Monitoring succession in temperate conifer forests with remote sensing: Implications for terrestrial carbon budgets

Forest succession is a fundamental ecological process which has significant implications for sustainable natural resource management as well as ecosystem biological, biophysical and biogeochemical processes. The capacity for carbon sequestration of terrestrial ecosystems is influenced by forest successional stage. A new forest ecosystem carbon flux model, which combines components of the ZELIG and CENTURY models, shows that net ecosystem production is strongly age dependent. A typical simulated forest stand in the Pacific Northwest of the United States can be a carbon sink for about 200 years, and eventually becomes carbon neutral. Based on a simplified age structure for the forests in Oregon and Washington, the ten million hectares of commercial forests were a significant carbon source from 1890 to 1990. Due to regrowth, this region is becoming a carbon sink and will remain one unless harvest rates are increased.; Improved knowledge of the age distribution of forests would enhance our ability to estimate regional terrestrial carbon budgets. Remote sensing offers the opportunity to map stand age over large areas, but existing methods are essentially empirical and thus difficult to apply outside the domain in which they are calibrated. A more comprehensive solution comes from integrating a forest ecosystem dynamics model (ZELIG) with a canopy reflectance model (GORT) to understand better the manifestation of forest succession in optical imagery. Changes in the reflectance of forest stands through time are highly nonlinear and strongly influenced by the understory reflectance over the first 15–20 years. Analyses of multitemporal Landsat TM data for the H. J. Andrews Experimental Forest have found the spectral/temporal patterns for young stands match the GORT-ZELIG simulation. Extension of the GORT-ZELIG linkage to the spatial domain reveals spatial patterns of multiresolution optical imagery are diagnostic of tree size. Tests using high resolution IKONOS imagery show the expected behavior of image variance as a function of resolution and indicate the potential for future automated mapping of tree size. Overall, both the spectral/temporal and spatial domains exhibit signatures related to stand age indicating the potential for better mapping of forest successional stage, and thus for improving estimates of regional carbon budgets.