Digital Capture of Geographic Feature Data for Surveys

In some surveys, it is of interest to record geospatial data such as coordinates for housing unit locations, road networks, agricultural field boundaries, and water body boundaries as part of the survey data collection process. These data objects are represented as shape data (i.e., points, lines, polygons), and are inherently different from coded and fixed format questionnaire responses. Geographic information system (GIS) software can be used to record shape data, but the differences between a GIS software environment and that of a standard computer-assisted survey instrument are substantial. Questions arise in balancing the linear structure of protocol enforcement with the iterative interactions that occur when working with geographic information; how to represent and configure feature editing environments for unskilled users; and whether constraints should be placed on the digitizing process to balance the accuracy of geographical feature delineation with the time spent recording the data. GIS software environments for survey data collection were explored via investigations with two agencies in the US Department of Agriculture. One study was conducted with National Agricultural Statistics Service field interviewers, who had not previously been exposed to computer-assisted data collection. Enumerators used pen-based tablet computers and a simplified ESRI ArcView application to delineate field boundaries on digital photographs during interviews with farm operators. Data were collected from observers and enumerators on the interview experience and software use. In a second study, Natural Resource Conservation Service staff were trained in the use of GIS, but varied in the depth of their prior experience with GIS software. Participants delineated water bodies on high resolution photographs using a simple ESRI ArcView application on desktop computers. The goals were to investigate the properties of area measurements at different scales in relation to work effort and to gain information on how users approached the delineation task in this environment through observation and digital capture of software sessions. We present results from these studies and on-going software development efforts.