Accuracy assessment of GPS and surveying technique in forest road mapping

摘要

Forest road networks provide access to the forest as a source of timber production and tourism services. Moreover, it is considered the main tool to protect forests from fire and smuggling. The prerequisite of road management and maintenance planning is to have spatial distribution and map of the roads. But newly constructed or some other forest road segments are not available in national maps. Therefore, mapping these networks is raised as a priority for a forest manager. The aim of this study was to assess accuracy of routine methods in road mapping. For this purpose, Patom district forest road was selected and road network map was extracted from the National Cartographic Center maps as the ground truth or base map. The map of the network was acquired using two methods, a GPS receiver and survey technique. Selecting 70 sample points on the network and considering the National Cartographic Center map as base map, accuracy was determined for two methods. The results showed that while the survey method was more accurate at the beginning of the path (first 500 meters), accumulation of errors resulted in higher rates of error in this method (up to 263 meters) compared to GPS. Mann-Whitney test revealed significant differences in accuracy of two methods and mean accuracies were 38.86 and 147.90 for GPS and surveying respectively. The results showed that for samples 1-15 there was no significant difference between the survey and GPS data but for samples 28-42 and 56-70 statistically significant difference were existed between the survey and GPS data. Regression analysis showed that the relation between GPS and surveying accuracies and distance were best defined by cubic (R2 adj = 0.65) and linear (R2 adj = 0.83) regression models respectively. Applying 10 and 5 meters buffers around base map, 68 and 41% of GPS and 44 and 21% of surveying derived road were overlapped with buffer zones. The time required to complete the survey was found to increase the overall cost of mapping road network. The lower time requirements associated with the GPS methods were found to reduce the costs associated with mapping forest roads.