Paper No. 6
Presentation Time: 9:25 AM


STEFFKE, Christy L., Geocognition Research Laboratory, Michigan State University, Department of Geological Sciences, 288 Farm Ln, 206 Natural Science Building, East Lansing, MI 48824 and LIBARKIN, Julie C., Department of Geological Sciences, Michigan State University, 288 Farm Ln, 206 Natural Science Building, East Lansing, MI 48824-1115,

Scientists use a variety of colors and textures to represent visual data. In the geosciences, several different color ramps have become pervasive for representing continuous value data such as elevation, precipitation, or temperature. In a recent eye tracking experiment, 23 participants were instructed to estimate values in a continuous-value digital elevation model (DEM) that was symbolized using four common color ramps. Participants were presented with the same DEM data subset symbolized using the four color ramps and instructed to estimate values at four locations with known values. These sixteen images were presented in a randomized sequence without the option to return to previous estimations in order to minimize the chance for estimation manipulation by the participants. The average difference between participant estimations and the known value was then calculated and compared with quantitative eye track characteristics measured using ArcGIS. These quantitative eye tracking metrics include total gaze path length (pixels), gaze plot length (pixels), and time spent estimating each value (milliseconds). Demographic characteristics of each participant (gender, age, ethnicity, etc.) were also considered. We found that participant interactions differed significantly as a function of color ramp, with some ramps producing widely divergent over or underestimations of data values. Comparisons of eye tracking metrics indicate that some ramps require significantly more gaze time or effort for data value estimation than others. Taken together, these data suggest that color ramps with a single hue, or ramps containing more than four colors, are the least communicative of data values. This and future studies will provide evidence for an ideal set of color ramp characteristics for communicating visual data.