Detecting decadal forest canopy change during a native insect outbreak in the Ozark National Forest, Arkansas

A model developed to detect changes in forest canopy cover was implemented in response to an outbreak of a native oak boring insect, red oak borer ( Enaphalodes rufulus (Haldeman) (Coleoptera: Cerambycidae)) in the Ozark National Forest of Arkansas. A Normalized Difference Water Index (NDWI) was used to detect forest cover change between two dates of Landsat satellite imagery. The resulting DeltaNDWI was obtained by subtracting one image from the other. Thresholds for extreme levels of positive and negative changes over time were derived from DeltaNDWI, and the resulting areas were described as either growth or decline. Forest variables were measured in situ within both growth and decline areas. Growth plot trees were smaller and younger, while decline plot trees were larger, older, displayed varying levels of crown dieback, and were often dead. Logistic regression was used to determine which forest measurement variable was the best predictor for the growth and decline categories. Age and DBH were the most successful predictors, DBH being preferable because it takes less time and effort to measure in situ. Growth plots were comprised mostly of sprouts from cut stumps, while decline plots had a relatively high frequency of northern red oaks in varying states of decline, with dead trees possessing numerous red oak borer heartwood galleries. Decadal detection of forest canopy change using NDWI can reveal areas of forest decline, and areas of regenerating even-aged stands that could be considered potential hazards for future disturbances.;Keywords: Landsat, NDWI, red oak borer, change detection, logistic regression...