The Development of a Predictive Hiking Travel Time Model Accounting for Terrain Variations

More and more GPS collected travel data is being shared via the internet. This is especially true with outdoor adventure travel including hiking. This research seeks to ascertain if this data can be used in the development of a predictive hiking speed model accounting for terrain variations. Fifty-two hikes represented by global positioning system (GPS) data were downloaded from an outdoor adventure travel web site to be used in the study. After identifying gross data errors several hikes were eliminated leaving 40 hikes which were used in the development of a hiking model. Regression analysis was carried out using slope and speed as the independent and dependent variables respectively. Two exponential curves were identified to model the slope – speed relationships in the GPS data, one for up slope travel and one for down slope travel. The study model developed and other speed slope models were loosely compared to observed travel times in nine additional GPS collected hikes. Spatial scale and spatial dependency in the GPS dataset was also examined in the process of the study. A geographically weighted regression (GWR) was performed on the GPS data of a single hike. Results indicated a better fit than classical regression analysis. A spatial autocorrelation analysis was performed on the residual values of the GWR. The results of the spatial autocorrelation analysis on the residuals of the GWR model on one hike showed that they are spatially dependent, which implied that other factors such as GPS accuracy impact the slope-speed model. The study shows that with some additional research and changes to data scrubbing routines used; there is a possibility for developing a better more robust hiking slope – speed model using this type of GPS data.