as a PDF
Transcription
as a PDF
A. Wagtendonk, N. de Reus 323 Development and Use of Digital Fieldwork Tools for Academic Purposes Alfred Wagtendonk, Nils de Reus Vrije Universiteit Amsterdam Spatial Information Laboratory (SPINlab) {alfred.wagtendonk | nils.de.reus}@ivm.falw.vu.nl ABSTRACT Over the last decade, fieldwork data collection methodology has fallen increasingly behind office data processing techniques. This has resulted in a gap between the form in which raw data is brought in from the field and the form in which the office systems require their input. Taking digital fieldwork techniques to the field with mobile computing offers an opportunity to reduce or close the gap, eliminating to a significant extent both the problem of expensive post-processing and the need for intermediate procedures which are a known source of transcription errors. A case study is explained in this paper wherein is tested what immediate effect the introduction of mobile computing has on efficiency, quality and reliability of fieldwork results. This paper was compiled for the Münster GI-Days 2004. It describes the expected or sought advantages of digital fieldwork techniques and what was done in the case study ‘CropSpy’, but contains no technical information for implementation. INTRODUCTION Fieldwork occupies an essential niche in such scientific disciplines as geology, biology, archaeology and others as both a means of collecting field data and an important opportunity for teaching students in the subject through hands-on experience. An inventory of such fieldwork reveals that while they differ in scope, objective and methodology, it is common that the collection of raw data is carried out in approximately the same way today26 as it was ten years ago, using pen and paper as the main recording instruments. On the office side however, where both fieldwork preparation and post-processing of raw field data takes place, procedures have changed considerably due to advances in databases, expert systems, desktop GIS software and other forms of automation. Consequently, there is a world of difference between the use of computers, databases and advanced analysis 26 Up to 2003, statistics for 2004 are not yet available 324 A. Wagtendonk, N. de Reus techniques in the office environment and the use of traditional pen and paper in the field to collect raw data as inputs for these systems. Even though field procedure can be in itself very advanced and developed in methodological sense, they usually fail in their integration with digital procedures and systems in the office, making several time consuming and error prone transcription steps necessary (i.e. digitising marks from analogue maps and introducing the hand-written field data into the database in the post-processing process). An opportunity to bridge this gap between the reality of the field and the reality of the office lies in developments in mobile computing of the last 10 years, which allow mobile computing technology to be brought into the field so that computer assisted (digital) working methods can be extended to the remote fieldwork area. As much of computer science is not only about data processing, but also about communication of data in electronic form, it is clear that the potential of computer assisted methods in the whole data and information flow between field and office can be enormous. A few examples of possible advantages are the following: x Reduction of data transcription errors by eliminating in whole or part the need to translate between digital and analogue formats x Automated uniform input (e.g. time and location stamping of measurements) with possibilities of automatic data validation during collection (Nykänen, 2002) x Indirect operation (automatic completion of fields, voice recognition, etc.) x Increase of spatial accuracy (taking coordinates straight from GPS, not always an improvement over existing method) x Possible wireless information retrieval x Field communication and data-exchange possibility for guidance and feedback from remote specialists at the office x Use and knowledge of modern field techniques is important for the university to be able to compete with other education institutions and for the career perspectives of the students themselves x Higher education yield, students are forced to keep on thinking instead of ‘blind’ recording during the day and processing data in the night (Groen, 2003). Recording data directly into a device capable of also performing data analysis means that students, as well as experienced fieldworkers, can perform preliminary analysis while out in the field to A. Wagtendonk, N. de Reus 325 find out patterns and, importantly, apparent anomalies to an expected pattern that would require additional data collection in certain areas while the team is still there. In spite of these advantages, in practice researchers often hesitate to make the step from the classical to the computer-assisted fieldwork. One of several explanations for this is that researchers, apart from all the involved work and preparation in changing and improving of existing procedures, are concerned that the introduction of computer assisted methods will have too deep an impact on the way the fieldwork is carried out and on the organisation of the fieldwork, effectively taking away their existing advantage of experience, and that the use of this technology in the field creates all kind of technology related problems distracting from the real subject of the fieldwork. Also the technology in itself, particularly those requiring new skills such as the necessary application development for the mobile GIS software, can be a barrier to start with these new methods. Last but not least researchers and teachers have expressed concern that using the computer and other digital equipment in the field prevents students from developing specific skills they need in the field, such as orientation and sketching. As was said in our own Faculty of Earth Science (paraphrased and translated); "Maps inside computers are very useful for people who already understand how to use maps properly. But early in the curriculum, we should not have to teach students about maps and computers at the same time." CASE STUDY To study and promote the possibilities of using computer assisted field methods at our own university we addressed the above mentioned problems in several case studies carried out in the framework of a Dutch project about wireless learning and mobile fieldwork with the financial support of the SURF foundation27. 27 SURF is the Dutch higher education and research partnership organisation for network services and information and communications technology (ICT). 326 A. Wagtendonk, N. de Reus The starting points for the case studies were the following: x The computer assisted field method had to be based on an existing field method to make comparison possible and to demonstrate the added value of the new method. x The method should be developed with existing ‘off-the-shelf’ technology and be easy to operate and to customise. x Specific requirements of the mobile user should be addressed x The added value of the computer assisted method has to be proven in terms of speed, efficiency and reliability The Situation The case study that we want to discuss here, the development of the ‘CropSpy’ was carried out in co-operation with a Dutch company specialised in the production of remote sensing products for among others agriculture. Through this co-operation the application and method could be evaluated by both field teams of students and professional fieldworkers who could compare the new method with the conventional method. The computer assisted fieldwork method is aiming to improve the current fieldwork method that forms an integral part of the process leading to the yearly updated crop map of the Netherlands (1:10,000). The digital crops map of the shows what crop(s) are grown on every parcel of agricultural land in the Netherlands. Crops include grass, corn, potatoes, sugar beets, wheat, barley, flax, onions, vegetables, flower bulbs, etc. The map is used by local and national government bodies, water authorities, and agri-businesses. Its uses include yield forecasts, disease and crop protection management, land administration, environmental quality monitoring (e.g. estimation of pollution loads), manure management and a variety of others. The map is compiled based on interpretation of optical and radar satellite images collected at different phases of the growing season. This automatic phase is usually complemented by fieldwork to validate results and to complete data acquisition where satellite images are insufficient. The tested ‘CropSpy’ solution CropSpy allows fieldworkers to visualise a high-resolution satellite map of the fieldwork area on a handheld. The area displayed on the device is centred on the position of the user, determined by GPS. The user can inset a point on the map and associate crop information to it, such as type, description etc. The information is associated to the entire crop parcel con- A. Wagtendonk, N. de Reus 327 taining the point. Through a wireless connection, each point added is sent to a central server in real-time. The solution is based on ArcPad and consists of three main components: x a mapping component x an entry form x a storage and exchange component The mapping component includes the data installed on the handheld and the customised mapping functionality of ArcPad for collecting point information. As the shape of the crop fields can be determined in most cases with the help of the satellite images, crop identification corresponds to adding a point – called crop point - within a field area, and associating crop attribute information to it. An entry form assists the user determining crop type and adding additional information. An application made with ArcPad Studio (selected for its relatively low learning curve to encourage use by otherwise non-technical teachers and researchers) ensures that once a point is added by tapping on the screen the map the entry form pops up to assist form identification and specification. A particular requirement concerns the possibility of adding the x,y position of the fieldworker to the crop point information. This position may be different from the crop point, for instance, when the fieldworker stands on road and verifies inaccessible fields at the roadside. This information is stored in the attribute table of each crop point and serves for data validation and quality control. As soon as the user has tapped the screen to place a crop point, an entry form appears showing seven of the most prevalent crops in the fieldwork area (see Figure 1 a. b. and c. on next page). This list is based on the frequency of each crop type in past fieldwork. If the observed crop is not among the top seven, the user can visualise a comprehensive list of crops, classified into crop families. Visual and descriptive information for each crop is also made available, and can be used by less experienced fieldworkers to aid in crop identification. 328 A. Wagtendonk, N. de Reus The user can also record comments and text information, or add a picture taken on the field if a wireless camera is available. The information collected includes: Automatically recorded Entered by fieldworker Map coordinates of crop point (x,y) Name of crop to be classified: “crop type” code GPS coordinates of fieldworker (x,y,z) GPS data quality parameters (number of visible satellites, hdop) Date and time of point recording Comment about the crop to be classified (optional) Code and file location of picture taken with digital camera (optional) Project year, map sheet, fieldworker code Fig. 1a, 1b, 1c: Quick-list, full list and view information tool. On the left, a list of crops with the most common ones presented first. The centre screen shows the categorized crop lookup form, while the right picture assists fieldworkers in the correct crop assessment. The application uses two methods for data storage. The first is the local storage of data collected, which can be transferred to a central GIS at the end of the fieldwork. To prevent data loss caused by battery failure or technical malfunction of the handheld, the application saves data on a removable memory card instead of using the internal memory of the handheld. A second safeguard mechanism employs remote data storage when a wireless network connection is available. The application allows users to send the information collected about each crop point to a remote database through a wireless internet connection (a GPRS connection). In this scenario the information is stored both locally and remote. The remote synchronisation is optional. A. Wagtendonk, N. de Reus 329 The database on the remote web server, located at the premises of the user organisation or hosted by a third party, operates as an input device for a real-time ArcIMS webmapping service that displays the crop points collected against the background maps. Even though this feature is not required for this application, it does give the possibility of exchanging data remotely between different field teams and the office. This facilitates monitoring of the whole field campaign and allows the introduction of process optimisations. Since standard communication facilities such as email and ftp are available on handheld, fieldworkers can also use the wireless internet connection for other purposes, such as to consult colleagues at the office (e.g. to send a picture of a field back to the office for expert advise on its identification), or to exchange logistics information with other fieldworkers. Fig. 2: Fieldworkers are not limited to mobility on foot alone, and the suitability of digital tools had to be tested under different field conditions. Here, a handheld device is mounted on a (device specific) support and connected to a GPS receiver with a cable). In this case the handheld has an integrated wireless internet connection (GPRS) that is always open in the background, allowing full internet access while in the field. The field test Initial testing of the application developed for this case study was done by a team of six students for five days, during two of which members of the SPINlab accompanied and monitored the teams on location. After refinement based on the student team reports, three more days of testing were done by a trained ‘cropmapping’ employee with prior experience with the analogue version of the fieldwork, again two of which were participated 330 A. Wagtendonk, N. de Reus in by SPINlab personnel. The testing days were all between June and September due to the necessity for there to be real crops in the fields for identification. Weather conditions during testing ranged from dry hot weather to rain. The findings presented are the result of all testing days. FINDINGS The computer-assisted method has been evaluated both from a business perspective and from an academic perspective. Important criteria from a business perspective were the improvements in crop-point collection speed and the end-to-end cost saving for the entire process, that is from the initiation of the fieldwork to the check-out and integration of the data collected into the crop map. Other more academic criteria included the increase of data quality, the reduction of collection and transcription errors, the availability of fieldwork planning information, optimising schedule and location of field work. Table one shows that the business requirements are easily met and that the biggest cost reduction is not in the field itself but because of the elimination of post-processing costs. Tab. 1: Summary of the results of fieldwork. The costs are expressed in percentage of the total costs of the traditional method. The mobile equipment costs (hardware and software) are estimated assuming that a device has a life span of 18 months and is in use for 50% of the time. Figures in both columns are based on the total cost of the traditional method being equal to 100%. Activity Traditional CropSpy Fields per day 400 550 Costs Traditional Digital Fieldwork 39% 28% Coordination 7% 4% Fieldwork preparation 3% 12% Post processing of field 51% data 0% Mobile equipment 4% 0% A. Wagtendonk, N. de Reus 331 100% 48% The tests carried out showed also positive results for the academic requirements, with an increase in quality control in the fieldwork, and the possibility of better managing separate fieldworkers’ teams. They are achieved by actively supporting crop recognition in the field, by minimising position error through a direct link between GPS, map image and coordinates listed in the form, and by verifying the progress of separate teams in real time. However, behind the promising results there are also several practical issues to overcome. New problems and necessary precautions were uncovered that to some extent eliminate some of the intended advantages of digital fieldwork techniques. One of the ideas behind digital fieldwork for example was that it was to reduce the need to return to the fieldwork area during post processing for verification of anomalous data. In its place however has now come a need to visit the fieldwork area, which is often remote if not at least rural, during the preparatory stage to verify the availability and quality of GPRS coverage (if such functionality is to be used). Limited battery capacity also remains problematic for those who need their application to be running for the full length of a fieldwork day, especially if there is no wall socket to recharge spent batteries from at the end of the day. On the organisational side, there is initial unfamiliarity with both the technology, and the need for exact preparations to ready the background and data files for the fieldwork area ahead of time. In the case of the CropSpy, sending out a team to get started before an up-to-date satellite image is available is now undesirable, because doing so re-introduces a need for extra post-processing work to match the collected data to the updated field outlines. As the timely availability of satellite imagery depends on whether unpredictable weather patterns allow for a clear picture to be taken, some risk is involved here. Furthermore, effect of the communication and co-ordination facilities was insufficiently tested, as due to the limited size of the testing group, the number of teams in the field simultaneously has at no time exceeded two. To draw more firm conclusions, more extensive tests with more fieldworkers have to be carried out. 332 A. Wagtendonk, N. de Reus REFERENCES Groen, M. (2003): Van potloodje en veldboekje naar DPA, GPS en GIS. Internal memorandum Faculty of Earth and Life Sciences, Vrije Universiteit Amsterdam Nykänen, V. (2002): Capturing digital data in the field - digital field data capture at the Geological Survey of Finland. Abstract and presentation in workshop on capturing digital data in the field, British Geological Survey (BGS), Nottingham, Arill 2002 (http://www.bgs.ac.uk/dfdc/ vnykanen.html)