| Increasing amounts and types of timely and accurate data are required for monitoring to ensure compliance with natural resource regulatory requirements. This study developed a cost-effective method to partially fulfill these data requirements using super large scale aerial photography (Scale: greater than 1:2,000). Two synchronized, metric, Rolleiflex 70mm (2.76in) cameras mounted 12m (40ft) apart on a rigid platform and carried at 5.6 km/hr (3 knots) by a helicopter collected this high resolution, 3D imagery from Alaska and Washington. The overlapping photo pairs provided 3D views of natural resource objects as fine as twigs. The 12m (40ft) inter-camera distance improved ground visibility between tree crowns of dense old growth forests. Analytical stereoplotters and the application of photogrammetric principles enabled measurement and interpretation of photo objects such as trees and their height in a cost-effective way. Horizontal and vertical measurement accuracy was within 2% and 3% of field measurement, respectively.; Forest inventory and riparian buffer monitoring applications were used to test this method. Although field work is still required to develop photo-field relationships unique to each ecosystem and for quality assurance, the photo estimates of individual tree height, volume, diameter, type, and location, as well as down tree decay class and landing spot, plot timber volume, and area were comparable to and may replace approximately 95% of field effort. For example, the average of the absolute differences between field and photo estimates for tree height was 2.4m (7.8ft) (s.d. = 2.1m (6.8ft), n = 376), diameter at breast height (1.4m (4.5ft) above ground on uphill tree side) was 5.8cm (2.3in) (s.d. = 5.6cm (2.2in), n = 109), and plot volume in gross board feet was within 10.9% to 13.4% (n = 10) depending on the estimator used. Forest type was correctly classified 99.4% (n = 180) of the time.; Timber inventory, species identification, sample distribution, down wood detection, and mapping of habitat features such as streams and trees provided improvement over field methods alone. For example, tree species was correctly identified 90% (n = 176) of the time for alder, Alaska cedar, lodgepole pine, mountain hemlock, Sitka spruce, western hemlock, and western redcedar. The spatial position of tree bases using digital elevation models was within 2.6m (8.4ft) of the actual position (s.d. = 1.4m (4.5ft), n = 20).; Tree height, riparian buffer, and forest inventory plot volume data photo collection costs were one-fourth to one-half those of field methods. The new variables of tree crown closure and tree branch density were developed but did not contribute significantly to tree volume estimation. Physiographic limitations and sampling bias were eliminated by helicopter use. |
