Estimating forest restoration treatment effects on forest structure: Multi-temporal UAV SfM model and multispectral image analysis

Abstract

The ponderosa pine (Pinus ponderosa) forests across Arizona have undergone significant changes in their structure and composition since European-American settlement in the 1860’s. At present, Southwestern ponderosa pine forests are dense with homogenous tree stands rendering the forest more susceptible to catastrophic wildfire and insect outbreaks. In an effort to mediate the present and future impacts of overcrowded forests, the Four Forests Restoration Initiative (4FRI) was formed in 2010 to plan, implement, and monitor restoration projects across a 2.4 million acres of federally managed forests. Restoration treatments are now well underway and there is a need to rapidly monitor and assess the changes to better inform adaptive management decisions. This study aimed to provide an accurate, cost-effective, and timely solution for forest restoration treatment monitoring by using multi-temporal high resolution multispectral images and structure from motion (SfM)-derived point cloud data from a fixed-wing unmanned aerial vehicle (UAV). Images were acquired before and after a forest thinning treatment. We estimated changes in stand-level canopy cover, tree density, forest patch and interspace characteristics, and quantified individual tree locations and characteristics. Results from a 93 ha restoration treatment study site indicate 42% reduction in tree density, and a 28% reduction in canopy cover from 45% to 17%. We discuss considerations related to data acquisition, SfM algorithm parameterization, and the effectiveness in deriving specific forest structure metrics for restoration planning, which can substantially affect the processing workflow and subsequent data products.