Map Moose Group Populations

The Coeur d’Alene Regional Office retained Vision Air Research to conduct a moose count within a portion of Wildlife Management Unit # 5.   The goal of the project was to conduct an aerial infrared survey for moose and map moose group locations, and provide a count of moose observed.   

Conifer forests including some areas that have been harvested, dominate the survey area.  The area also includes residential area that include house, barns, other buildings, recreational fields, open parks, hardwood forests/shrublands and agricultural areas.  The lower elevations were free of snow while the upper elevations were covered with snow packs deep enough to cover any shrub layers.

Methods

The survey was conducted mid March, 2011 between 1200 and 2000 hours.  Daylight surveys were used to allow for flight safety in this mountainous terrain.  In addition, daylight conditions allowed the use of the color video camera as well as the IR to allow confirmation of the various large mammals in the area.

Detection Potential

Cover type influences the availability of the subject animal to be detected by the sensor.  A dense canopy will make it more difficult to detect a moose since infrared doesn’t see through the vegetation canopy.  A moose that is completely hidden by vegetation canopy cover is “unavailable” to be detected – more on this below.  Detection rates for open areas such as agricultural fields and meadow were 100%, deciduous forests were roughly 86 %, and conifer can range from 50 – 80 % depending on the canopy closure.

The other variable which has shown a strong influence on detect was “sky” or the effect the cloud deck had on how quickly infrared energy was emitted.  A cloud layer allows the animals to glow hot compared to the radiant energy emitted by rocks, soil, and vegetation.  A cloud layer enhances detection.  It was cloudy during these surveys, which provides optimal conditions for detection while daylight hours provided the ability for verification with the color camera.

There were no “controls” or known moose to allow developing a search image of moose in this study area.  Other research I’ve conducted to determine detection rates have been based on known target subjects.  For example, one or more individuals in a group had radio collars.  The location of the target subject was monitored by a second aircrew in another airplane or via ground based crews to avoid any detection bias.  These controls allowed me to determine if the individual or groups were detected, were available to be detected and subsequently missed, or unavailable to be detected because they were no longer in the search area.  In areas where no collared animals were available, previously detected animals were used as targets in subsequent replicates.  This is similar to a mark – recapture method for determining detection.  These efforts have revealed a consistency as to which variables influence detection.  The vegetation cover type is the primary variable to confound detection rates.  Infrared cannot detect or “see” through leaf cover.  As such, evergreen species can thwart detection.  Branches and tree boles can also influence detection based on the size of the animal. Some animals may be able to effectively hide behind tree boles or masked by dense branches.  This variable is fairly easy to comprehend – if the animal is hidden it is not available to be detected.  If the animal can’t be seen by visual methods (e.g., a deer is bedded behind a tree bole) it can’t be see or was considered “unavailable”.  What was not obvious was the effect of bud break on detection.  Although the deer, for example, could be seen visually during bud break, the deer can be masked by the energy given off by the bud break.  Buds effectively “glow” masking deer behind the canopy.  Bud break was not an issue during this survey.