Modelling deciduous forest ice storm damage using aerial CIR imagery and hemispheric photography

An evaluation of low cost aerial colour infrared (CIR) imagery and hemispheric photography for assessment of forest canopy structural damage caused by a severe ice storm is presented. Much of eastern Canada and the northeastern United States was affected by the five storms in January, 1998. Forest damage in Gatineau Park, Quebec was measured in August 1998 using several structural variables that were either directly representative of damage, or are expected to change over time in response to damage levels. Direct measures of individual tree damage included numbers of stems, first order branches and second order branches that were broken, and a visual health index. Canopy level structural measures expected to vary as the forest responds to damage included effective leaf area index (LAIe) and canopy closure measured using hemispheric photographs, and basal area. Aerial CIR photographs were acquired with a Vinten 70 mm camera and Kodak 2443 CIR film in mid-August 1998. They were scanned to 0.25m ground pixel spacing and corrected for bi-directional and optical light fall-off effects. Hemispheric photography proved to be a simple, repeatable, and low cost technique for derivation of LAIe and canopy closure, although the analysis algorithm slightly underestimates LAIe Stepwise multiple regression of image spectral, textural and shadow fraction measures produced significant relations (p ≤ 0.05) for LAIe canopy closure, basal area and the health score. In these models image texture and shadow fractions were generally not individually significant, but improved the overall significance and R2 by partitioning out some of the irrelevant variance in the other predictor variables. The multiple variable models were therefore more precise than single spectral variable models, illustrating the additional information content of image texture and shadow fractions. Weaker relations were found for models of the numbers of fallen branches and stems.