by the area of each survey area, and input this into the algorithm to determine how much further survey we should do, if any, in each landscape element. This somewhat complicated procedure was made possible only by the fact that each day’s recorded data was already available in a digital form ready for download and manipulation. Adding the extra step of transferring data from paper record to digital forms before this process could commence would have been prohibitively time consuming and would have rendered this novel survey method impossible. Ideally, we would be able to document our coverage of survey spaces in detail and use this information to update the posterior probabilities that each landscape element in our survey universe contained significant remains of the target periods (Epipaleolithic, Neolithic, and Chalcolithic). To this end, the required calculation of the length of transects— or, more accurately, transect segments, since each transect walked by individual team members could be broken into pieces of a particular length due to obstructions (trees, shrubbery, stones, crevices, etc.) or changes in direction on the terrain features being surveyed— was rendered quite simple by way of the digital recording methods. Here, a certain field in the database could be dedicated to the calculation of the distance between two coordinates. Start and end coordinates automatically entered into separate fields from the onboard GPS could be used as inputs for the calculations attributed to that field, and the results displayed. Knowing this distance, and having it calculated on the fly, allowed us to generate the proportion of each survey area that was effectively covered by each surveyor. This proportion was instrumental in the decision-making process for further allocation of survey effort, in particular, target areas within the survey universe. Should a particular survey area have seen only very minimal proportional coverage, this would be readily apparent and further survey effort could be allocated to that area accordingly. These types of quick calculations and their attendant consequences on how work was divvied up between survey areas throughout the seasons were made possible by the use of tablet computers for digital recording. In order for these proportions to be calculated, another piece of information was required in addition to the lengths of each transect segment walked by team members: the sweep width. Sweep width refers, in the most basic sense, to the effective width of a “band” perpendicular to the surveyor’s line of travel within which he or she can be thought to reliably spot artifacts (Banning et al. 2011) . The width of this band changes depending on a number of factors, including the expertise of individual surveyors, the height of their detection devices (eyes, in this case) from the ground, the obtrusiveness of the artifacts themselves, the terrain, groundcover, lighting conditions, etc. In order to calculate these, several experimental runs were established and executed in order to calculate an average effective sweep width among team members in different terrain conditions. These sweep width calibrations also took place with the help of the iPads (Figure 4). Transects were walked, with surveyors noting the distances along their line of travel and perpendicularly away from that line of travel where they thought they spotted artifacts of different classes that were created and planted by team members as stand-ins for real artifacts. These “calibration runs” were carried out in conditions that were thought to be close approximations to the conditions that would be encountered during survey. During these calibration runs, surveyors noted start and finish times in order to calculate the time spent walking transects of known distance (generally 150 m), time of day, and any positive identifications into the digital calibration database. This digital information was then easily manipulated using statistical software to come up with the appropriate sweep widths for the team, given different terrain and visibility factors. Coupling the results of the effective sweep widths of the survey team in different conditions with the length of transects walked during actual survey allowed for an area of coverage to be calculated and, after dividing this number by the area of target survey areas, the proportions of coverage could be calculated. Such a detailed survey strategy would not have been possible without the easy manipulability afforded by the digitally recorded data. Additionally, we were also able to include such data as counts of different classes of artifacts along each survey transect. Tracking and recording these counts was accomplished so that artifact densities could be readily calculated in order that the identification of significant artifact clusters perhaps not recognized in the field could be identified in the lab. These were recorded on the tablet computers with the help of “counter buttons” in the File- MakerGo database, which, when pressed, would increment the number of certain artifact classes. These incremental counts were automatically associated with a particular segment of each transect, of which the particular length and sweep width were known. This made the calculation of artifact densities and thus the possible identification of archaeological sites efficient. Thus, more importantly than the ease afforded by the elimination of several pieces of equipment, the digital recording method allowed, in our case, for the daily calculation of important metrics crucial to the experimental survey being conducted that would have been difficult to accomplish, from the point of view of time, effort, and efficiency, had we been using a traditional paper/map survey. These include the coordinates of the start and finish of each transect or transect segment, the length of transect walked, and the effective sweep width of each of those transects as they were being walked. Given that the survey methods we were experimenting with required all of these pieces of information to be readily calculated and input into spreadsheets containing mathematical algorithms for directing our survey and allocating our effort in the field as efficiently as possible, having them calculated in the field at the touch of a button and easily exportable and importable at the touch of another button for inclusion in these spreadsheets was invaluable. As alluded to earlier, however, the adoption of digital recording methods was not without its own set of difficulties, which we became aware of through our use of tablet computers as our primary recording devices in the field. While the disadvantages were few, and were outnumbered by the advantages that the tablets provided, especially given the survey methods that we employed, they certainly served to keep the field crew on their toes. One of the most glaring disadvantages of the use of tablet computers in the field was exactly that— glare. Often in the bright sun so prevalent in the Jordan Valley, the glare on the screen made the content of pages in the database difficult to read. Surveyors ended up having to spend time attempting to position themselves so that they could block the sun with their own bodies and provide appropriate shade to see the screen well enough to input their survey data. Much of the screen was rendered almost invisible in the direct sunlight, including the buttons that allowed for counts of different types of artifacts and the digital keyboard needed for taking notes on the terrain being surveyed or the artifactual evidence encountered. While this may not seem like too stressful a situation on the face of it, having to contend with this problem over a period of six or seven working hours on a very hot day can and did become quite frustrating for several of the crew members. Another disadvantage was the difficulty that some crew members had in manipulating the digital keyboard on the iPad and the buttons created for data input within the database itself, even when they were completely visible. Such minor annoyances included the autocorrect function when trying to take notes, accidental zooming in on a page when trying to select a field (which occurred often when a button was not responsive at first, prompting team members to tap it again, effectively “double clicking” an area on the page and zooming in accidentally), difficulty getting the buttons to respond to finger taps, and difficulties navigating the different pages set up for transects, sites, and so on. Some of these issues were, of course, due to the glare mentioned above. Other times it was simply unfamiliarity with the tactile responsiveness (or lack thereof, in some cases) of the tablets themselves that led to these issues. In either case, these, too, often became annoyances for some crew members. Because of these difficulties, typos in notes and the input of certain types of data, such as coordinates, transect numbers, and polygon numbers, in the incorrect fields was not uncommon. For this reason, even though not nearly as much transfer of data was required as would be from a strictly paper to a digital form, some level of corrections after exporting tables from the iPads was still necessary. However, as mentioned earlier, such corrections could be made in a much more timely fashion due to the original digital nature of the data. Simple cutting and pasting of data from the incorrect field into the correct one was certainly much more efficient and less time consuming than transferring data from one format to another would likely have been. The most serious difficulty that we encountered was the loss of small portions of data due to either human error or errors on the part of the applications or hardware being used. The final cause of the loss of some transect data— that is, start and end coordinates of transect segments, as well as artifact counts along those transects— has yet to be fully discovered. However, one hypothesis is that the overheating of the machines in the hot Jordanian sun may have resulted in some glitches that did not allow the information to be stored correctly. Another hypothesis is that the accidental pressing of buttons by the surveyors, or errors in the transfer of data from the iPads to the main hub computer, may have been the cause. In any case, somewhere along the line between accumulating, recording, transferring, and viewing certain pieces of data collected on the tablet computers in the field, some digital information did indeed go missing. The likelihood that this was human error is high and is the digital equivalent to the possibility that human error could lead to the loss of data in a paper-based recording system for survey. Forgetting sheets of paper in the field, mis-recording information during data transfer, and failing to record certain pieces of information in the first place are sources of error that can— and do— result in the loss of data when paper recording is the main method of recording data. Thus, although it was a disadvantage, loss of data was not a reason for the researchers to reject digital recording methods altogether. In sum, there were several reasons why the move to a fully digital data collection and recording method was favored. Not least of these was the necessity of efficiently implementing daily survey results in an experimental method that resulted in the allocation of survey effort based on information gathered in previous days. This information needed to be easily manipulated so as to carry out the process of allocating survey effort. Despite several difficulties encountered during our trials with this new digital recording method in the field, such a process would not have been possible without it. The advantages of using tablet computers to implement a digital recording method in the field thus far outweighed any difficulties we may have encountered, including the difficulties posed by the sun and the intermittent loss of small portions of collected data. Moreover, the most serious of these difficulties, the loss of data, has largely been dealt with as our expertise with both the hardware and the software has increased during our seasons of fieldwork. Only with the use of tablet computers was such an experiment, which produced exciting results, possible. References Cited Banning, E.B, Alicia L. Hawkins, and S.T. Stewart 2011 Sweep Widths and the Detection of Artifacts in Archaeological Survey. Journal of Archaeological Science 38:3447–3458. Cascalheira, João, Célia Gonçalves, and Nuno Bicho 2014 Smartphones and the Use of Customized Apps in Archaeological Projects. The SAA Archaeological Record 14:20–25. Fee, Samuel B., David K. Pettegrew, and William R. Caraher 2013 Taking Mobile Computing to the Field. Near Eastern Archaeology 76:50–55. Introducing the SAA Knowledge Series A New Member Benefit Launching in September 2015! Dr. Kenneth M. Ames, Current Perspectives on Complex Hunter Gatherers Dr. Jeffrey M. Altschul, Cultural Heritage Management in Developing Countries: Challenges and Opportunities for Archaeology and Archaeologists Dr. Luis Jaime Castillo Butters, Flying in Llama-Land: Aerial Photography using Drones, Photogrammetry and 3D Modeling of Archaeological Sites and Features—Experiences from the Field Dr. Paul Andrew Mayewski, Timing Is Everything: Humans and Climate Change For more information, visit saa.org/AbouttheSociety/KnowledgeSeries/tabid/1525/Default.aspx Registration opens 30 days before the presentation SAA Online Seminars Professional Development the Easy Way! Visit www.saa.org for detailed information, registration, and schedules for these upcoming courses: • Introduction to Digital Repositories for Archaeological Materials: tDAR (the Digital Archaeological Record) • The National Science Foundation Archaeology Program: Insight into the Grant Application and Award Process • Archaeological Curation for the 21st Century • Fundamentals of Budgeting for CRM Projects • Geophysical Remote Sensing in Archaeology: An Overview and Practical Guide for Beginners and Intermediate Users, Teachers, and Consumers • Archaeological Applications of Airborne Laser Scanning Check saa.org for course updates and additions!
Published by Society for American Archaeology. View All Articles.