Moskal, L.M., Jakubauskas, M. E., Price, K. P., and E. A. Martinko, 2002. High-resolution digital photography for forest characterization in the Central Plateau of the Yellowstone National Park. ASPRS 2002 Annual Conference and FIG XXII Congress, Washington, D.C., April 22-26, 2002.

HIGH-RESOLUTION DIGITAL PHOTOGRAPHY FOR FOREST CHARACTERIZATION IN

THE CENTRAL PLATEAU OF THE YELLOWSTONE NATIONAL PARK

Ludmila M. Moskal1, 2

PhD candidate moskal@ku.edu

Mark E. Jakubauskas1, 2

Assistant Research Scientist, Courtesy Assistant Professor mjakub@ku.edu

Kevin P. Price1, 2

Associate Director, Professor price@ku.edu

Edward A. Martinko1, 3

Director, Professor martinko@ku.edu

1Kansas Applied Remote Sensing Program

2Department of Geography 3Department of Ecology and Evolutionary Biology

2335 Irving Hill Road, University of Kansas Lawrence, KS 66045

ABSTRACT Detailed knowledge of forest structure is an important component in research focusing on forest biodiversity monitoring, carbon budgeting studies, fire modelling and forest inventory estimation. In forest inventory mapping, high spatial resolution multispectral imagery are becoming a valuable and often a critical tool to effective forest resource management planning. In this research, we describe and illustrate the forestry applications of an aerial multispectral digital imaging camera system (DuncanTech MS3100) that was experimentally flown over several study sites in the Central Plateau of the Yellowstone National Park, in July 2001. The purpose of this demonstration is three fold; first we discuss the application of geostatistical methods, co-kriging specifically, to model forest canopy components such as height, canopy thickness, species, density and structure. Second, investigate how estimate forest stand characteristics such as stem counts and crown diameter can be estimated using object based methodology on high resolution imagery. Finally, we plan to examine the spectral and spatial characteristics, captured by the aerial camera system, of seedling regeneration in the 1988 burn sites.

INTRODUCTION

Resource managers traditionally use field and photo interpretation methods for mapping vegetation, but these techniques are time consuming, expensive, and inherently subjective, however new digital and geospatial technologies are beginning to replace the airphoto data with digital information. Specifically, in forest inventory mapping, high spatial resolution multispectral imagery are becoming a valuable and often a critical tool to effective resource management planning. Furthermore, aerial sensors allow a more detailed look at specific areas of interest, and can even provide a means of calibration for predictive models. Here, we describe and illustrate the forestry applications of the Kansas Applied Remote Sensing (KARS) Program multispectral digital imaging camera system (Duncan Tech MS3100) that was experimentally flown over several study areas in the Central Plateau of Yellowstone National Park in July, 2001.

1

The purpose of all consequent research with this data set is three fold;

• To apply geostatistical methods, specifically co-kriging, to model forest

canopy components such as height, canopy thickness, species, density and structure;

• To estimate forest stand characteristics such as stem counts and crown diameter, and;

2

• To examine the spectral and spatial characteristics of seedling regeneration sites burned in the 1988 fires.

Study Area

Figure 1. Location of the study area in the Central Plateau of Yellowstone National Park. The A and B on the map indicate the two regions where KARS DuncanTech MS3100 digital multispectral camera imagery was acquired in July 2001.

The Yellowstone National Park, the

oldest National park in the Unites States, is located in the northwest corner of Wyoming encompasses about 9000 km2. The region is primarily a high, forested plateau. The climate in this region is generally cool with relatively moist springs and dry summers (Marther 1986). The vegetation of the plateau is controlled mainly by elevation, with moisture generally increasing with elevation, and the soil formation related to the geological substrate (Despain 1990). The coniferous forest canopy of Yellowstone is dominated by lodgepole pine (Pinus contorta). However, older stands, approximately 250 to 350 years old are comprised mostly of subalpine fir (Abies lasiocarpa) and Engelmann spruce (Picea engelmanni). Douglas-fir (Pseudosuga menziesii) is also present in the region at lower elevations. Historically, fire was a common natural disturbance in the region. Romme and Despain (1989) have reported fires of such scale occurring in the past, most recently in the early 1700s.

The Central Plateau of the Yellowstone National Park (Figure 1) provides ideal conditions for studying the spectral reflectance characteristics of coniferous forests for several reasons. First, the ecology and succession of the Yellowstone forest have been well-documented (Despain, 1990). Second, much of the Yellowstone coniferous forest occurs as a mosaic of succession stages on extensive, gently rolling plateaus. The forest canopy of Yellowstone is dominated by lodgepole pine (83% of the forested area), minimizing variations potentially introduced by mixtures of tree species, in particular mixtures of coniferous and deciduous species.

The cover type classification for Yellowstone uses a letter code to designate the species (LP = lodgepole pine) and a numerical code (0, 1, 2, 3) to indicate the relative stage of development of the forest stand (Despain 1990). This study focused on three development stages or age classes of the coniferous forest. Forest defined as LP0, indicated by the blue caption in Figure 1, are at the initial post fire regeneration stage, about 0-70 years old. The texture of this type of a stand is usually in the mid and upper range (Franklin et al., 2000; Gerylo et al., 1998; Jakubauskas 1996; Wulder, 1996). The NDVI values for this type of a stand are low but can be a poor indicator of biomass if cover is poor (Cohen and Spies 1992; Running et al., 1986). Other vegetation as well as snags and deadfall can play a large role in the reflectance characteristics of this type of stand.

The LP1 abbreviation, indicated by the blue caption in Figure 1, refers to stands 40 to 150 years post fire. These are the intermediate age stand with a very even canopy and little understory. These type of stands tend to be spatially uniform and have a low texture measure (Franklin et al., 2000; Gerylo et al., 1998; Jakubauskas 1996; Wulder, 1996). The NDVI values for this type of a stand range from mid to high (Cohen and Spies 1992; Running et al., 1986).

The LP3 stands, indicated by the blue caption in Figure 1, are approximately 250 to 350 years old and are comprised mostly of subalpine fir (Abies lasiocarpa) and Engelmann spruce (Picea engelmanni). The vertical structure in these types of stands is very complex, causing high shadows on high-resolution imagery. The texture of this type of a stand, especially on high-resolution imagery, is typically high (Franklin et al., 2000; Gerylo et al., 1998; Jakubauskas 1996; Wulder, 1996). The NDVI values for this type of a stand are usually high but can be influenced by the shadows (Cohen and Spies 1992; Running et al., 1986).

3

Figure 2. Landsat TM (left) showing the flight lines for the aerial imagery acquisition in the Thumb Lake region. Example of the multispectral digital imagery flight line (right).

Field data

Three years of field campaigns have been carried out in the study area in support of the heir described projects as well as other research projects (Moskal et al. 2000). Field sampling during summers of the years 1999 and 2000 were directed toward the acquisition of forest structural and biophysical parameters in the Central Plateau

of Yellowstone National Park at two spatial scales. Particular care was taken in the field to ensure that sites were located in the center of a homogenous stand to avoid edge effects. Broad scale sampling was performed at points on a 1 km grid interval corresponding to major UTM easting/northing intersections. Intensive sampling was performed at three sets of fine grids with points placed at 100m intervals, nested within the coarse sampling grid. The three intensive grid sites were located in areas dominated by each of three distinct age classes of the lodgepole pine cover types. All three intensively sample sites consisted of 50 plots each, and additional 200 plots were collected on the broad grid to be applied in future studies. During the summer of 2001 additional data was collected in the 1988 fire regenerating sites as well as a fourth nested grid in the 1km grid sample. Sites were characterised for forest composition, basal area and tree heights at plots located at 100m, 355m and 500m intervals. The data set for this study area consists of over 500 field sites.

KARS DuncanTech MS3100 digital multispectral camera system

A customised DuncanTech MS3100 digital

multispectral camera (Figure 2 a) was used to collect 40 cm per pixel resolution imagery over two study sites in the Central Plateau of Yellowstone National Park, between July 18th and 20th, 2001. The image area for each frame was 1392 X 1040 pixels or approximately 0.4 by 0.5-km on the ground. A progressive scan was used to acquire clear images of moving targets at frame rates of up to 7.5 fps. Over 20 flight lines were required to cover the West Thumb study site (shown as A in light green in Figure 1). The Lower Geyser Basin overflight required 15 flightlines that were about 15 km in length (shown in Figure 1 in light yellow, under caption B). Three spectral bands collected were blue (450-520nm), red (630-690nm) and near-infrared (760-900nm). Sample imagery and the West Thumb flightlines are shown in Figure 2.

KARS camera internal system set-up The image data was captured from the digital

camera using National Instrument's PCI-1424 Frame Grabber. The camera's output provided parallel pixel data, pixel clock, line valid, and frame valid signals. The digital frame grabber was connected to the digital video output connector on the camera using a 100-pin data cable. A 600Mhz Pentium III computer with 128Mb RAM running Windows 98 was used to control camera configuration and image acquisition software. While in flight, the system saved captured imagery to a 75Gb hard drive. The system also had a CD-R drive for more permanent storage of the imagery. A Sony 15-inch TFT LCD colour monitor was used to view input from either the computer or camera using a video switch box. With a resolution of 1024 x 768 @ 75Hz, this monitor also had high contrast that allows images to be viewed under bright cockpit conditions. The power supply for the computer and camera, consisted of a self-contained power supply system which was constructed

Figure 3. KARS DuncanTech MS3100 digital multispectral camera system set-up. Photo plate a) the DuncanTech MS3100 digital multispectral camera, photo plate b) internal system set-up, and photo plate c) the aircraft.

4

5

using two 12-volt deep-cycle batteries and a Whistler 1500W DC to AC power inverter. The power supply included a digital voltmeter and an emergency power disconnect switch. The computer, camera, inverter, and monitor were remounted to a wooden pallet and connected together. The batteries were mounted on a separate pallet and automotive battery cables used to connect the batteries to the inverter. After determining available mount points in the single engine Cessna 172 aeroplane (Figure 3), holes were drilled in the pallet and the pallets secured to the rear seat rails on the floor of the aircraft. The camera was positioned over a hole cut in the bottom of the aeroplane and secured with nylon straps. The passenger controlled the entire imagery acquisition process using a mini keyboard with touchpad mouse.

GEOSTATISTICAL METHODS

Traditional spectral classification of remotely sensed images applied on a pixel-by-pixel basis ignores the potentially useful spatial information between the values of proximate pixels (Atkinson and Lewis 2000). Image texture is the spatial variation in image tones (Haralick 1979), and has long been recognized by both air photo-interpreters and digital image analysts as a powerful source of information in forestry image analysis (Jensen 1996). With the increased availability of new high spatial detail digital imagery, from airborne (Anger 1999) and satellite platforms (Barnsley 1999), it is apparent that digital image based texture should be considered as a potentially important information source for forestry purposes (Green 2000).

Spatial analysis of forest digital imagery is based on the close link between forest canopy structure and image texture as well as on the relationship between the image pattern and the processes that take place in the scene. Forest stands have been separated by differences in digital texture related to species composition (Franklin et al. 2000), age-class and structure (Cohen et al. 1990), and crown closure (St-Onge and Cavayas 1995, 1997). These authors have concluded that tree size and density greatly influence the spatial structure of high-resolution images of forest stands and that the variogram is an effective tool to measure this structure. Thus, the 40 cm per pixel imagery collected in the 21 flight lines (Figure 2) at the West Thumb area of the Central Plateau, Yellowstone National Park, can be used to extract forest stand characteristics such as per species basal area, tree height and canopy structure (roughness/thickness). Furthermore, geostatistical methods can be used to exploit the spatial information pertaining to forest structure inherent in image data. By calibrating remotely sensed multispectral data with a small number of ground measurements, and information extracted from the aerial sensor, characteristics of the forest measured at sample points can be extrapolated across a large geographic region. This has significant advantages for forest management, especially when forests are in remote or inaccessible locations. Thus, we plan to use the forest measurements and information obtained from the high resolution aerial sensor to develop robust geostatistical models applicable to the Central Plateau forests of Yellowstone.

STAND CHARACTERIZATION METHODS

Similar to the new developments in precision agriculture, precision forestry applications are becoming increasingly useful to forest managers seeking detailed resource information. The spatial and temporal availability of remotely sensed data and the advancements in digital technology are providing the tools modern forest managers have been seeking. However, the conventional methods of dealing with coarse resolution remotely sensed imagery, (such as Landsat TM) are not necessarily suitable for high-resolution imagery analysis. For example, maximum likelihood supervised classification or unsupervised classification are not suitable for the analysis of very high-resolution images such as those captured by high-resolution aerial multispectral sensors. Thus, other methods that are specifically designed for high-resolution sensors are becoming essential to forest managers interested in the application of such data. Image analysis methodologies are actively being developed for the handling of high-resolution data. These methodologies include object oriented image classifications and image segmentation. Following in this direction are a growing group of researchers including: Culvenor (2002), Wulder et al. (2000), Walsworht and King (1999), Gougeon (1995), also see Hill and Leckie (1998). These scientists have been developing computer algorithms that automate the delineation of the tree crowns and stem counting procedures. Many of these applications have focused

6

Figure 4. High-resolution digital multispectral camera based stand characterisation

on the identification, classification and mapping of individual trees and tree crowns (Gougeon 1995; Brantberg 1997; Niemann et al. 1999; Pollack 1998). A few studies have also attempted volume estimation from airborne (Franklin and McDermid 1993; Hall et al. 1998; Magnussen and Boudewyn 1998) and satellite data (De Wulf et al. 1990; Ardo 1992; Gemmell 1995).

The two most

fundamental assumptions to these new approaches are that image resolution is fine enough to capture individual trees with a number of pixels, and that the centre of the tree crown appears

radiometrically brighter then the edge of the tree crown. In coniferous forests the top of the tree crown forms an apex that is illuminated more brightly then the surrounding branches. This apex is captured as a very bright pixel in high-resolution imagery (Figure 4). By applying image-filtering methods that capture this apex pixel, stem counts and per hectare forest densities can be estimated. Furthermore, object oriented image analysis, based on the concept that information necessary to interpret an image is not necessarily represented in single pixels, but in meaningful image objects and their mutual relationships, can be used for tree crown delineation. Using this approach these objects can be quantified, measured and compared to ground based measurements and satellite remotely sensed estimates.

SEEDLING REGENERATION METHODS

Fire is an important agent of change in an ecosystem. The large fires of 1988 in Yellowstone National Park demonstrated how dramatically and rapidly the vegetation and consequently the state of an ecosystem can change. The 250,000 ha of burned forest created a striking mosaic of burn severities on the park landscape. Both the ecological and economic impact of these fires have been significant (YNP 1993; Polzin et al. 1993). The burns have begun to naturally regenerate with lodgepole pine (Pinus contorta) seedlings (Reed et al. 1999). The influence of this regeneration on ecological processes affecting the fauna of that ecosystem will persist for decades (Norland et al. 1996). For example, populations and movements of animals are directly and indirectly influenced by the vegetation present in their habitat (Boyce 1998 and 1997; Merrill et al. 1993; Boyce and Merrill 1991). Therefore, knowing where and how the burns are regenerating is an important aspect of sustainable park management strategies.

Figure 5 illustrates four seedling density regeneration sites. The presence of other vegetation in the very low seedling density regeneration site produces a similar spectral response to other seedling density sites. Thus, the redness in the aerial images indicates the presence of vegetation. However, the spatial component of the aerial data

clearly differs for all four-seedling density regeneration sites because the seedling crowns produce different spatial patterns. Therefore, spatial image information captured by image texture can be incorporated to improve the distinction between these different sites (Moskal et al. 2001).

7

Figure 5. Seedling regeneration as observed in the field and from the KARS DuncanTech MS3100 digital multispectral camera system, top images show low regeneration, bottom images show high regeneration.

CONCLUSIONS

The modelling of forest canopy structure components and characteristics as well as the monitoring of regeneration success based on field data alone is a daunting task involving large amounts of time and financial resources. New geospatial technologies, such as remote sensing, have made information collection possible where field surveying has fallen short due to such factors as cost, time and terrain difficulties (Smith et al. 1990). The spatial extent of remotely sensed data as well as the temporal availability of such data are proving to be a very important tool in the task of accurate and timely forest inventory mapping and thusly, successful sustainable forest management. Therefore, the development of the discussed applications based on a high resolution aerial sensors can prove to be a cost saving forest characterisation tool applicable to many forest management objectives.

ACKNOWLEDGMENTS

This project was conducted at the Kansas Applied Remote Sensing (KARS) Program (Edward A. Martinko, Director). The research described in this paper was funded by the National Aeronautics and Space Administration (NASA) Earth Science Enterprise Food and Fiber Applications of Remote

Sensing (FFARS), Project NAG13-99019. The authors would like to thank the Yellowstone 2001 field campaign team including Matt Dunbar, Dana Peterson, Andrea Repinsky and Dr. Mike Shaughnessy. Finally, acknowledgements are extended to the airborne data acquisition team: Ron Renz (Pilot), Michael Houts (KARS system operator), Dr. Mark Ewing, (University of Kansas Aerospace Engineering) and Dr. David Downing (Kansas NASA EPSCOR).

8

REFERENCES Ardo, J., 1992. Volume quantification of coniferous forest compartments using spectral radiance recorded by Landsat Thematic

Mapper. Int. J. Rem. Sen., (13) 9, 1779-1786. Atkinson, P. M. and P. Lewis, 2000. Geostatistical classification for remote sensing: an introduction. Com. & Geosci., 26(4),

361-371. Avery, T. E., 1978. Forester’s guide to aerial photo interpretation. USDA Forest Service, Agriculture Handbook No. 308. Boyce, M. S., 1998. Ecological-process management and ungulates: Yellowstone’s conservation paradigm. Wildl. Soc. Bull., 26,

391-398. Boyce, M. S., 1997. Review essay: The grizzly bears of Yellowstone. Yellowstone Sci., 5(1), 18-20. Boyce, M. S. and E. H. Merrill, 1991. Ungulate responses to the 1988 fires in Yellowstone National Park. Tall Timbers Fire Ecol.

Proc., 17, 121-132. Brantberg, T., 1997. Towards structure-based classification of tree crowns in high spatial resolution aerial images. Scandinavian

J. Forest Res., 12, 89-96. Chen, J. M., 1996. Evaluation of vegetation indices and a modified simple ratio for boreal applications. Can. J. Rem. Sen., 22

(3), 229-242. Cohen, W. B., and T. A. Spies, 1992. Estimating structural attributes of Douglas-Fir/Western Hemlock forest Stands from

Landsat and SPOT imagery. Rem. Sen. Env., 41(1), 1-17. Cohen, W. B., T. A. Spies, and G. A. Bradshaw, 1990. Semivariograms of digital imagery for analysis of conifer canopy

structure. Rem. Sen. Env., 29, 669-672. Culvenor, D. S., 2002. TIDA: an algorithm for the delineation of tree crowns in high spatial resolution remotely sensed imagery.

Com. & Geosci., 28, 33-44. Despain, D., 1990. Yellowstone vegetation: consequences of environment and history in a natural setting. Roberts Rinehart

Publishers, Santa Barbara. De Wulf, R. R., R. E. Goossens, B. P. De Roover, and F. C. Borry, 1990. Extraction of forest stand parameters from

panchromatic and multispectral SPOT-1 data. Int. J. Rem. Sen., (11) 9, 1571-1588. Dungan, J., 1998. Spatial prediction of vegetation quantities using ground and image data. Int. J. Rem. Sen., 19, 267-285. Franklin, S. E., R. J. Hall, L. M. Moskal, A. J. Maudie, and M. B. Lavigne, 2000. Incorporating texture into classification of

forest species composition from airborne multispectral images. Int. J. Rem. Sen., 21, 61-79. Franklin, S. E., M. Lavigne, M. Deuling, M. Wulder, and E. Hunt, 1997. Estimation of forest leaf area index using remote

sensing and GIS data for modelling net primary production. Int. J. Rem. Sens., (18)16, 3459-3471. Franklin, S. E., 1994. Discrimination of subalpine forest species and canopy density using digital casi, SPOT and Landsat TM

data. In: Photogram. Eng. & Rem. Sens., 60 (10), 11233-1241. Franklin, S., and G. J. McDermid, 1993. Empirical relations between casi spectral response and lodgepole pine (Pinus contorta)

Forest Stand Parameters, Int. J. Rem. Sen., 14, 2331-2348. Gemmell, F. M., 1995. Effects of forest cover, terrain, and scale on timber volume estimation with Thematic Mapper data in a

rocky mountain site. Rem. Sen. Env., 51, 291-305. Gerylo, G., R. J. Hall, S. E. Franklin, A. Roberts, and E. J. Milton, 1998. Hierarchical image classification and extraction of

forest species composition and crown closure from airborne multispectral images, Can. J. Rem. Sen., 24, (3), 219-232. Gougeon, F. A., 1995. A crown-following approach tot he automatic delineation of individual trees in high spatial resolution

digital images. Can. J. Rem. Sen., 21(3), 274-284. Hall, R. J., G. Gerylo, and S. E. Franklin, 1998. Estimation of stand volume from high resolution multispectral images, In

Proceedings, 20th Can. Rem. Sen. Symp., Calgary, AB, 191-196. Haralick, R., 1979. Statistical and structural approaches to texture. Proceedings of the IEEE, 65(5), 786-804. Hawkes, B., D. Goodenough, B. Lawson, A. Thomson, W. Sahle, K. O. Niemann, P. Fuslem, J. Beck, B. Bell, and P. Symington,

1995. Forest fire fuel mapping using GIS and remote sensing in British Columbia. Summ. GIS'95, Vancouver, BC. Heute, A. R., 1988. A Soil-Adjusted Vegetation Index (SAVI), Rem. Sens. Env., 25, 295-309. Hill, D. A. and D. G. Leckie, 1998. Proceedings: Automated interpretation of high-resolution digital imagery for forestry.

Canadian Forest Service, Victoria, BC, 402. Hunt, E. R. Jr, 1994. Relationship between woody biomass and PAR conversion efficiency for estimating net primary production

from NDVI. Int. J. Rem. Sens., 15, 1725-1730. Jakubauskas, M. E., 1996. Canonical correlation analysis of coniferous forest spectral and biotic relationships. Int. J. Rem. Sens.,

17(12), 2323-2332. Knight, D. H., and L. L. Wallace, 1989. The Yellowstone fires: issues in landscape ecology. BioScience 39(10), 700-706. Leckie, D. G., and F. A. Gougeon, 1998. An assessment of both visual and automated tree counting and species identification

with high spatial resolution multispectral imagery, In: Automated interpretation of high spatial resolution digital imagery for forestry, Victoria, B.C., 141-152.

9

Magnussen, S. and P. Boudewyn, 1998. Derivations of stand heights from airborne laser scanner data with canopy-based quantile estimators, Can. J Forest Res., 28, 1016-1031.

Marther, B. E., 1986. Wyoming climate atlas. Lincoln: University of Nebraska Press. Merrill, E. H., R. W. Marrs, M. B. Brodahl, and M. S. Boyce, 1993. Estimations of green herbaceous phytomass in Yellowstone

National Park using remote sensing, J. Range Manage., 46, 151-157. Moskal, L. M., K. Price, M. E. Jakubauskas and E. Martinko, 2001. Comparison of hyperspectral AVIRIS and Landsat TM

imagery for estimating burn site pine seedling regeneration densities in the Central Plateau of Yellowstone National Park. In: Proceedings of The 3rd International Forestry and Agriculture Remote Sensing Conference and Exhibition, Denver, CO.

Niemann, K. O., S. Adams, and G. Hay, 1998. Automated tree crown identification using digital orthophoto mosaics, In Automated Interpretation of High Spatial Resolution Digital Imagery for Forestry, Victoria, B.C., 105-113.

Norland, J.E., F.J. Singer, L. Mack, 1996. Effects of the Yellowstone fires of 1988 on elk habitats, in: Ecological implications of fire in greater Yellowstone. I. Ass. Wildland Fire, Fairfield, WA., 223-232.

Pollack, R., 1998. Individual tree recognition based on a synthetic tree crown image model, In: Automated interpretation of high spatial resolution digital imagery for forestry, Victoria, B.C., 25-34.

Polzin, P. E., M. S. Yuan and E. G. Schuster, 1993. Some economic impacts of the 1988 fires in the Yellowstone area. United States Department of Agriculture, Forest Service, Intermountain Research Station, Research Note INT-418.

Price, K. P, X. Guo and J.M. Stiles, 1999. Discriminant analysis of Landsat TM multi-temporal data for six grassland management practices in eastern Kansas, Proceedings of the ASPRS 1999 Annual Convention, Portland, Oregon, 498-508.

Reed, R. A., M. E. Finley, W. H. Romme and M. G. Turner, 1999. Aboveground net primary production and leaf area index in early postfire vegetation in Yellowstone National Park, Ecosystems, 2, 88-94.

Rogan, J. and Yool, S. R. 2001. Mapping fire-induced vegetation depletion in the Peloncillo Mountains, Arizona and New Mexico. Int. J. Rem. Sen., 22 (16), 3101-3121.

Romme, W.H. and D.G. Despain, 1989. Historical perspective on the Yellowstone Fires of 1988. Bioscience 39(10), 696-699. Root, R. R., and J. W. van Wagtendonk, 1999. Hyperspectral analysis of multi-temporal Landsat TM data for mapping fuels in

Yosemite National Park. Proceedings of the Annual Conf. Amer. Soc. Photog. and Rem. Sen. Portland, OR. Running, S. W., and J. C. Coughlan, 1988. A general model of forest ecosystem processes for regional applications. I: hydrologic

balance, canopy gas exchange and primary production. Processes, Ecol. Model., 42, 125-154. Running, S. W., Peterson, D. L., Spanner, M. A., and K. Teuber, 1986. Remote Sensing of coniferous forest leaf area. Ecology,

67(1), 273-276. Smith, O., M., Ustin, L., S., Adams, B., J., and R. A. Gillespie, 1990. Vegetation in deserts: I. a regional measure of abundance

from multispectral images, Rem. Sens. Env., 31, 1-26. St-Onge, B. A., and F. Cavayas, 1995. Estimating forest stand structure from high resolution imagery using the directional

variogram. Int. J. Rem. Sen., 16, 1999-2021. St-Onge, B. A., and F. Cavayas, 1997. Automated forest structure from high resolution imagery based on directional

semivariogram estimates. Rem. Sen. Env., 61, 82-95. Teillet, P. M., K. Staenz K. and D. Williams, 1997. Effects of spectral, spatial, and radiometric characteristics on remote sensing

vegetation indices for forested regions; Rem. Sens. Env., 61 ,139-149. Tucker, C. J., W. W. Newcomb, S. O. Los, and S. D. Prince, 1991. Mean and inter-year variation of growing season normalised

difference vegetation index for the Sahel 1981-1989. Int. J. Rem. Sens., 12, 1133-1135. Uhl, C., and J. B. Kauffman, 1990. Deforestation, fire susceptibility, and potential tree responses to fire in the eastern Amazon.

Ecology 71, 437-449. Walsworth, N. A. and D. J. King, 1999. Image modelling of forest changes associated with acid mine drainage. Computers &

Geosciences, 25(5), 567-580. White, J. D., K. C. Ryan, C. C. Key, S. W. Running, 1996. Remote sensing of forest fire severity and vegetation recovery. Int. J.

of Wildland Fire, 6 (3), 125-136. Wulder, M., Niemann, K. O. and D. G. Goodenough, 2000. Local maximum filtering for the extraction of tree locations and

basal area from high spatial resolution imagery. Rem. Sen. Env., 73, 103-114. Wulder, M. A., 1996. Airborne remote sensing of forest structure: estimation of leaf area index. Unpublished Master of

Environmental Studies Thesis, University of Waterloo, Waterloo, ON. YNP, 1993. The ecological implications of fire in Greater Yellowstone: second biennial scientific conference on the Greater

Yellowstone Ecosystem, September 19-21, 1993, Mammoth Hot Springs, Yellowstone National Park.