MAPPING DISTURBANCE FROM SELECTIVE LOGGING IN TROPICAL FORESTS OF THE YUCATAN PENINSULA, MEXICO

Background. Mapping selective logging impacts on the Yucatan Peninsula is important to pursuing carbon emissions reduction and biodiversity conservation goals. Objective. To evaluate the effectiveness of applying remote sensing techniques using LANDSAT 8 OLI imagery to detect tropical forest disturbance from timber harvesting in four communally managed forests (ejidos). We further assess differences among them in terms of implementing improved forest management (IFM) and reduced impact logging (RIL). Methodology. Vegetation indices were calculated, and forest cover classification was performed to map logged and unlogged forest and specific harvest disturbances (e.g. felling gaps, skid trails, logging roads and log landings) in annual cutting areas of 2014. Accuracy assessments were conducted based on validation points collected in the field after logging. Results. We found that 75% of the binary classifications (logged and unlogged forest) had mean overall accuracies greater than 60%, representing a fair (40 to 70%) accuracy, although mapping of specific harvesting disturbances had poor accuracy (<40%). Vegetation indices that performed the best were normalized vegetation index (NDVI), Tasseled Cap Greenness and Tasseled Cap Wetness. Ejidos that applied IFM and RIL impacted a smaller percentage of their cutting areas and less area of forest per cubic meter of timber extracted, despite similar or higher logging intensities than ejidos without improved practices. Implication. Monitoring selective logging disturbance is important to improved forest management and certification of sustainability.  Conclusion. Mapping and monitoring impacts from selective logging by forest managers and technicians can be performed in a cost-efficient manner using LANDSAT 8 images, although accuracy could be improved with higher resolution imagery.