A comparison of cloud droplet radii measured from space

Cloud droplet effective radius (CDR) can be estimated from the spectral signature of cloud reflectance. The technique has been applied to measurements of the Advanced Very High Resolution Radiometer instrument and more recently to the Moderate Resolution Imaging Spectroradiometer (MODIS). Another technique relies on the directional signature of the polarized reflectance and has been applied to observations from Polarization and Directionality of the Earth's Reflectances (POLDER) onboard Advanced Earth Observation Satellite (ADEOS). Although the latter technique requires very specific conditions, we argue that, when applicable, it is very accurate. A large fraction of successful POLDER estimates are derived from measurements over stratocumulus cloud fields. During portions of 2003, POLDER and MODIS acquired near coincident observations. The data can then be used for an evaluation of the two CDR products. The two datasets are highly correlated over the oceans albeit with a MODIS high bias of about 2 /spl mu/m. The correlation breaks down when POLDER retrieves small droplets (less than 7 /spl mu/m), which occurs over most land surfaces as well as polluted oceanic areas. We discuss the possible causes for biases and errors. Although differences in the two CDR estimates are expected because of the differences in the spatial scale and vertical weighting function, we did not find a fully satisfactory explanation for the bias and lack of correlation over land surfaces. It seems, however, that the spatial variability as seen by MODIS is larger than that deduced from POLDER measurements, in particular over land surfaces.