A framework to retrieve ice water paths over a broad area by combining observations of surface cloud radar and satellite high-frequency microwave measurements in a physically consistent way is presented. The surface cloud radar provides the statistics of vertical distribution of ice water content, while the satellite retrievals use those statistics to broaden a point measurement to an areal measurement. This study seeks to improve retrievals by adopting newly available ice microphysical properties from recent in situ observations, and by treating single scattering properties based on discrete dipole approximation simulations of realistic nonspherical ice particles. First, a new radar reflectivity-ice water content relation is derived using backscattering cross sections calculated from six types of nonspherical ice particles. Ice water content profiles derived from radar reflectivity profiles are then used to calculate brightness temperatures at satellite observing frequencies by a radiative transfer model, which forms an a priori database for a Bayesian ice water path retrieval algorithm. Comparison between the satellite and surface radar retrievals shows that the two agree better for cases with ice water paths greater than 100 g m-2. For lower ice water paths, the difference between satellite and surface radar retrievals becomes large, presumably because of the insensitivity of the microwave scattering to optically thin ice clouds.