Modelling and Mapping Likely Soil Rutting Occurrences across Forested Areas

Abstract

This article addresses where ruts are likely to occur during in-field forest operations. This was done by inspecting high-resolution surface images across New Brunswick (NB) and elsewhere to mark where ruts have (1) and have not (0) occurred in harvested cutblocks. This marking revealed 1) where off-road operations were likely done on moist to wet and unfrozen soils; and 2) whether the ruts so incurred were water-logged at the time of imaging. Through geospatial processing of the NB-wide digital elevation model (DEM, available at 1 m resolution), the following attributes were added to each of the marked rut and no-rut locations: 1) the cartographic depth-to-water (DTW) as referenced to the nearest flow channels with >1 and >4 ha upslope flow accumulation areas (FA); 2) the topographic position index (TPI) in reference to the mean annulus elevation 50 m away from each DEM cell; 3) mean slope and curvatures within each cell-surrounding 10-m circle; 4) the terrain wetness index (TWI); 5) soil association type according to the NB forest soil map, adjusted for NB’s most recent hydrographic network delineations for waterbodies and wetlands. Subjecting these data to logistic regression analysis revealed that image-located off-road rutting occurred at about 90% probability in water-accumulating zones where TPI is <0 m and DTW is <1 m. Using slope, curvature, TWI, and soil type as additional rut occurrence predictors did not affect this zonation significantly.