Rapid land use changes are substantially altering the global carbon budget, yet quantifying the impact of these changes, or assessing efforts to mitigate them, remains challenging. Methods for assessing forest carbon range from precise ground surveys to remote-sensing approaches that provide proxies for canopy height and structure. We introduce a method for extracting a proxy for canopy heights from Interferometric Synthetic Aperture Radar (InSAR) data. Our method focuses on short-spatial scale differences between forested and cleared regions, reducing the impact of errors from variations in atmospheric water vapor or satellite orbital positions. We generate time-varying, Landsat-based maps of land use and perform our analysis on the original wrapped (modulo-2π) data to avoid errors introduce by phase unwrapping and to allow assessment of the confidence of our results (within 3–4 m in many cases). We apply our approach to the Pacific Northwest, which contains some of the world’s tallest trees and has experienced extensive clearcutting. We use SAR imagery acquired at L-band by the PALSAR instrument on the Japanese Aerospace Exploration Agency’s (JAXA) Advanced Land Observation Satellite (ALOS). As SAR data archives expand, our approach can complement other remote-sensing methods and allow time-variable assessment of forest carbon budgets worldwide.