Augmenting size models for Pinus strobiformis seedlings using dimensional estimates from unmanned aircraft systems

In forestry, common garden experiments traditionally require manual measurements and visual inspections. Unmanned aircraft systems (UAS) are a newer method of monitoring plants that is potentially more efficient than traditional techniques. This study had two objectives: to assess the size and mortality of Pinus strobiformis Engelm. seedlings using UAS and to predict the second-year seedling size using manual measurements from the first year and from UAS size estimates. Raised boxes containing 150 seedlings were surveyed twice, one year apart, using multispectral UAS. Seedling heights and diameters at root collar (DRC) were measured manually in both years. We found that size estimates made using a vegetation mask were suitable predictors for size, while spectral indices were not. Furthermore, we provided evidence that inclusion of UAS size estimates as predictors improves the fit of the models. Our study suggests that common variables used in forest monitoring are not necessarily best suited for seedlings. Therefore, we created a new variable, called the longitudinal area (height x DRC), which proved to be a significant predictor for both height and DRC. Finally, we demonstrate that seedling mortality can be effectively measured from remotely sensed data, which is useful for common garden and regeneration studies.