The processing and evaluation of digital airborne imagery for detection, monitoring and modeling of mountain pine
beetle (MPB) infestations is evaluated.
The most efficient and reliable remote sensing strategy for identification and mapping of infestation stages ("current" to
"red" to "grey" attack) of MPB in lodgepole pine forests is determined for the most practical and cost effective
This research was planned to specifically enhance knowledge by determining the remote sensing imaging systems and
analytical procedures that optimize resource management for this critical forest health problem. Within the context of
this study, airborne remote sensing of forest environments for forest health determinations (MPB) is most suitably
undertaken using multispectral digitally converted imagery (aerial photography) at scales of 1:8000 for early detection of
current MPB attack and 1:16000 for mapping and sequential monitoring of red and grey attack. Digital conversion
should be undertaken at 10 to 16 microns for B&W multispectral imagery and 16 to 24 microns for colour and colour
From an "operational" perspective, the use of twin mapping-cameras with colour and B&W or colour infrared film will
provide the best approximation of multispectral digital imagery with near comparable performance in a competitive
private sector context (open bidding).
A variety of aerial mapping cameras were adapted and developed into simulated multiband digital photogrammetric
mapping systems. Direct digital multispectral, two multiband cameras (IIS 4 band and Itek 9 band) and paired mapping
and reconnaissance cameras were evaluated for digital spectral performance and photogrammetric mapping accuracy in
an aquatic environment.
Aerial films (24cm X 24cm format) tested were: Agfa color negative and extended red (visible and near infrared)
panchromatic, and; Kodak color infrared and B&W (visible and near infrared) infrared. All films were negative
processed to published standards and digitally converted at either 16 (color) or 10 (B&W) microns. Excellent precision
in the digital conversions was obtained with scanning errors of less than one micron. Radiometric data conversion was
undertaken using linear density conversion and centered 8 bit histogram exposure. This resulted in multiple 8 bit spectral
image bands that were unaltered (not radiometrically enhanced) "optical count" conversions of film density. This
provided the best film density conversion to a digital product while retaining the original film density characteristics.
Data covering water depth, water quality, surface roughness, and bottom substrate were acquired using different
measurement techniques as well as different techniques to locate sampling points on the imagery. Despite extensive
efforts to obtain accurate ground truth data location errors, measurement errors, and variations in the correlation between
water depth and remotely sensed signal persisted. These errors must be considered endemic and may not be removed
through even the most elaborate sampling set up.
Results indicate that multispectral photogrammetric systems offer improved feature mapping capability.
Conference Committee Involvement (1)
Remote Sensing of the Ocean, Sea Ice, and Large Water Regions 2008
15 September 2008 | Cardiff, Wales, United Kingdom