Content uploaded by Leif L. Richardson
Author content
All content in this area was uploaded by Leif L. Richardson on Jan 14, 2014
Content may be subject to copyright.
A preview of the PDF is not available
AIMS-thermal--a thermal and high resolution color camera system integrated with GIS for aerial moose and deer census in northeastern Vermont
Abstract
A GIS analysis of landscape scale distribution of moose (Alces alces) in northern Vermont during winter 2010 showed that most moose were located at elevations of 300 – 600 m, with little discernible elevational gradient. Slope and aspect were not correlated with locations as moose were distributed in the study area with the relative amount in each descriptive class. The distribution of >85% moose based on NOAA cover types was in deciduous, mixedwood, and coniferous stands relative to their availability; locations in scrub/shrub and wetlands were higher and lower than expected, respectively. Higher resolution AIMS imagery indicated that moose used mixed woods more and coniferous stands less than available. The most significant landscape characteristic influencing the location of moose was proximity to forest openings/timber cuts that presumably provide important seasonal browse.
... S2n heterogeneous environments can impede or even lead to false identification. Using multiple spectral bands, most prominently TIR, adds further information and improves detection (see Chrétien et al. 2016;Kissell and Nimmo 2011;Millette et al. 2011). TIR light is electromagnetic radiation known as heat. ...
... Kissell and Nimmo 2011). On the other hand, Millette et al. (2011) did not find such a parallax effect in their study. Especially with the use of small strip widths of about 80-150 meters, we do not see a correlation between detection probability and distance to the centre line (Deuker and Franke 2015). ...
... The method assumes perfect detection without any masking, as well as masking independent of animal distribution. While the former can be met by flying under perfect conditions (Bernatas and Nelson 2004;Millette et al. 2011), the latter has to be carefully ensured considering species and study area. As a measure for detection probability we want to use the masking factor, here vegetation cover and environmental conditions affecting sight ability derived from available TIR and VIS imagery, as well as further variables (Beaver 2011;Deuker and Franke 2015;Franke et al. 2012). ...
Building up a model using different covariates to improve wildlife density estimations from aerial surveys.
... Results of numerous studies indicate the great potential of thermal infrared aerial surveys for wildlife monitoring (Burn et 4 al. 2009;Conn et al. 2014;Havens and Sharp 1998;Millette et al. 2011;Wiggers and Beckerman 1993). The use of manned aircraft, justified for monitoring diurnal species over vast areas (e.g. ...
... Moreover, ungulate detection with high resolution RGB cameras is impractical due to the animals' camouflaged coloration. More recently, some attempts have been made to overcome these limitations using multispectral systems of sensors -high resolution RGB digital camera coupled with a TIR camera -for aerial ungulate surveys from both manned and unmanned aircrafts (Chrétien, Théau, and Ménard 2016;Franke et al. 2012;Millette et al. 2011). The TIR sensor was used to detect animals, while species identification was based mainly on characteristics derived from RGB pictures. ...
Effective wildlife management and conservation require reliable assessments of animal abundance. However, no ungulate monitoring methods is entirely satisfying in terms of cost-effectiveness and accuracy. A new method combining unmanned aerial vehicles (drones) and thermal infrared (TIR) imaging may have great potential as a tool for ungulate surveys. Drones enable safe operations at low flying altitudes, and at night – a time that often offers the optimal conditions for wildlife monitoring. To assess the feasibility of the proposed method we used fixed-wing drones with TIR cameras to conduct test surveys in Drawieński National Park, Poland. We demonstrated that ungulate thermal signatures are visible both in leafless deciduous and in pine-dominated coniferous forests. Survey timing highly influenced the results – the best quality thermal images were obtained at sunrise, late evening, and at night. Our preliminary results indicated that thermal surveys from drones are a promising method for ungulate enumeration. We demonstrated that with ground resolution of ~10 cm it is possible to visibly distinguish large species (i.e. red deer) and achieve a good level of area coverage. The main challenges of the method are difficulties in species identification due to relatively low resolution of TIR cameras, regulations limiting drone operations to visual line of sight, and high dependence on weather.
... Consequently, new models must be created for their specified study area, which requires invasive and potentially dangerous moose capture for radio-collaring or marking during initial model development (Anderson and Lindzey 1996). Other aerial survey methods, such as mark-resight (Bowden and Kufeld 1995), mark-recapture distance sampling (Oyster et al. 2018), and infra-red thermal imaging (Millette et al. 2011), share similar limitations and challenges. In addition, capture, handling, and marking of animals is sometimes discouraged in areas such as national parks because of concern for the animals' welfare and intrusion on visitor experiences (Berger et al. 1999, Mech andBarber 2002). ...
Acquiring demographic data for moose (Alces alces) can be difficult because they are solitary in nature, they prefer densely vegetated and mountainous habitats, and they often occur at low density. Such data, however, are essential for long‐term population monitoring, evaluating management practices, and effective conservation. Winter aerial surveys are the standard method for estimating moose population parameters, but they can be logistically challenging, expensive, and subject to sightability correction, which necessitates the capture of study animals for initial model development. Herein, we demonstrate a noninvasive alternative approach for estimating population parameters of moose in northern Yellowstone National Park, where aerial surveys were attempted but proved ineffective. We determined individual moose genotype and sex using microsatellite polymerase chain reaction amplification of DNA extracted from fecal pellets, integrated ancillary pellet sample data (i.e., metadata) in genotype analysis to aid in the identification of matching genotypes, and used spatially explicit capture‐recapture (SECR) modeling to estimate sex‐specific density and abundance. We collected 616 samples over 3 consecutive winters (Dec 2013–Apr 2016) and within 2 sampling occasions each winter. We recorded 514 captures of 142 individual moose (69 males, 73 females). Overall density ranged between 0.062 moose/km2 and 0.076 moose/km2 and averaged 0.034/km2 for females and 0.033/km2 for males. Abundance estimates were 150 moose in 2013 (female = 76, 95% CI = 55–105; male = 74, 95% CI = 54–103), 186 in 2014 (female = 95, 95% CI = 63–142; male = 91, 95% CI = 60–138), and 160 in 2015 (female = 79, 95% CI = 58–108; male = 81, 95% CI = 59–110). Average population sex ratio was 0.99 males/female. We demonstrate that SECR analysis of fecal DNA genotypes, using metadata in genotype analysis to help identify matching moose genotypes, is a promising alternative method for estimating sex‐specific density and abundance of a low‐density moose population in a mountainous and forested landscape. Our findings demonstrate that SECR analysis of genotypes derived from noninvasively collected fecal pellets, coupled with the inclusion of metadata in genotype analysis, is a viable alternative method for estimating winter population parameters of moose, especially in areas where dense forests and mountainous terrain make aerial survey methods ineffective. Our methods should be applicable to population studies of other ungulate species that are difficult to monitor.
... Solar radiation heats the ground and non-target objects (e.g., rocks, stumps, trees), which in turn create noise, or thermal bright spots in an image, leading to potential false-positive detections or masking of target animals (Dunn et al. 2002, Chrétien et al. 2016, Lethbridge et al. 2019). Similar to Millette et al. (2011), who conducted thermal surveys of moose from a manned aircraft, our results for cloud cover and temperature on moose detection demonstrated this phenomenon. Increased cloud cover improved thermal detection probability because clouds blocked some amount of solar energy from reaching objects on the ground, thus maximizing the thermal contrast between moose and their surroundings and reducing the potential for misidentifying a non-moose object as a moose. ...
The use of unmanned aerial systems (UAS) for wildlife surveying and research has widely expanded in the past decade, but with varying levels of success. Applying UAS paired with Forward Looking Infrared (FLIR) technology to survey forest-dwelling species has been particularly challenging because of unreliable animal detection. We describe our application of UAS and FLIR technology to detect GPS-collared moose (Alces alces) and their calves in the heavily-forested region of northeastern Minnesota, USA, during 2018 and 2019. We conducted grid-pattern UAS thermal surveys over GPS-collared cows during the calving seasons (April to June) of 2018 and 2019 to determine the feasibility of using a FLIR-equipped UAS for detecting cow moose, and for quantifying the number of calves. We also collected data on environmental and flight characteristic variables to model moose detection. Our best fitting model of moose detection showed increased detection with more cloud cover at the survey site ( β = 1.13, SE = 0.43), whereas increased forest canopy ( β = −1.10, SE = 0.38), and vegetative greenness (enhanced vegetation index, EVI; β = −1.37, SE = 0.32) both reduced detection success. By adjusting our methodology based on our detection model findings, we increased our adult moose detection success from 25% during our first season, to 85% during our second season, and calf detection from 27% to 79%, respectively. We report on our methodological improvements and identify limitations to UAS-based wildlife research in forested systems. Overall, we found that UAS with FLIR sensing is a promising tool for quantifying moose calving success, twinning rate, and calf survival, and may be effective for monitoring the reproductive success and survival of other wildlife species in densely forested regions.
... However, attempts thus far to mitigate, or develop alternatives to, population estimation under poor sightability conditions have been isolated, unfruitful, or narrowly applied (White 2007, Christ 2011, Seaton 2014, Wald and Nielson 2014, Frye 2016. Further, few attempts have been made in Alaska to investigate performance of methods that detect moose by means other than human observers (e.g., radiometric thermal imaging systems, Millette et al. 2011). Downscaled climate projections under even midrange emission scenarios demonstrate that this problem is expected to worsen, 1 magnifying the need for moose monitoring methods that do not rely on a constant, high detection rate for accuracy. ...
Monitoring moose (Alces alces) populations is a key component of wildlife management in Alaska. In response to reports of recent difficulties implementing the existing techniques for monitoring moose, an interagency work group identified the monitoring techniques currently in use, characterized technique performance, and examined commonality and geographic patterns of problems encountered when applying techniques in the field. Field biologists engaged in monitoring moose in Alaska were emailed an online questionnaire designed to organize information about overall program satisfaction, population parameters monitored, techniques for estimating parameters, and current impediments to monitoring.
During 2007–2017, biologists failed to complete 42% of scheduled surveys to estimate abundance (n = 295 surveys, 42 respondents). Survey failure rates differed across ecoregions: failure rates were highest in the Kenai/Southcentral (57%), Eastern Interior (43%) and Coastal Subarctic (41%) ecoregions, but lower rates of survey failure were reported for Western Interior (20%) and Arctic Slope (15%) ecoregions of Alaska. Patterns of survey failure were similar for composition.
Lack of adequate snow cover and poor flying weather were the first and second most commonly cited reasons, respectively, for failure to complete scheduled surveys. Where surveys were successfully completed, estimates generally had less precision than desired, with only 50% of respondents achieving intended precision goals for abundance estimation. Biologists indicated a strong willingness to use a new method for monitoring moose if it 1) did not rely on complete snow cover, 2) was more accurate, 3) provided higher precision, 4) provided continuity with previous estimates, 5) could be used where inclement flying weather is frequent, 6) could be used in areas with dense vegetative cover, 7) was accompanied by technical assistance or a user manual, and 6) was similar in cost to existing methods. They indicated mild unwillingness to use a new method that 1) used ground observations, 2) required hunters to turn in specimens, 3) used helicopters for aerial observation, or 4) required more resources than current methods.
These results highlight the need to develop new survey and measurement techniques that can be conducted independently of problematic snow and weather conditions, or at least have far more flexibility in implementing survey protocols. Indeed, the problem of monitoring moose in areas with poor snow conditions is so challenging and pervasive that solutions may require a concentrated, cooperative effort among agencies, including practical feedback from field biologists. Precision of existing techniques may also be improved through better optimization of survey design, the integration of more historical population information in estimation, and perhaps by better clarifying precision requirements relative to program goals.
... While heavily forested, timber harvesting is common throughout as the majority of the land is privately owned and commercially harvested (NEFA 2007). The 2011 moose density was estimated at 0.77 moose/km 2 (1.96 moose/ mi 2 ) based on a rolling 3-year average of moose sightings by early winter (November) deer hunters, and was previously estimated in 2010 as 0.93 moose/km 2 (2.41 moose/ mile 2 ) based on aerial surveys (Millette et al. 2011). ...
Drones are growing in popularity and are used to locate individual animals, estimate ungulate population size, and monitor endangered species. Our goal was to assess the efficacy of using drones to locate neonatal white‐tailed deer (Odocoileus virginianus) for capture. During May–June 2019 and June 2020, we used a drone equipped with thermal‐infrared (TIR) and conventional multispectral red‐green‐blue (RGB) cameras to locate and confirm fawn thermal signatures in Minnesota's southern farmland region. We confirmed 29 and 71 fawn thermal signatures in 2019 and 2020, respectively. We captured 13 fawns to validate the methodology (detection and identification) of using drones for capture in 2019. We flew the drone for 47.3 hours covering 792 hectares (17 hectares per hour) in 2019 and flew for 41.1 hours and covered 1640 hectares (40 hectares per hour) in 2020. We worked 201.5 person‐hours with a mean crew size of 4 people that required 6.9 person‐hours to locate each fawn in 2019, whereas we worked 145.7 person‐hours with a mean crew size of 4 people that required 70% less effort and resulted in 2.1 person‐hours to locate each fawn in 2020. Flight efficiency was highest at 60 m above ground level (AGL) with a speed of 9 m/s; however, diurnal use of drones made identifying fawn thermal signatures difficult as thermal loading of the environment quickly reduced the temperature differential between vegetation and signatures. We found TIR‐equipped drones to be an efficient method to locate and subsequently capture fawns in open habitats in comparison to other capture methods such as vaginal implant transmitters (VITs), ground searches, or monitoring doe behavior. We recommend flying at night or in cloudy conditions to avoid false positives. © 2021 The Wildlife Society. Our goal was to assess the efficacy of using drones to locate and neonatal white‐tailed deer (Odocoileus virginianus) for capture. We found TIR‐equipped drones to be an efficient method to locate and subsequently capture fawns in open habitats in comparison to other capture methods, but recommend flying overnight or in cloudy conditions to avoid false positives.
Drones equipped with thermal sensors have shown ability to overcome some of the limitations often associated with traditional human‐occupied aerial surveys (e.g., low detection, high operational cost, human safety risk). However, their accuracy and reliability as a valid population technique have not been adequately tested. We tested the effectiveness of using a miniaturized thermal sensor equipped to a drone (thermal drone) for surveying white‐tailed deer (Odocoileus virginianus) populations using a captive deer population with a highly constrained (hereafter, known) abundance (151–163 deer, midpoint 157 [87–94 deer/km2, midpoint 90 deer/km2]) at Auburn University's deer research facility, Alabama, USA, 16–17 March 2017. We flew 3 flights beginning 30 minutes prior to sunrise and sunset (1 morning and 2 evening) consisting of 15 nonoverlapping parallel transects (18.8 km) using a small fixed‐wing aircraft equipped with a nonradiometric thermal infrared imager. Deer were identified by 2 separate observers by their contrast against background thermal radiation and body shape. Our average thermal drone density estimate (69.8 deer/km2, 95% CI = 52.2–87.6), was 78% of the mean known value of 90.2 deer/km2, exceeding most sighting probabilities observed with thermal surveys conducted using human‐occupied aircraft. Thermal contrast between animals and background was improved during evening flights and our drone‐based density estimate (82.7 deer/km2) was 92% of the mean known value. This indicates that time of flight, in conjunction with local vegetation types, determines thermal contrast and influences ability to distinguish deer. The method provides the ability to perform accurate and reliable population surveys in a safe and cost‐effective manner compared with traditional aerial surveys and is only expected to continue to improve as sensor technology and machine learning analytics continue to advance. Furthermore, the precise replicability of autonomous flights at future dates results in methodology with superior spatial precision that increases statistical power to detect population trends across surveys. © 2020 The Wildlife Society. Drones equipped with thermal sensors have shown ability to overcome some of the limitations often associated with traditional human‐occupied aerial surveys (e.g., low detection, high operational cost, human safety risk); however, their accuracy and reliability as a valid population technique have not been adequately tested. We tested the effectiveness of using a drone equipped with a nonradiometric thermal infrared imager for surveying a captive deer population with a known abundance and observed the ability to conduct precisely replicated surveys with sighting probabilities (92% during optimized flight conditions) that equal or exceed traditional airborne techniques.
Estimating the abundance of wide‐ranging wildlife, difficult under any circumstances, is particularly challenging when detection is low and affected by factors that also influence density and distribution. In northeastern Washington, moose (Alces alces) have evidently increased since the 1970s but spend most of their time under coniferous cover that makes detection from the air difficult. We used a Bayesian hierarchical approach to incorporate habitat use (in the form of availability as a function of canopy closure) into a detection model within a mark‐recapture distance sampling framework to estimate moose density. Our model of availability used a latent density surface employing habitat use data obtained from 17 adult female moose wearing global positioning system (GPS) collars. Distance sampling data, obtained from helicopter surveys in winters 2014, 2015, and 2016, consisted of double‐observer detections of 166 moose groups along 2,241 km of systematically placed line transects within 29 survey blocks selected using a stratified‐random design. We estimated moose density over the entire survey area as 0.49/km² (95% credible interval = 0.33–0.67/km²). Extrapolated to the 10,513‐km² survey area, we estimated 5,169 moose (95% credible interval = 3,510–7,034). Our methodology allowed us to adjust for availability bias and produce an estimate even where detection was difficult but required many hours of helicopter flights, acceptable weather conditions, and the availability of GPS collared‐moose. © 2018 The Wildlife Society
Unregulated hunting and habitat loss led to a near extirpation of moose (Alces alces) in New Hampshire in the 1800s. After state protection in 1901, the estimated population increased slowly to ~500 moose in 1977, then increased rapidly in the next 2 decades to ~7500 following an increase in browse habitat created by spruce budworm (Choristoneura fumiferana) and related timber salvage operations, and then halved from 1998-2016 despite highly available optimal habitat. The declining population was partially related to the specific management objective to reduce moose-vehicle collisions, and a possible change in deer hunter and moose behavior that influence population estimates. But given the substantial decline in productivity and condition of cows, and frequent episodes of high calf mortality in April, the primary cause of decline was presumed to be is an increase in winter tick abundance.
This study examined the relationships among moose density, optimal habitat, weather/ground conditions, winter tick abundance, and natal dispersal in northern New England. Comparing movement data from the previous (2002-2006) and current (2014-2016) productivity studies in New Hampshire and Maine, the distance of natal dispersal, home and core range size, and home and core range overlap did not significantly (P > 0.05) change despite an increase in optimal habitat and a decrease in moose density.
Geographic changes in tick abundance were related to an interaction between moose density, and the onset and length of winter. Annual changes in tick abundance in northern New Hampshire are driven by desiccating late summer conditions, as well as the length of the fall questing season. Lower precipitation (6.4 cm) and higher minimum temperatures (9.8 °C) specifically concentrated during larval quiescence from mid-August through mid-September reduces winter tick abundance and the likelihood of an epizootic event. The onset of winter, defined by the first snowfall event (> 2.54 cm), influenced the length of the questing season relative to the date of long-term first snowfall event (14 November). In the epizootic region, average winter tick abundance on moose harvested in mid-October indicated a threshold of 36.9 ticks, above which an epizootic is like to occur unless an early snowfall event shortened the fall questing season. Optimal habitat created by forest harvesting was produced at an annual rate of 1.3% (1999-2011) and is not considered limiting in northern New Hampshire, but likely concentrates moose density locally (~4 moose/km2) facilitating the exchange of winter ticks. In northern New Hampshire, snow cover late into April did not reduce tick abundance in the following year and cold temperatures (< 17 °C) that induced replete adult female mortality are extremely rare in April.
Given a continuation of warming climate and conservative moose harvest weather conditions and high local moose densities will continue to favor the life cycle of winter ticks, increasing the frequency of winter tick epizootics and shift the epizootic region slowly northward. Conversely, temporary reduction of moose density may substantially reduce parasite abundance and support a healthier and more productive moose population.
Multi-Resolution Land Characterization 2001 (MRLC 2001) is a second-generation Federal consortium designed to create an updated pool of nation-wide Landsat 5 and 7 imagery and derive a second-generation National Land Cover Database(NLCD 2001). The objectives of this multi-layer, multi-source database are two fold: first, to provide consistent land cover for all 50 States, and second, to provide a data framework which allows flexibility in developing and applying each independent data component to a wide variety of other applications. Components in the database include the following: (1) normalized imagery for three time periods per path/row, (2) ancillary data, including a 30 m Digital Elevation Model(DEM) derived into slope, aspect and slope position, (3) per-pixel estimates of percent imperviousness and percent tree canopy, (4) 29 classes of land cover data derived from the imagery, ancillary data, and derivatives, (5) classification rules,
confidence estimates, and metadata from the land cover classification. This database is now being developed using a Mapping Zone approach, with 66 Zones in the continental United States and 23 Zones in Alaska. Results from three initial mapping Zones show single-pixel land cover accuracies ranging from 73 to 77 percent, imperviousness accuracies ranging from 83 to 91 percent, tree canopy accuracies ranging from 78 to 93 percent, and an estimated 50 percent increase in mapping efficiency over previous methods. The database has now entered the production phase and is being created using extensive partnering in the Federal government with planned completion by 2006.
A commercially available thermal-infrared scanning system was used to survey populations of several wildlife species. The
system's ability to detect species of different sizes in varying habitats relative to conventional survey methods, to differentiate
between species in the same habitat, and the influence of environtmental factors on operational aspects of employing this
technology in the field were evaluated. Total costs for the surveys were approximately $0.36/ha. There were marked discrepancies
in the counts of untrained observers and those from trained analysis. Computer-assisted analysis of infrared imagery recorded
52% fewer deer than were estimated from drive counts, and densities of moose were five times those estimated from conventional
aerial methods. By flying concentric circles and using telephoto, detailed counts of turkeys and deer were possible. With
the aid of computer-assisted analysis, infrared thermography may become a useful wildlife population survey tool. More research
is needed to verify the actual efficiency of detection by combining aerial scans with ground truthing for a variely of species
and habitals.
This paper concerns the potential uses of infrared scanning for remote sensing of big game animals. The basic mechanics of the equipment are described and the method of aerial scanning explained. An operational trial of the method was performed on the George Reserve white-tailed deer (Odocoileus virginianus) herd. In addition to the regular deer population, three deer were placed in small pens under different cover conditions-open grassland, hardwoods (with the leaves down), and pine coniferous canopy. Radiometer readings on the penned deer and their backgrounds showed approximately 7 C differential in apparent temperature. The two deer in the grassland and oak woodland were easily detected, but the deer under the conifer canopy would have been missed if its location had not been known. The infrared count of the Reserve herd was 98 as compared to a population estimated by other methods at 101 head. Under the right conditions, infrared scanning will probably give better counts over large areas than any other technique available at present. However, the inability of infrared to penetrate green leaf canopy, variability of animal and background apparent temperatures depending upon weather and other factors, difficulty in distinguishing between species of animals, and high initial cost of the scanning device are substantial limitations to the use of the technique.
An airborne, thermal infrared scanner was tested for deer detection over penned mule deer (Odocoileus hemionus hemionus) near Fort Collins, Colorado. The animals were detected at 300- and 500-ft altitudes, but not at 1,000 ft.
The applicability of two types of airborne infrared detector systems for censusing white-tailed deer (Odocoileus virginianus) was studied during different seasons and at different altitudes. This method of detecting deer was checked against ground censusing along an interstate highway that traversed the heavily wooded, mountainous study area. We found that detectability from the air was related to time of day, season, altitude, and wavelength sensitivity of the infrared detectors. In preliminary studies, individual cattle were easily detected from an altitude of 1,000 feet, using infrared equipment sensitive in the 3-14 micrometer region of the light spectrum. Deer in shrublands were detected at 100, 250, and at 500 feet during nighttime summer flights, using a 3-14 micron detector; the 3-5 micron detector was found to be a better hot spot detector and gave the best image during the summer surveys. In contrast, the 3-4 micrometer detector provided superior images during a nighttime winter flight. The interaction of seasonal and wavelength sensitivity effects were attributed to differential thermal radiation. During winter, the deer's hair surface is more nearly equal to the average background temperature than it is during summer; this would be adaptive in that body heat is apparently freely radiated during the summer and retained during the winter. Although of great potential value as a censusing technique for big game animals, current limitations of the equipment used makes large-scale airborne infrared detection of deer impractical unless the census area is flat and relatively free of obstructing vegetation. Extensive work with the equipment described here disclosed that nighttime surveys are more likely to be successful than daytime flights unless the sky is heavily overcast. Difficulties with interference from large amounts of reflected solar radiation in the shorter wavelengths of the infrared spectrum are likely to be encountered in daytime operations, especially on clear days. Refinement of the technique will undoubtedly provide a powerful tool for studying population dynamics and behavior of big game species.
Estimating abundance of free-ranging white-tailed deer (Odocoileus virginianus) with traditional methods often produces variable results, is labor-intensive, requires specific weather conditions, and is expensive. Thermal imaging from an airborne platform promises advantages over conventional techniques. We estimated variation in the detection rate of white-tailed deer in a deciduous forest landscape in central Missouri using aerial surveys (n = 10) with a state-of-the-art thermal imaging system. Replicated aerial mark-resight surveys (n = 11) provided an independent abundance estimate for evaluating thermal imaging estimates. Based on an aggregated mark-resight estimate of 311 deer, detection rates for thermal imaging surveys ranged from 31-89% (x̄=56%, SE=6.7). Variability in detection was attributed to inconsistent sensor operation and variable thermal contrast between deer and background objects. Operator bias likely will remain problematic until hardware and software developments can automate documentation of ground coverage. Further study is needed to identify factors impacting thermal contrast and detection rate. Until the capabilities of thermal imaging are more fully understood and the sampling protocols refined, detection rates may be too variable to provide reliable counts of animal abundance.
Evaluation of infrared technology for aerial moose surveys in New Hampshire
- K P Adams
- P J Pekins
- K A Gustafson
- K Bontaites
ADAMS, K. P., P. J. PEKINS, K. A. GUSTAFSON,
and K. BONTAITES. 1997. Evaluation
of infrared technology for aerial moose
surveys in New Hampshire. Alces 33:
129-139.