Loading…
US-IALE 2018 has ended
Back To Schedule
Wednesday, April 11 • 4:30pm - 4:45pm
REMOTE SENSING/IMAGE ANALYSIS: Reconstructing Historical Boreal Forest Structure by Using Archived Landsat and National Forest Inventory

Sign up or log in to save this to your schedule, view media, leave feedback and see who's attending!

AUTHORS: Bina Thapa*, Peter Wolter – Department of Natural Resource Ecology & Management, Iowa State University; Brian Sturtevant, Institute for Applied Ecosystem Studies, Northern Research Station, USDA Forest Service; Phil Townsend, Department of Forest and Wildlife Ecology, University of Wisconsin-Madison

ABSTRACT: Recurrence of wide-spread biological disturbances in North American boreal and sub-boreal forests results in substantial economic loses annually and generates biological feedbacks that are not yet fully understood. Defoliator insect outbreaks and associated feedback dynamics are influenced by forest conditions at landscape scales. Effective forecast modeling of future defoliator disturbance dynamics dictates that we must first be able to “hindcast” past insect disturbance events. To perform the latter, we need to know the forest state prior to a disturbance event. Here, we use the U.S. Forest Service’s Forest Inventory and Analysis (FIA) data from three periods (1970s, 1990s, and 2003-2006) to calibrate forest structure models. Near concurrent Landsat Thematic Mapper (TM) sensor data were used with these FIA data to recreate forest state conditions leading up to past insect disturbance events in northern Minnesota. Model calibration results between Landsat reflectance and FIA data for both total forest basal area and relative basal area of preferred host species (Abies balsamea) were poor to moderate (R^2 values 0.27 and 0.43 respectively). Results for relative BA of aspen (Populous tremuloides) and remaining host species (spruce: Picea glauca and P. mariana) yielded substantially better accuracies (R^2 values 0.64 and 0.78; RMSE values 15.56 m^2 ha^-1 and 10.65 m^2 ha^-1, respectively). We determined that FIA data from the early 1990s generated stronger model calibration results compare to models calibrated using late 1970s and 2003-2006 FIA data. We posit four potential contributing factors: 1) differences in time-lag between FIA and Landsat dates used, 2) change in FIA plot design after 1995s, 3) poor spatial accuracy associated with FIA plot center locations and 4) species abundance. Nevertheless, in combining these historical FIA and Landsat data to recreate spatially explicit maps of past forest states we bridge a critical data gap in spruce budworm disturbance dynamics research.

Wednesday April 11, 2018 4:30pm - 4:45pm CDT
Water Tower Parlor