Invasive pests cause major ecological and economic damages to forests around the world. Satellite imagery is an important tool for monitoring defoliation at broad scales, but environmental conditions can affect whether defoliation leads to mortality. In this study, we modeled forest mortality resulting from a 2015-2017 Spongy moth outbreak in the temperate deciduous forests of Rhode Island (northeastern U.S.). We used Landsat-based defoliation mapping and geospatial environmental data with Random Forest to model mortality severity of canopy trees at a 100 m spatial resolution.
Defoliation was mapped based on declines in Normalized Difference Vegetation Index (NDVI) in outbreak years compared to baseline NDVI from pre-outbreak years. Other predictors included geospatial data representing soil characteristics, drought condition, and forest characteristics as well as proximity to coast, development, and water. The Random Forest tool in Python Sklearn was used to model forest mortality with 2 classes (low/high) and 3 classes (low/med/high). The best models had overall accuracies of 82% and 65% for the 2-class and 3-class models, respectively. The most important predictors of forest mortality were defoliation, distance to coast, and canopy cover. Soils and topography had minimal importance in the models possibly due to limitations of the data and limited variability within our study area. Repeated defoliations were relatively rare during the outbreak. Model performance improved only slightly with the inclusion of more than 3 variables. The models classified 35% of forests as having canopy mortality > 5 trees/ha and 21% of Rhode Island forests having mortality >11 trees/ha. The study shows the benefit of Random Forest models that use both defoliation maps and geospatial environmental data for classifying forest mortality.