Predictions of understorey light conditions in northern hardwood forests following parameterization, sensitivity analysis, and tests of the SORTIE light model

We parameterized the light model of SORTIE for northern hardwoods in eastern Canada, and performed a sensitivity analysis and validation tests of the model before using it to predict the effect of various types of partial cutting on understorey light conditions. The parameterization was done by characterizing the crown geometry and openness of sugar maple (Acer saccharum Marsh.), yellow birch (Betula alleghaniensis Britt.), and beech (Fagus grandifolia Ehrh.). Those results indicated that beech casts a deeper shade than sugar maple and yellow birch. The sensitivity analysis showed that the model predictions were more sensitive to variations in the crown geometry parameters, especially the crown radius parameter, than to variations in crown openness. Validation tests of the model were performed in both mapped and unmapped plots by comparing light predicted by SORTIE to light measured in the field using hemispherical photographs and sensor-based measurements. In mapped stands, the model provided reasonably accurate predictions of the overall variation in understorey light levels between 2 and 30% full sunlight, but the predictions tended to lack spatial precision. In unmapped stands, SORTIE accurately predicted stand-level mean light availability at 5 m aboveground for stands ranging in basal area from 19 to 27 m(2)/ha. At heights lower than 5 m, SORTIE accurately predicted the light availability in a recent selection cut with a low density of understorey vegetation, but tended to overestimate light availability in stands with relatively dense undergrowth. Finally, a demonstration of the possible usefulness of the SORTIE light model is presented by using the model to compare the proportion of various light microsites created by a variety of selection cutting systems in use in eastern Canada (selection cutting with different harvesting intensities, group selection, and patch selection). (C) 2002 Elsevier Science B.V. All rights reserved.