Graduation date: 2007To effectively study dynamic processes like forest succession over long time periods one must effectively integrate data collected at many different times, locations and spatial scales. The purpose of this research is to integrate forest inventory data collected by the USDA Forest Service’s Forest Inventory and Analysis (FIA) Program with multi-temporal satellite data to better understand early successional forest regrowth patterns and carbon storage in western Oregon forests. To detect and characterize continuous changes in early forest succession however, optical satellite images must first be transformed to a common radiometric scale to minimize sun, sensor, view-angle and atmospheric differences among images. We present a comparison of five atmospheric correction methods used to calibrate a nearly continuous, 20-year Landsat TM/ETM+ image data set (19-images) over western Oregon (path 46 row 29). We found that an automated ordination algorithm called multivariate alteration detection (MAD) (Canty et al., 2004), which statistically locates invariant pixels between a subject and a reference image yielded the most consistent common scale among images. Using the crossnormalized image-series we modeled percent tree cover measurements derived by ground survey and airphoto interpretation to the greater landscape. Developing a series of forest regrowth classes we identified a wide range of successional regrowth pathways 18 years after clearcut harvesting. We observed the propensity for faster regrowth on north facing aspects, shallow slopes and at low elevations. Finally, we utilized two sets of forest inventory data to evaluate a Landsat based curve-fitting model for predicting live forest carbon. At the pixel level, the model tended to over-predict carbon and performed better (i.e., higher correlation, lower RMSE) in the Coast Range ecoregion, likely the result of faster, less variable growth patterns. At the landscape scale, we found that the flux of forest carbon predicted by the curve-fit model was in absolute terms, well within the standard error of the inventory estimates. In the process of evaluating the curve-fit model, we discovered a new method for detecting subtle (i.e., forest to non-forest) land-use shifts with Landsat data. Identifying these types of land-use shifts is critically important to developing a more accurate comprehensive carbon budget from forests. We were also able to identify several potential improvements to estimating live forest carbon with the curve-fitting approach.