Open Collections

UBC Undergraduate Research

Using remote sensing to detect Dendroctonus ponderosae (Mountain Pine Beetle) impact in British Columbia Chiu, Horton 2011-04-18

You don't seem to have a PDF reader installed, try download the pdf

Item Metadata

Download

Media
52966-Chiu_Horton_FRST_497_2010.pdf [ 1.04MB ]
Metadata
JSON: 52966-1.0075504.json
JSON-LD: 52966-1.0075504-ld.json
RDF/XML (Pretty): 52966-1.0075504-rdf.xml
RDF/JSON: 52966-1.0075504-rdf.json
Turtle: 52966-1.0075504-turtle.txt
N-Triples: 52966-1.0075504-rdf-ntriples.txt
Original Record: 52966-1.0075504-source.json
Full Text
52966-1.0075504-fulltext.txt
Citation
52966-1.0075504.ris

Full Text

Using Remote Sensing to detect Dendroctonus ponderosae (Mountain Pine Beetle) Impact in British Columbia Faculty of Forestry The University Of British Columbia Vancouver, B.C By: Student: Horton Chiu Bachelor of Science in Forestry, Forest Resource Management 4/18/2011 1Primary advisor: Nicholas Coops Secondary advisor: Michael Meitner 2Abstract Dendroctonus ponderosae, mountain pine beetle infestation in BC has become a great issue over the years. Remote sensing is used to monitor mountain pine beetle impact so that damage can be mitigated, prevented, and controlled. Aerial overview survey, aerial photography and different satellite applications are compared. Future advancement in remote sensing applications can save our Pinus contorta (lodgepole pine) forests in BC. *Keywords: Aerial Overview Survey, Aerial Photography, IKONOS, Landsat, Quickbird, Spot 3Contents Abstract                                                                                                                                                                           .......................................................................................................................................................3  Contents                                                                                                                                                                          ......................................................................................................................................................4  1. Introduction                                                                                                                                                                .............................................................................................................................................6                                                                                                                                                                                  .............................................................................................................................................................6  2.1 Mountain Pine Beetle Biology                                                                                                                               ................................................................................................................6  2.1 Relationship with fungi                                                                                                                                      ......................................................................................................................8  2.2 Tree attack stages                                                                                                                                               ..............................................................................................................................8  3. Remote Sensing Background                                                                                                                                   ...................................................................................................................9  3.1 Definition                                                                                                                                                              ...........................................................................................................................................9  3.2 Spectral, Spatial, and Temporal Resolution                                                                                                    .......................................................................................9  3.3 Assessing Accuracy                                                                                                                                           ..........................................................................................................................10  4. Using Remote Sensing to detect Mountain Pine Beetles                                                                                 .......................................................................12  4.1 Spatial resolution                                                                                                                                              .............................................................................................................................12  4.2 Spectral resolution                                                                                                                                            ...........................................................................................................................13  4.3 Temporal Resolution                                                                                                                                        ........................................................................................................................15  5. Remote Sensing Instruments and Their Capacities to detect Beetle Attack                                                 ..........................................16  5.1 Introduction                                                                                                                                                       .....................................................................................................................................16  5.2 Aerial Overview Survey and Aerial Photography                                                                                         .............................................................................17  5.2.1 Introduction                                                                                                                                               ..............................................................................................................................17  5.2.2 How It Works                                                                                                                                             ............................................................................................................................18  5.2.3 Applications and Techniques                                                                                                                  ....................................................................................................18  5.3 Satellite Imagery                                                                                                                                               ..............................................................................................................................21  5.3.1 Introduction                                                                                                                                               ..............................................................................................................................21  5.3.2 How It Works                                                                                                                                             ............................................................................................................................21  5.3.3 Applications and Techniques                                                                                                                  ....................................................................................................22  5.4 Limitations and Benefits                                                                                                                                  ..................................................................................................................26  5.5 Future Directions                                                                                                                                              .............................................................................................................................27  6. Conclusion                                                                                                                                                                .............................................................................................................................................27  47. References                                                                                                                                                                .............................................................................................................................................28  Works Cited                                                                                                                                                                  ...............................................................................................................................................28  Figure 1. Mountain Pine Beetle (Province of British Columbia, 2010).........................................................6 Figure 2. Mountain Pine Beetle Phenology (Wulder, Dymond, White, Leckie, & Carroll, 2006)..................7 Figure 3. Fading Rate of Lodgepole Pine (Wulder, Dymond, White, Leckie, & Carroll, 2006)......................9 Figure 4. Spatial Resolution Comparison. Photo on the left is an aerial photography that has 0.5m resolution. The other picture is showing a Landsat imagery with a pixel size of 30m. (Province of British Columbia)..................................................................................................................................................10 Figure 5. Illustration of Low, Medium, and High Spatial Resolutions. From low to high resolution, they are 30x30m, 4x4m, and 1x1m. (Wulder, Dymond, White, Leckie, & Carroll, Surveying mountain pine beetle damage of forests: A review of remote sensing opportunities, 2006)............................................13 Figure 6. Spectral Characteristics of Green-Attack and Non-Attack Trees (Wulder, Dymond, White, Leckie, & Carroll, Surveying mountain pine beetle damage of forests: A review of remote sensing opportunities, 2006)..................................................................................................................................14 Figure 7. Wavelength Reflectance of Different Tree Conditions (Wulder, Dymond, White, Leckie, & Carroll, Surveying mountain pine beetle damage of forests: A review of remote sensing opportunities, 2006)..........................................................................................................................................................15 Figure 8. Electromagnetic Spectrum (University of Nebraska-Lincoln, 2005)............................................18 Figure 9. Double Sampling Example. The large polygon is the study area and the squares are photo-plots. The white squares are ground plots as well. (Wulder, Dymond, & White, Remote sensing in the survey of mountain pine beetle impacts: Review and recommendations, 2005).....................................................20 Figure 10. Multi-Stage Sampling Example. Primary samples of the study area are shown in picture A. Squares are secondary samples (photo-plots) and circles are tertiary samples (ground plots), as shown in picture B. (Wulder, Dymond, & White, Remote sensing in the survey of mountain pine beetle impacts: Review and recommendations, 2005).......................................................................................................20 Figure 11. Cross-track Scanner and Along-track Scanner (Coops, 2011)....................................................22 Figure 12. NDMI Annual Trajectories.........................................................................................................24 Table 1. Accuracy: Error Matrix Example (Wulder, White, Coops, Alvarez, Butson, & Yuan, 2006)...........10 Table 2. Characteristics and Sensor Spatial Requirement of Different Population States (Wulder, Dymond, White, Leckie, & Carroll, Surveying mountain pine beetle damage of forests: A review of remote sensing opportunities, 2006)........................................................................................................12 Table 3. Remote Sensing Application Costs and Spatial Resolutions.........................................................17 51. Introduction Dendroctonus ponderosae (mountain pine beetle) is causing significant problems to forestry in the interior of BC (Figure 1). It infests mature Pinus contorta (lodgepole pine) and in just 5 years, the beetle expansion increased more than 6-fold from 164,999 ha in 1999 to 7,089,901 ha in 2004. Beetle-killed wood die gradually in three stages, green, red, and then grey. Two main factors that have caused the beetle outbreak are a very successful fire suppression program and increased minimum temperature in the interior of BC. Mapping beetle impact is important because it provides detailed information on forest health for planning, modeling, forest inventory updates, sanitation logging, and block layout. The information required for projects is determined by the level of detail and precision required (Wulder, Dymond, White, Leckie, & Carroll, Surveying mountain pine beetle damage of forests: A review of remote sensing opportunities, 2006). Figure 1. Mountain Pine Beetle (Province of British Columbia, 2010) Remote Sensing has been used since the 1960’s to detect mountain pine beetle impact. But because of the beetle outbreak in the 2000’s, more remote sensing research have been conducted to record the extensive damage to forest areas caused by the beetles. Due to advancement of a variety of remote sensing devices and improvement in image-processing capabilities, more information about the beetle impact can be obtained (Wulder, White, Coops, Alvarez, Butson, & Yuan, 2006). This paper discusses applications of different remote sensing instruments, namely aerial photography and satellite imagery about mountain pine beetle detection. 2.1 Mountain Pine Beetle Biology The current Mountain Pine Beetle infestation was established in the northern region of British Columbia in the early 1990’s (Nelson, Boots, Wulder, & Carroll, 2007). The infestation is endemic to British 6Columbia as a natural disturbance but it has become an epidemic infestation because of climate change. It has a one year life cycle and it is a small, cylindrical bark beetle species (Wulder, Dymond, White, Leckie, & Carroll, Surveying mountain pine beetle damage of forests: A review of remote sensing opportunities, 2006). The areas most heavily attacked by the mountain pine beetle are Houston and Prince George in British Columbia, and the Alberta border (Lee, Kim, & Breuil, 2006). The primary tree species attacked by the mountain pine beetles is lodgepole pine in terms of the commercial impact (Wulder, Dymond, White, Leckie, & Carroll, Surveying mountain pine beetle damage of forests: A review of remote sensing opportunities, 2006). It feeds on fungi and the phloem of the tree (Lee, Kim, & Breuil, 2006). It attacks mature trees by boring into the bark and creating tunnels in the phloem, which is the layer between the bark and the xylem (Bentz, Logan, & Amman, 1991). This bark beetle species attacks its host species in July and August, when trees are often stressed by water deficiency (Lee, Kim, & Breuil, 2006). In the winter, mountain pine beetle larvae stay under the bark of the tree and they continue to digest until spring. The larvae then turn into pupae in June and July and the adults arise from the infested trees over the summer and into the early fall to cause further infestations (Figure 2) (Bentz, Logan, & Amman, 1991). Figure 2. Mountain Pine Beetle Phenology (Wulder, Dymond, White, Leckie, & Carroll, 2006) Mountain pine beetles are susceptible to cold weather, especially the eggs, pupae, and young larva. It can be killed successfully in the cold weather under the following conditions: persistent temperatures under -35 Celsius to -40 Celsius for a few consecutive days in mid-winters, maintained temperatures of -25 Celsius in early fall or late spring. The most effective time to kill the beetles with cold weather is in the fall because the mountain pine beetle has not developed it’s natural anti-freeze levels at that time. Another effective time to kill the mountain pine beetle is before a deep layer of snow accumulates, which acts as an insulator against the chill if the beetle is in the base of the tree (BCMOF, 2010). Tree defence is overcome by the mountain pine beetles with two tactics: attacking in large numbers and inoculating trees with invasive fungi. Mass attack is when mountain pine beetles cooperatively attack a selected tree to overcome its defence (Wulder, Dymond, White, Leckie, & Carroll, 2006). 72.1 Relationship with fungi The mountain pine beetle lives in a symbiotic relationship with the fungi that it carries. In BC, nine fungi species are found to associate with the beetles (Lee, Kim, & Breuil, 2006). Before maturing beetles break out of the pupal chambers, they feed on fungi and fungal spores are deposited in their mycangia, guts, and exoskeleton (Lee, Kim, & Breuil, 2006). The beetles transmit the fungi to the trees when they attack new hosts (Lee, Kim, & Breuil, 2006). The most studied fungus species associated with the mountain pine beetle is Ophiostoma spp. (blue stain fungus) (Parks Canada, 2009). To initiate tree invasion, the blue-stain fungus extends its mycelia into the sapwood quickly (Parks Canada, 2009). The blue stain fungus helps the beetles attack the tree successfully by lowering the tree host’s defense and slowing the movement of fluids (Parks Canada, 2009). The blue-stain fungi obstructs the tubes that bring water up into the tree, which causes problems for the tree to produce sticky sap. The sap production is a tree defence mechanism to trap invading insects. The combined damages caused by the blue-stain fungus allow the mountain pine beetles to dig galleries and lay eggs while staying clear of the tree’s defences (Wulder, Dymond, White, Leckie, & Carroll, 2006). 2.2 Tree attack stages Trees attacked by mountain pine beetles go through three stages: green attack, red attack, and grey attack. During green attack, beetles have already invaded the trees and the needles will stay green for a few months after the tree dies. One year after tree infestation, infested tree foliage turns red and the infestation is now in red attack stage due to nutrient supply being eliminated by the beetle invasion (Figure 3) (BCMOF, 2010). At the same time, chlorophyll and other pigment molecules were broken down (Wulder, Dymond, White, Leckie, & Carroll, Surveying mountain pine beetle damage of forests: A review of remote sensing opportunities, 2006). Two to four years after red attack initiated, the trees start to turn grey while displacing their red needles and exposing bare branches. Grey attack trees decay as time passes and are sensitive to falling and blow-down (BCMOF, 2010). 8Figure 3. Fading Rate of Lodgepole Pine (Wulder, Dymond, White, Leckie, & Carroll, 2006) 3. Remote Sensing Background 3.1 Definition Remote sensing is the measurement or acquisition of information from an object by a recording device that is not in contact with the object. Examples include taking pictures by an aircraft, observing the earth and weather through satellites, and examining fetal development through ultrasound (Lillesand, Kiefer, & Chipman, 2004). In forestry, remote sensing involves of sensors aboards an aircraft or spacecraft for forest characteristics measurement (Province of British Columbia). 3.2 Spectral, Spatial, and Temporal Resolution Resolution is the capability of a sensing device to detect features in the spatial, spectral, and temporal disciplines (Lillesand, Kiefer, & Chipman, 2004). Spatial resolution is the smallest distance of unit that can be resolved by a sensor. Various sensors have different spatial resolutions; a sensor that has a small pixel size has a high spatial resolution (Figure 4). Spectral resolution depends on three criteria: the number of spectral bands used, their location in the electromagnetic spectrum, and the bandwidth of each band. From low to high spectral resolution, the categories are panchromatic, multispectral, and hyperspectral (Lillesand, Kiefer, & Chipman, 2004). Temporal resolution is the time it requires for a satellite complete an orbit around the earth. For example, the temporal resolution of Landsat is 16 days but for a NOAA satellite is four hours. In other words, the NOAA satellite has a higher temporal resolution than Landsat (University of Wisconsin-Madison). As a general rule, as spatial resolution increases, spectral resolution decreases (Wulder, Dymond, & White, Remote sensing in the survey of mountain pine beetle impacts: Review and recommendations, 2005). 9Figure 4. Spatial Resolution Comparison. Photo on the left is an aerial photography that has 0.5m resolution. The other picture is showing a Landsat imagery with a pixel size of 30m. (Province of British Columbia) 3.3 Assessing Accuracy Accuracy of object detection provides to the researcher with the success of detection methods used and the identification of possible error sources. Accuracy in remote sensing is shown using the error matrix, which allows a quantitative comparison of classes between mapped results with validated data (Wulder, White, Coops, Alvarez, Butson, & Yuan, 2006). For example, to validate red attack detection in an area, a ground survey crew needs to go to the forest to see whether the detection is accurate or not. Overall accuracy is the percentage of sample points that are classified correctly along the diagonal of error matrix divided by the total. In the example given, the overall accuracy is 96.5% (Table 1). For the overall accuracy to map accurately, the classes need to be of equal importance (Wulder, White, Coops, Alvarez, Butson, & Yuan, 2006). In table 1, both ground-validated sites of attacked and non-attacked trees are equally represented, so the overall accuracy represents the map accuracy. Table 1. Accuracy: Error Matrix Example (Wulder, White, Coops, Alvarez, Butson, & Yuan, 2006) 10Two measures of accuracy are producer’s and user’s accuracy. Producer’s accuracy is the percentage of ground class correctly classified. In table 1 for instance, 97 out of 101 ground-validated attacked trees were correctly classified. Contrastingly, omission error is the percentage of reference sites for a specific class which are omitted from the remotely sensed class. User’s accuracy is the percentage of correctly classified remotely sensed sites. In table 1, the user’s accuracy is 97%. On the other hand, commission error is the quantity of reference sites for a specific class that were mistakenly included in the remotely sensed class (Wulder, White, Coops, Alvarez, Butson, & Yuan, 2006). The commission error is 3% in the example (Table 1). 114. Using Remote Sensing to detect Mountain Pine Beetles The effects of spatial, spectral, and temporal resolution on the attack characteristics require the researcher to consider the most appropriate remote sensing application for mountain pine beetle detection. 4.1 Spatial resolution There are 4 different population level states: endemic, incipient, epidemic, and post-epidemic (Wulder, Dymond, White, Leckie, & Carroll, Surveying mountain pine beetle damage of forests: A review of remote sensing opportunities, 2006). Remote sensing applications used to detect beetle infestation are different depending on the population state (Table 2). Table 2. Characteristics and Sensor Spatial Requirement of Different Population States (Wulder, Dymond, White, Leckie, & Carroll, Surveying mountain pine beetle damage of forests: A review of remote sensing opportunities, 2006) Mountain pine beetle population level Forest damage characteristics Spatial resolution requirements Endemic Single or small groups of trees High Incipient Small to large groups of trees High or moderate Epidemic Large groups of trees over large areas Moderate In the endemic population state, mountain pine beetles concentrate on weak trees and are spread sparsely over mature lodgepole pine forests. High spatial resolution imagery, 5mx5m or smaller, is required to detect this population state but even that may not work because tall and healthy trees may make the shorter infested trees undetectable (Figure 5) (Wulder, Dymond, White, Leckie, & Carroll, Surveying 12mountain pine beetle damage of forests: A review of remote sensing opportunities, 2006). Figure 5. Illustration of Low, Medium, and High Spatial Resolutions. From low to high resolution, they are 30x30m, 4x4m, and 1x1m. (Wulder, Dymond, White, Leckie, & Carroll, Surveying mountain pine beetle damage of forests: A review of remote sensing opportunities, 2006) An incipient population can successfully mass attack a big diameter lodgepole pine. Depending on the coverage required, high or medium spatial resolution can detect incipient populations (Wulder, Dymond, White, Leckie, & Carroll, Surveying mountain pine beetle damage of forests: A review of remote sensing opportunities, 2006) When an incipient population becomes larger and combine into a bigger population, it becomes an epidemic population. An epidemic population impacts large number of mature lodgepole pine trees, which can be detected with a moderate spatial resolution sensor, which is in the range between 5x5m to 30x30m  (Wulder, Dymond, White, Leckie, & Carroll, Surveying mountain pine beetle damage of forests: A review of remote sensing opportunities, 2006). The epidemic population may then decline due to adverse conditions such as very cool weather, depletion of lodgepole pine hosts, and/or increased resistance of host (Wulder, Dymond, White, Leckie, & Carroll, Surveying mountain pine beetle damage of forests: A review of remote sensing opportunities, 2006). 4.2 Spectral resolution Green attack and red attack have different spectral properties. Because green attack is a non-visual stress, it is very difficult to detect using remote sensing and its spectral signature overlaps greatly with a healthy 13tree’s spectral signature (Figure 6). Green attack may be detected by observing the water stress of foliage and is detectable within 45 to 90 days since attack initiation. However, detection of green attack is poor because branches and other background objects in the forest obstruct foliage detection (Wulder, Dymond, White, Leckie, & Carroll, Surveying mountain pine beetle damage of forests: A review of remote sensing opportunities, 2006). Figure 6. Spectral Characteristics of Green-Attack and Non-Attack Trees (Wulder, Dymond, White, Leckie, & Carroll, Surveying mountain pine beetle damage of forests: A review of remote sensing opportunities, 2006) Red attack has visual symptoms and displays high reflectance of wavelengths in the 850 to 1100nm range that is caused by decline in moisture content (Figure 7). However, the decline in foliage moisture content may not be due to mountain pine beetle infestation; it may be due to other insects, root rot, fungi, or drought (Wulder, Dymond, White, Leckie, & Carroll, Surveying mountain pine beetle damage of forests: 14A review of remote sensing opportunities, 2006).  Figure 7. Wavelength Reflectance of Different Tree Conditions (Wulder, Dymond, White, Leckie, & Carroll, Surveying mountain pine beetle damage of forests: A review of remote sensing opportunities, 2006) Complexities arise in remote sensing application selection when different population level is combined with different attack stage. For example, red attack epidemic population is easy to detect due to explicit spectral properties over large areas. However, red attack endemic population needs to be combined with a high spatial resolution sensor due to the population’s small and sparse concentration (Wulder, Dymond, White, Leckie, & Carroll, Surveying mountain pine beetle damage of forests: A review of remote sensing opportunities, 2006). 4.3 Temporal Resolution Foliage temporal resolution is related to its fade rate. Fade rate is the rate in which foliage changes color and it is caused by climate and phenology. Different trees may have different fade rates; tt relies upon tree genetics and condition, and the local environments (Figure 3) (Wulder, Dymond, White, Leckie, & Carroll, Surveying mountain pine beetle damage of forests: A review of remote sensing opportunities, 2006). 155. Remote Sensing Instruments and Their Capacities to detect Beetle Attack 5.1 Introduction Use of remote sensing for forestry in BC started with aerial surveys back in 1946 with increased usage in the 1950’s. The benefits of using aerial surveys include: low cost per hectare, fast, and good positional accuracy. It has since become the operational standard in BC for red attack surveys (Wulder, Dymond, & White, Remote sensing in the survey of mountain pine beetle impacts: Review and recommendations, 2005). Beetle infestation is visually interpreted or analyzed using image processing software (Wulder, Dymond, & White, Remote sensing in the survey of mountain pine beetle impacts: Review and recommendations, 2005). Techniques such as double sampling and multi-stage sampling are used to see how accurate the remotely sensed data interpretation is (Wulder, Dymond, & White, Remote sensing in the survey of mountain pine beetle impacts: Review and recommendations, 2005). Just like aerial surveys, data interpretation made from satellite imagery can be verified by conducting ground surveys. When verifying for beetle infestation, ground crews hike into the forest to prove that most of the affected trees are purely infested by mountain pine beetles and not other agents (NASA, 2010). A remote sensing study consists of data acquisition (sketches or images), pre-processing (digitizing sketches, scanning images, georeferencing images), processing (interpretation, classification), and validation (ground-checking and accuracy assessment (Wulder, Dymond, & White, Remote sensing in the survey of mountain pine beetle impacts: Review and recommendations, 2005). Data acquisition cost depends on the sensor used and the size of the study area. Aerial surveys are low cost if the weather is suitable and waiting and travelling time for the aircraft are lessened. Satellite imagery is expensive if the satellite is ordered to target a specific area at a certain time when viewing condition is optimal. Otherwise, it is cheaper to acquire than aerial photographs (Wulder, Dymond, & White, Remote sensing in the survey of mountain pine beetle impacts: Review and recommendations, 2005). With the availability of free satellite imagery taken from Landsat 7 ETM+  since October 1st, 2008, it is a cost-free option (U.S. Geological Survey, 2008). Cost trade-off exists between area size and the spatial resolution desired. Generally, a sensor with high spatial resolution has a lower image extent. For example, Landsat TM satellite has spatial resolution of 30x30m, covers about 32,225 km2, and it used to cost $0.02 per km2 (Table 3) (Wulder, Dymond, & White, Remote sensing in the survey of mountain pine beetle impacts: Review and recommendations, 2005). On the other hand, IKONOs multispectral sensor has spatial resolution of 4x4m and covers about 100 km2 at a higher cost of $7 per km2. Processing cost depends on the sensor and study design. For instance, the cost of aerial sketch mapping is around 10 percent of aerial photograph interpretation cost because it takes about 14 hours more to do (Wulder, Dymond, & White, Remote sensing in the survey of 16mountain pine beetle impacts: Review and recommendations, 2005). In the validation process if the study uses multi-stage sampling, 55% of the study cost can be tied to ground surveys. Table 3. Remote Sensing Application Costs and Spatial Resolutions  Aerial Sketch Mapping Aerial Photographs Landsat IKONOS Quickbird Spot-5 Cost High High Free High High Medium Spatial Resoltuio n High Very high 30m 4m 2.4 or 2.8m 20m  Aerial sketch mapping Aerial Photographs Landsat IKONOS Quickbird Spot-5 Cost High High Free High High Medium Spatial Resolutio n High Very high 30m 4m 2.4m 10m Swath Width N/A N/A 185km 11km 16.5km or 544km 2250km Accuracy N/A 92.60% 71% 70% to 93% N/A 71% Ground Track N/A N/A 16 days 11 days 1 to 3.5 days 1-4 or 26 days at nadir 5.2 Aerial Overview Survey and Aerial Photography 5.2.1 Introduction Aerial photographs are the most commonly obtained remotely sensed imagery and they are in high spatial resolution. True color and color infrared photographs are used to map numerous forest damaging agents at different scales (Wulder, Dymond, White, Leckie, & Carroll, Surveying mountain pine beetle damage of forests: A review of remote sensing opportunities, 2006). Interpretation of aerial photographs involves looking for pattern, size, color, shadow, texture, and site association (Lillesand, Kiefer, & Chipman, 2004). Aerial overview surveys require technicians to draw on a map the mountain pine beetle damage. 175.2.2 How It Works The basic concepts of aerial photography include the type of film, scale, focal length, overlap, and photo identification. Films are either in black and white, color, infrared, or false colour infrared (Figure 8) (University of Nebraska-Lincoln, 2005). Color-infrared film, used sometimes for mountain pine beetle applications captures the near-infrared portion of the electromagnetic spectrum, then colors are assigned to the bandwidths so that users can see the near-infrared effects. Figure 8. Electromagnetic Spectrum (University of Nebraska-Lincoln, 2005) 5.2.3 Applications and Techniques 5.2.3.1 Broad Scale Red attack detection is done through general and detail surveys. Through aerial overview surveys (general survey), the government of BC is primarily interested in remotely sensing red attack to complement strategic planning and to minimize the loss of timber values. The information is used to define management zones, allocate forest resources to mitigate infestation, select areas which may need more detailed surveys. For tactical planning during epidemics, aerial surveys provide valuable data (Wulder, Dymond, White, Leckie, & Carroll, Surveying mountain pine beetle damage of forests: A review of remote sensing opportunities, 2006) A general survey such as aerial overview survey is generally for non-operational purposes and uses a fixed-wing aircraft. The aircraft technician draw the captured info such as beetle location, extent, and severity onto either a hardcopy map or a tablet computer at the regional scale of 1:100,000 or 1:250,000. It is then followed up with a detailed survey for operational purposes (Wulder, Dymond, White, Leckie, 18& Carroll, Surveying mountain pine beetle damage of forests: A review of remote sensing opportunities, 2006). Aerial overview survey is beneficial in terms of cost and mapping efficiency. It is cost effective because it provides an extensive of a range of forest health issues in a limited time-frame cheaper than any other remote sensing applications. Also, an experienced technician can interpret cues such as tree species, knowledge of insect habitat, past infestation zones, and the spatial traits of different insects to accurately map the extent and severity of infestation (Wulder, Dymond, White, Leckie, & Carroll, Surveying mountain pine beetle damage of forests: A review of remote sensing opportunities, 2006). Another great advantage is that it can be used to find which areas need more detailed surveys. There is a number of issues associated with aerial interpretation however. Off-nadir viewing (oblique view) can produce location errors which make it implausible to send out a ground crew. Aerial surveys data is inconsistent and has low accuracy of infested tree quantity estimates. Its infested tree quantity estimates differ from the actually number of trees by -40 to 73 percent, which increases with higher red- attack density. Infested areas estimated using aerial surveys are 34 percent bigger than areas identified in aerial photographs (Wulder, Dymond, White, Leckie, & Carroll, Surveying mountain pine beetle damage of forests: A review of remote sensing opportunities, 2006) It is recommended that because of the inconsistency of aerial survey data, confidence intervals and other statistical error estimating tools should be used to inform the reliability of infested area information. The aircraft observer needs more training and data collection standards need to be more rigorous and extensive. Infestation position accuracy can improve by using satellite imagery such as Landsat ETM+  as part of the base map because it provides a continuous view of the landscape and the extra data it provides allows for higher red-attack mapping accuracy. Another method is to implement a digital sketch mapping system which can incorporate GPS and GIS, but it is costly to implement (Wulder, Dymond, White, Leckie, & Carroll, Surveying mountain pine beetle damage of forests: A review of remote sensing opportunities, 2006). 5.2.3.2 Detailed Scale A detailed survey consists of interpreting aerial photography or helicopter GPS on a rotary or fixed-wing aircraft at the landscape and local scales of 1:40,000 and 1:50,000 to map location, extent, and severity of beetle infestations for operation accurately. The data is then used by ground crew to initiate a more detailed ground survey (Wulder, Dymond, White, Leckie, & Carroll, Surveying mountain pine beetle damage of forests: A review of remote sensing opportunities, 2006) 19Aerial photography is used with the double sampling method. Double sampling happens when the samples are measured twice and the great sample size is used for the statistical inference of the population. The U.S. Forest Service uses a technique known as “double sampling for stratification,” where aerial photographs are taken of weighted samples, timber volume classes, and within each class, precise volume estimates are measured using ground plots. Another variation is “double sampling with regression,” which connects aerial photographs and ground survey information to estimate the larger population (Figure 9). The advantage of this method is that it is informative while minimizing expensive ground plots. Also, cheaper aerial photographs become higher quality data (Wulder, Dymond, & White, Remote sensing in the survey of mountain pine beetle impacts: Review and recommendations, 2005).     Figure 9. Double Sampling Example. The large polygon is the study area and the squares are photo-plots. The white squares are ground plots as well. (Wulder, Dymond, & White, Remote sensing in the survey of mountain pine beetle impacts: Review and recommendations, 2005)  Multistage sampling is another technique that is used with aerial photography and for this technique, sub- samples are chosen within samples for sample unit estimation (Figure 10). This technique’s advantage is that sampling efficiency increases as cost reduces. The primary samples are areas surveyed within a large mountain pine beetle infestation area using aerial overview survey. Aerial photographs are used as secondary sampling units that cover 10% of the primary sampling unit areas and are used to estimate beetle impact on timber volume. Subsequent sub-sample of 1% of area is used as field verification (Wulder, Dymond, & White, Remote sensing in the survey of mountain pine beetle impacts: Review and recommendations, 2005). Figure 10. Multi-Stage Sampling Example. Primary samples of the study area are shown in picture A. Squares are secondary samples (photo-plots) and circles are tertiary samples (ground plots), as shown in picture B. (Wulder, Dymond, & White, Remote sensing in the survey of mountain pine beetle impacts: Review and recommendations, 2005) 20At the landscape level which generally deals with timber supply reviews and resource management planning, the use of aerial photography is popular. This level provides data for implementing broad objectives such as harvesting schedule and road construction (Wulder, Dymond, White, Leckie, & Carroll, Surveying mountain pine beetle damage of forests: A review of remote sensing opportunities, 2006). At this level, estimates of the number of red-attack trees can be estimated and which areas need more detailed ground surveys can be decided (Wulder, Dymond, White, Leckie, & Carroll, Surveying mountain pine beetle damage of forests: A review of remote sensing opportunities, 2006). The BC government focuses on sites with minimal level of new infestation, which is detectable at the local level, so that it can be taken care of before the infestation expands out. The information needed is the quantity, location, and species of trees in red attack stage. Red-attack area data is used to determine probable green-attack trees location. Ground survey is then conducted to look for damaged trees and develop a block layout for salvage logging (Wulder, Dymond, White, Leckie, & Carroll, Surveying mountain pine beetle damage of forests: A review of remote sensing opportunities, 2006). At the local level, Helicopter GPS is the benchmark for operational accuracy. This method has horizontal accuracy of about 20m and a low error of commission of less than 5% (Wulder, Dymond, White, Leckie, & Carroll, Surveying mountain pine beetle damage of forests: A review of remote sensing opportunities, 2006). However, it contains positional error due to flying altitude and velocity, and attribute errors such as quantifying the number of infested trees. Positional error can be as high as 45m at a flying altitude of 500m with a 5 degree view angle. Also, infestation number may be hard to estimate due to lack of experience, plane velocity, and the weather (Coops, Wulder, & White, Assessment of QuickBird high spatial resolution imagery to detect red attack damage due to mountain pine beetle infestation, 2006). 5.3 Satellite Imagery 5.3.1 Introduction Space remote sensing has helped people to understand earth’s processes better and the impact of humans on the planet. The benefits of space remote sensing include the ability to survey vast amount of land, which provides a holistic view of the study area. Space remotely sensed data can also be used with forest inventory data and spatial data such as roads, elevation, and climate data. Interpretation bias is reduced because of the use of automated delineation methods which may increase consistency and reliability of mapping (Coops, Wulder, & White, Assessment of QuickBird high spatial resolution imagery to detect red attack damage due to mountain pine beetle infestation, 2006). 5.3.2 How It Works 21Satellite system concepts are sensor types, orbits, receiving stations, and reference systems. There are two types of sensors: cross-track and pushbroom scanner (Figure 11). The cross-track sensor scans in a back and forth motion perpendicular to the satellite movement direction and each ground resolution cell is scanned one by one. Landsat satellites are examples of this type of sensor (Lillesand, Kiefer, & Chipman, 2004). On the other hand, the pushbroom scanner scans the ground in the same direction the satellite is moving. SPOT, IKONOS, Quickbird are examples of this type of sensor.  Figure 11. Cross-track Scanner and Along-track Scanner (Coops, 2011) The two types of orbits that satellites follow are sun-synchronus and geostationary. Sun-synchronous orbit is orbiting the illuminated side of the earth and geostationary orbit follows the earth movement and remains static on top of a geographical location (Lillesand, Kiefer, & Chipman, 2004). In Canada, the Canada Center for Remote Sensing has 2 ground receiving stations in Cantley, Quebec and Prince Albert, Saskatchewan. Their spatial locations allow the download of imagery from any satellites over Canada (Coops, 2011). 5.3.3 Applications and Techniques 5.3.3.1 Landsat 22The Landsat program started off as “Earth Resources Technology Satellites (ERTSs), which initiated in 1967. Altogether, 7 satellites were created with Landsat-7, which is equipped with the Enhanced Thematic Mapper (ETM+) being the most advanced. It has an orbit period of 16 days and 8 bands including the panchromatic band. While its sensor generally has 30m resolution with a swath width of 185km, it has 15m resolution in the panchromatic band. Detailed 15m resolution images can be produced when the panchromatic band combines with ETM+ bands 1 to 5 and 7 (Lillesand, Kiefer, & Chipman, 2004). It is appropriate for detecting larger infestations at epidemic population levels and not for smaller or spatially distributed beetle infestations at endemic or incipient population levels. Landsat ETM+ has 30m resolution and can detect mountain pine beetle infestation with more than 90% accuracy if the red attack trees distribution zone is larger than 1.5ha (Wulder, Dymond, White, Leckie, & Carroll, Surveying mountain pine beetle damage of forests: A review of remote sensing opportunities, 2006). In one study, Landsat TM had an accuracy of 73.3 percent in red-attack tree detection at the landscape level. The training areas for the study were developed from 50m field plots and aerial photos and maximum likelihood classification was used to classify red-attack trees (Wulder, Dymond, & White, Remote sensing in the survey of mountain pine beetle impacts: Review and recommendations, 2005). Maximum likelihood is one of the most powerful classification system, which uses training class inputs such as spectral signatures and statistical tools such as Bayesian probability, means, and covariances to determine what class the questionable pixel is (Lillesand, Kiefer, & Chipman, 2004). In 2008, BC Ministry of Forest started using NDMI index on Landsat scenes to monitor annual moisture change that is associated with mountain pine beetle attack. NDMI evolved from the Enhanced Wetness Difference Image (EWDI), which looks at change over two-image dates to detect mountain pine beetle impact in the forest. For imagery that contained numerous scenes (ex. 8+ Landsat scenes annually) and were acquired over a long time, NDMI is a better approach to detect mountain pine beetle impact. The index is calculated using the following formulation (BCMOF, 2009). NDMI = [Band 4 – Band 5] / [Band4 +Band 5] Band 4 records near-infrared (NIR) wavelengths and is sensitive to the chlorophyll reflectance. Band 5 records mid-infrared (MIR) and is sensitive to leaf moisture absorbance.  A higher ratio signifies a higher leaf reflectance and lower ratio displays a higher leaf moisture content (BCMOF, 2009). A rapid decrease in NDMI means that the stand was attacked by mountain pine beetles (Error: Reference source not found). For example, in 2004 the blue line that represents a pine stand was attacked. The other stand represented by the pink line was attacked in 2007. A year of attack is given to a Landsat pixel if there is a rapid decrease in NDMI in the 30m area that is represented by the Landsat pixel. The year of attack pixel is then scaled up to the polygon from the Vegetation Resource Inventory (VRI) and the year of attack pixel that occurs most frequently will determine the polygon’s year of attack 23Figure 12. NDMI Annual Trajectories 5.3.3.2 IKONOS The IKONOS satellite was launched on September 24th, 1999. It stays in altitude of 682km while occupying a sun synchronous orbit. It crosses the equator at 10:30 am and has a ground track of 11 days. It has the ability to collect images at angles up to 45 degrees from vertical. Its swath width, the strip of Earth of which imagery is captured, is 11km. This satellite captures images in ground resolution of 4m and its panchromatic band resolution is 1m. Panchromatic bands capture data in black and white but it can be combined with the multispectral bands to enhance image sharpness. One special feature of this satellite is that it can be programmed to follow along features such as power lines and meanders (Lillesand, Kiefer, & Chipman, 2004). IKONOS is appropriate for detecting small groups or individual red-attack trees. However, it is expensive to use which limits its use to situations where practicality of other methods are not feasible (Wulder, White, Coops, Alvarez, Butson, & Yuan, 2006). In stands with less than 5% red-attack trees, red-attack estimates from IKONOS imagery have been shown to have an accuracy of 70% and 93% for stands with 5 to 20% red-attack trees (White, Wulder, & Grills, 2006). 245.3.3.3 Quickbird Quickbird was once the highest resolution satellite available to the public. It launched on October 18th, 2001 and was operated by DigitalGlobe, Inc. It is sun-synchronous at a relatively low altitude of 450km and has a revisit time of 1 to 3.5 days depending on the sensor’s angle and latitude. This satellite’s panchromatic sensor has a resolution of 0.61m and a four-band multispectral sensor with resolution of 2.40m. Its swath width is 16.5km at nadir and its accessible ground swath is 544km (Lillesand, Kiefer, & Chipman, 2004). Quickbird has shown its ability to detect red-attack trees in low to moderate levels. Its high resolution imagery can be used to estimate the total number of damaged and at-risk trees. It can be a substitute for heli-GPS for location and damage estimation if the latter is unavailable (Coops, Wulder, & White, Assessment of QuickBird high spatial resolution imagery to detect red attack damage due to mountain pine beetle infestation, 2006). A high resolution satellite like Quickbird provides one benefit that allows the interpreter to count the number of red-attack pixels as the red-attack damage increases in level. Imagery from regular satellites produce the saturation effect, which is the inability to count individual red-attack trees when stands are interpreted as gravely damaged (more than 25% stand attacked). The survey technician then can no longer quantify each individual red-attack crowns (Coops, Wulder, & White, Assessment of QuickBird high spatial resolution imagery to detect red attack damage due to mountain pine beetle infestation, 2006). 5.3.3.3.1 Normalized difference vegetation index (NDVI) and Red-Green Index (RGI) In one study using Quickbird to detect red-attack, Normalized difference vegetation index (NDVI) was found to be less effective than Red-Green Index (RGI) at differentiating non-attack crowns from red attack crowns. NDVI is a numerical indicator that uses various satellite sensors to analyze whether the observed target contains green vegetation or not. Below is the NDVI formulation. NDVI= near-infrared (NIR)(630-690nm)-red(780-900nm)              near-infrared +red Chlorophyll in leaf cells strongly reflect near-infrared light and greatly absorb visible red light. Non- vegetation structures such as clouds, water, and snow produce a negative NDVI value because they reflect 25NIR greater than red light. The different amount of reflectance of a leaf cell’s NIR and red light allow NDVI to compute a ratio that represents live leaf presence (Lillesand, Kiefer, & Chipman, 2004). Red-green  index (RGI) accents the spectral change in foliage colour from green to red, therefore making it a suitable index to sense mountain pine beetle damage. Statistically, it can differentiate fader and red attack from healthy trees. However, it cannot distinguish between shadowed crowns and non-attack crown (Coops, Wulder, & White, Assessment of QuickBird high spatial resolution imagery to detect red attack damage due to mountain pine beetle infestation, 2006). This can be a problem if the shadowed crowns are attacked by mountain pine beetles because RGI will misinterpret it. 5.3.3.4 SPOT Like Landsat data, SPOT data has been used long term historically and widely available. The SPOT program was developed by the French government and was created as a commercial program for operations. SPOT-5 was launched on May 3rd, 2002. SPOT-5 satellite bears two high resolution geometric (HRG) instruments, a high resolution stereoscopic (HRS) instrument, and a vegetation instrument. The HRG instruments provide 2.5 or 5m resolution panchromatic imagery, 10m resolution in the green, red, and near-infrared multispectral bands, and 20m resolution in the mid-IR band. The HRS instrument uses panchromatic imagery to prepare digital elevation models (DEMs) at 10m resolution. Its vegetation instrument has a very great swath width of 2250km but low spatial resolution, which makes it suitable for frequent and large coverage applications (Lillesand, Kiefer, & Chipman, 2004). In general, Spot-5 imagery has a 71% accuracy which is comparable to Landsat imagery accuracy. Therefore Spot-5 can be considered as a replacement of Landsat if the latter is unavailable, however it is a very expensive alternative (White, Wulder, & Grills, 2006). Even though Spot-5 has a higher spatial resolution than Landsat, it has lesser spectral resolution which gives Landsat the ability to discriminate different vegetation types better. Some studies indicate that Landsat data outperforms Spot-5 data for mountain pine beetle detection (White, Wulder, & Grills, 2006). 5.4 Limitations and Benefits Compared to aerial sensors, satellite sensors have several benefits. First, satellite capture continuous data across the extent of the sampled area but aerial sensors capture images in discrete units. During digital image processing, algorithms in imaging software reduces interpreter’s bias which is sometimes found in visual aerial photography interpretation. Satellite imagery also exerts homogenous sampling effort on every sample unit but aerial surveys are affected by viewing angle, flying conditions, and human error (Wulder, Dymond, & White, Remote sensing in the survey of mountain pine beetle impacts: Review and recommendations, 2005). 26However, satellite sensors have disadvantages as well. Quality can be an issue with satellite imagery because the sun angle is only ideal in beetle active period and the atmosphere should be free of clouds and haze. In other words, there is a short time for data acquisition (Wulder, White, Coops, Alvarez, Butson, & Yuan, 2006). Cost, which includes acquisition and processing cost, is also an issue. Processing cost can be broken down into ground truth data processing, masking of unnecessary objects, and data management (Wulder, White, Coops, Alvarez, Butson, & Yuan, 2006). The mixed pixel problem which is the pixel having both healthy and red-attack trees is a more prominent problem in satellite imagery because the spatial resolution is not as high as in aerial photography (Wulder, Dymond, & White, Remote sensing in the survey of mountain pine beetle impacts: Review and recommendations, 2005). 5.5 Future Directions In the future, new generation satellites which have improved spectral sensing ability such as ASTER or Hyperion, which have excess spectral analysis compared to Landsat satellites might trigger new development in mountain pine beetle detection. Also, high spatial resolution satellites such as CASI and Quickbird are unexplored. Quickbird may be used to calibrate helicopter surveys with its high spatial and attribute accuracy. Heli-GPS survey is cheaper to conduct so samples of Quickbird imagery could be to used in conjunction with heli-GPS to estimate red-attack. Currently, accuracy assessments are inconsistent from satellite imagery interpretation so there needs to be reliable standards for accuracy measurement to appeal to forest companies (Wulder, Dymond, & White, Remote sensing in the survey of mountain pine beetle impacts: Review and recommendations, 2005). Grey attack infestation displays success of mitigation efforts and detection of it at the landscape level allows beetle spread calculations and for planning salvage operations at the landscape level. Grey-attack detection techniques need to be refined in the future (Wulder, Dymond, White, Leckie, & Carroll, Surveying mountain pine beetle damage of forests: A review of remote sensing opportunities, 2006). Currently, green attack is subsequently detected after red-attack is detected. Advancement in future remote sensing sensor and application may allow stand-alone and cheap green-attack detection. The challenge to overcome in the future with satellite imagery uses is correlating the spatial location of measured ground data points to the pixel in the satellite imagery. 6. Conclusion The beetles attack mature trees and cooperatively damages lodgepole pines with  fungi. The different tree attack stages are green, red, and grey, with red-attack being the main concern in beetle infestation detection. Remote sensing uses an aircraft or spacecraft to measure forest characteristics. Image interpretation accuracy depends on the technician and the satellite used. The four different beetle population levels are endemic, incipient, epidemic, and post-epidemic. Mountain pine beetle infestation are characterized spectrally, spatially, and temporally. The different remote sensing applications include 27aerial sketch mapping, aerial photography, Landsat, IKONOS, Quickbird, and Spot-5. Each application has its own benefits and disadvantages. Forests have economic values and they are home to many wildlife creatures. With decaying lodgepole pine forests, many wildlife creatures and forest-dependant communities are impacted negatively. Using remote sensing to detect mountain pine beetle infestations is crucial to help mitigate and prevent susceptible forests from further degradation. Technological advancement in remote sensing technology in the future will help forest managers make wise decisions. A wise managing decision is needed because or else, the damage to our lodgepole pine forests will be catastrophic and we may never see the lodgepole pines that are so valuable to us in British Columbia ever again. 7. References Works Cited BCMOF. (2010). Facts About B.C.’s Mountain Pine Beetle. Retrieved November 25, 2010, from Ministry of Forests, Lands and Natural Resource Operations: http://www.for.gov.bc.ca/hfp/mountain_pine_beetle/MPB_Facts.pdf BCMOF. (2009, July 27). Normalized difference moisture index (NDMI) approach for year of death  determination in Mountain Pine Beetle forest stands. Retrieved December 11, 2010, from Ministry of Forest and Range: http://www.for.gov.bc.ca/hts/rs/mpb_impact/documents/YOD-NDMI_rev2.pdf Bentz, B., Logan, J., & Amman, G. (1991). Temperature-dependent development of the mountain pine beetle (Coleoptera:Scolytidae) and simulation of its phenology. The Canadian Entomologist , 1084-1094. Columbia University. (n.d.). Satellite Remote Sensing and its Role in Global Change Research. Retrieved February 10, 2011, from Center for International Earth Science Information Network: http://www.ciesin.columbia.edu/TG/RS/satremot.html Coops, N. (2011). FRST 443: Remote Sensing for Forestry and Agriculture. Retrieved March 1, 2011, from The University of British Columbia: http://www.forestry.ubc.ca/irss/Courses/FRST443RemoteSensingForestryAgriculture.aspx Coops, N., Wulder, M., & White, J. (2006). Assessment of QuickBird high spatial resolution imagery to detect red attack damage due to mountain pine beetle infestation. Remote Sensing of Environment , 67- 80. Lee, S., Kim, J.-J., & Breuil, C. (2006). Diversity of fungi associated with the mountain pine beetle, Dendroctonus ponderosae and infested lodgepole pines in British Columbia. Fungal Diversity , 22, 91- 105. 28Lillesand, T., Kiefer, R., & Chipman, J. (2004). Remote Sensing and Image Interpretation. New York: John Wiley & Sons. NASA. (2010, September 8). NASA Satellites Reveal Surprising Connection Between Beetle Attacks,  Wildfire. Retrieved February 11, 2011, from NASA: http://www.nasa.gov/topics/earth/features/beetles- fire.html Natural Resources Canada. (2007, September 25). Air Photos 101 - Principles of Aerial Photography. Retrieved February 14, 2011, from National Air Photo Library: http://airphotos.nrcan.gc.ca/photos101/photos101_e.php Natural Resources Canada. (2008, January 29). Tutorial: Fundamentals of Remote Sensing: Temporal  Resolution. Retrieved March 1, 2011, from Natural Resources Canada: http://www.ccrs.nrcan.gc.ca/resource/tutor/fundam/chapter2/06_e.php Nelson, T., Boots, B., Wulder, M., & Carroll, A. (2007). Environmental characteristics of mountain pine beetle infestation hot spots. BC Journal of Ecossytems and Management , 91-109. Parks Canada. (2009, April 15). Mountain Pine Beetle Initiative. Retrieved February 12, 2011, from Parks Canada: http://www.pc.gc.ca/docs/v-g/dpp-mpb/sec3/dpp-mpb3a.aspx Province of British Columbia. (2010). Bark Beetles in British Columbia. Retrieved November 23, 2010, from Ministry of Forests, Lands and Natural Resource Operations: http://www.for.gov.bc.ca/hfp/bark_beetles/index.htm Province of British Columbia. (n.d.). Remote Sensing and Geo-Spatial Applications . Retrieved November 11, 2010, from Ministry of Forests, Lands and Natural Resource Operations: http://www.for.gov.bc.ca/hts/rs/background.html U.S. Geological Survey. (2008, October 1). Landsat 7 ETM+ Data Now Available at No Charge . Retrieved March 23, 2011, from U.S. Geological Survey: http://landsat.usgs.gov/mission_headlines2008.php University of Nebraska-Lincoln. (2005). A Guide to the Practical Use of Aerial Color-infrared Photography  in Agriculture. Retrieved February 14, 2011, from Center for Advanced Land Management Information Technologies: http://www.casde.unl.edu/activities/cir-uses/em-energy.php University of Wisconsin-Madison. (n.d.). Image Processing and Analysis: Temporal Resolution, Part 1. Retrieved December 1, 2010, from Satellite Observations in Science Education : http://www.ssec.wisc.edu/sose/pirs/pirs_m3_topic1_p1.html White, J., Wulder, M., & Grills, D. (2006). Detecting and mapping mountain pine beetle red-attack damage with SPOT-5 10-m multispectral imagery. BC Journal of Ecosystems and Management , 105-118. Wulder, M., Dymond, C., & White, J. (2005). Remote sensing in the survey of mountain pine beetle  impacts: Review and recommendations. Victoria: Pacific Forestry Centre. 29Wulder, M., Dymond, C., White, J., Leckie, D., & Carroll, A. (2006). Surveying mountain pine beetle damage of forests: A review of remote sensing opportunities. Forest Ecology and Management , 221 (1- 3), 27-41. Wulder, M., White, J., Coops, N., Alvarez, M., Butson, C., & Yuan, X. (2006). A Procedure for Mapping  and Monitoring Mountain Pine Beetle Red Attack Forest Damage using Landsat Imagery. Victoria: Pacific Forestry Centre. 30

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.52966.1-0075504/manifest

Comment

Related Items