COS 60-7
Assimilating PALSAR remote sensing data to reduce uncertainty in ED2 predictions of forest biomass dynamics following disturbance

Wednesday, August 7, 2013: 9:50 AM
L100H, Minneapolis Convention Center
Brady S. Hardiman , Earth and Environment, Boston University, Boston, MA
Shawn P. Serbin , Forest and Wildlife Ecology, University of Wisconsin - Madison, Madison, WI
Joshua A. Mantooth , Earth & Environment, Boston University, Boston, MA
Robert Kennedy , Department of Earth and the Environment, Boston University, Boston, MA
Michael Dietze , Earth and Environment, Boston University, Boston, MA
Background/Question/Methods

Anthropogenic disturbance is a major and accelerating force shaping contemporary forests globally. The intensity of disturbances such as timber harvesting and agricultural clearing vary widely at local and regional scales and occur against a backdrop of natural disturbance, environmental heterogeneity, and global climate change. Forest responses to interacting sources of disturbance are critically important in determining biomass dynamics in these ecosystems.

The Ecosystem Demography 2.2 (ED2) model simulates forest biomass dynamics over short and long time scales by initializing ecosystem states from age (time since disturbance) and size (growth rate) data. Forest disturbances can be identified from LandTrendr products (from Landsat) and RADAR remote sensing of forest biomass (from ALOS-PALSAR) can quantify the biomass impact of such disturbances as well as spatial variability in biomass of regrowing forests. We combine these two remote sensing data products to characterize variability of forest responses to disturbance in north temperate mixed deciduous forest sites. Our objective is to reduce uncertainty in ED2 predictions of forest biomass dynamics following disturbance. Improved estimates of the effect of intensity, frequency, and scale of disturbance on forest biomass loss and regrowth can lead to more accurate predictions of forest carbon storage.

Results/Conclusions

We are using high resolution, time series synthetic aperture RADAR data to quantify the impact of forest disturbances previously detected by Landtrender on forest biomass in the Chequamegon Ecosystem Atmosphere Study (ChEAS) in northern Wisconsin, USA. Each of five sites has ongoing, long-term, ground-based measures of forest ecosystem processes including forest biomass production, mortality, and reproduction. We calibrated PALSAR backscatter data against ground-based, site-specific measures of forest biomass (R2=0.71) to produce maps of forest biomass. As a test, biomass maps correctly delineated boundaries between ecosystem types in the forest/wetland mosaic of northern WI. Areas identified by LandTrendr as recently disturbed corresponded with observations of abrupt biomass loss between consecutive PALSAR time points. Such areas were consistent with areas of known harvesting activity. Ongoing efforts will expand this process to ~25 forested sites throughout the Western hemisphere in a variety of forest types. Continuous assimilation of PALSAR data will allow us to follow forest responses to disturbance and improve understanding of how various forest types respond to disturbance. These findings will translate into more accurate long-term predictions of the resilience (or vulnerability) of forest growth to disturbances by the ED2 model.