Detecting Jack Pine Budworm Defoliation Using Spectral Mixture Analysis: Separating Effects from Determinants

Insect defoliation is a major disturbance force in forbetween 1987 and 1993, but single date NIR in each year was negatively correlated with budworm levels in 1993 ested ecosystems. Monitoring outbreaks, and estimating (r520.69 and 20.47). This was because hardwood trees the areas affected, is therefore important for both forest within jack pine stands caused higher NIR reflectance managers and forest ecologists. The objective of our study but limited jack pine budworm populations. The 10% was to classify jack pine budworm defoliation levels in NIR difference between pure and mixed jack pine stands Landsat TM imagery recorded previous to and during outweighed the 3–5% increase in NIR due to defoliation the 1990–1995 outbreak in our 450,000 ha study area in and necessitated stratification of the satellite data by tree northwestern Wisconsin (USA). Many previous studies species. Spectral mixture analysis performed on pure jack correlated insect defoliation and remotely sensed imagery pine stands resulted in a strong negative correlation bewith moderate to good success, but it often remained untween the 1993 green needle fraction and the 1993 budclear if actual defoliation effects or other forest attributes worm population data (r520.94). This study was, to our correlated with defoliation were detected. For example, a knowledge, the first that applied spectral mixture analysis larger deciduous component in mixed jack pine stands for forest damage detection, and also the first to use insect will limit budworm populations and thereby defoliation. population measurements as independent field data. These The deciduous component in stands is a determining facmethods, and the separation of determinants and effects tor for insect defoliation, whereas needle discoloration of jack pine budworm defoliation, enabled us to detect and tree mortality are their effect. We used pre-outbreak actual defoliation with high accuracy. Elsevier SciLandsat TM data (1987) to identify determining factors ence Inc., 1999 for jack pine budworm population levels and peak-outbreak imagery (1993) for detecting actual defoliation. Our satellite data were atmospherically corrected using a radiaINTRODUCTION tive transfer model (5S). Spectral mixture analysis was performed using spectrometer measurements of jack pine Insect defoliation is a major disturbance agent in forested needles and bark as representations of surface materials ecosystems, a concern for resource managers, and an im(“endmembers”). The explanatory power of the resulting portant area of research for ecologists. Remote sensing fraction images was evaluated using jack pine budworm has been widely used to monitor insect defoliation (e.g., population data collected at 33 sampling points. NearNelson, 1983; Buchheim et al., 1985; Mukai et al., 1987; infrared reflectance (NIR) increased in defoliated stands Hopkins et al., 1988; Leckie et al., 1989; Muchoney and Haack, 1994; Royle and Lathrop, 1997). Satellite imagery provides managers with rapid assessment of current dam* Department of Forest Ecology and Management, University of age so that stands with high mortality can be salvaged. Wisconsin–Madison, Madison It also provides scientists the opportunity to study insect † College of Natural Resources, University of Wisconsin–Stevens Point, Stevens Point defoliation over large areas so that outbreak dynamics Address correspondence to V. C. Radeloff, Dept. of Forest Ecolcan be related to other environmental parameters and ogy and Management, Univ. of Wisconsin–Madison, 1630 Linden Dr., thus be better understood and possibly forecasted (LuMadison, WI 53706, USA. E-mail: [email protected] Received 2 September 1998; revised 7 January 1999. ther et al., 1997).