In this study, we developed a new SAR/InSAR-based land-cover classification algorithm for estimating surface roughness, a key parameter that is critical for accurate compound flood modeling. We successfully classified the Greater Houston area and New Orleans into nine classes with distinct surface roughness using L-band ALOS PALSAR-1 data. The ALOS-derived surface roughness estimates show an excellent agreement with those derived from NOAA's Coastal Change Analysis Program (C-CAP, 2010) land-cover classification data acquired around the same time. The new algorithm allows us to fill the temporal gaps in the existing surface roughness database. The radar-derived surface roughness estimates will be fed into the the Advanced Circulation (AD-CIRC) modeling framework to understand the compound flood risk associated with rapid land-cover changes due to urban expansion over the last decade.