Interpretation of digital imagery to estimate juvenile stand attributes in managed boreal stands, density, stocking and height

Abstract

Forest regeneration monitoring is critical to inform forest management planning, evaluate silvicultural efficacy, and determine achievement of renewal standards in managed forests. We assessed the accuracy of operational monitoring using interpretation (INT) of true colour 7–10 cm digital stereo imagery in juvenile stands across a wide range of species compositions typical of northwestern Ontario’s boreal forest. Using the same grid of 16 m2 circular plots established at a density of 2 ha-1, interpreted stand-level estimates were compared to field survey estimates from summarized plot data. Using 1508 field plots, estimates of density, stocking and height were derived for species and species groups (e.g., poplars) across 46 stands. Species compositions were developed using two approaches (all stems and stocking) and accuracy of INT estimates of density, stocking, and height were analysed using an observed (field data) vs. predicted (INT data) linear modelling approach. The INT approach appears useful for monitoring regeneration and providing stand-level estimates of density and stocking, particularly for conifers as a group and for jack pine. However, INT underestimated deciduous tree density and stocking and failed to distinguish spruce from balsam fir or count white birch saplings. These errors have implications for determination of species composition from INT of leaf-off imagery. An approach to quality control is described, and recommendations for ways to improve operational estimates of height and species composition using INT assessments are provided.