Effects of threshold retrievals on estimates of the aerosol indirect radiative forcing

Empirical estimates of the aerosol indirect radiative forcing often rely on threshold cloud retrievals applied to multispectral satellite imagery data. In such retrievals, pixels having radiances that surpass prescribed thresholds are assumed to be overcast even if they are only partially cloud covered. This assumption leads to cloud visible optical depths that are underestimated and droplet radii that are overestimated. As regional cloud cover increases, overcast pixels become more common and the biases in cloud properties decrease. Because aerosol optical depths derived from cloud‐free pixels also increase with regional cloud cover (Loeb and Manalo‐Smith, 2005), the biases in threshold‐derived cloud properties can be mistakenly interpreted as being evidence for the effects of aerosols on clouds. Because of the biases, threshold retrievals of cloud properties are likely to lead to overestimates of the aerosol indirect forcing. A retrieval scheme that accounts for fractional cloud cover within an imager pixel is used to estimate the enhancement in the indirect radiative forcing that arises from threshold cloud retrievals. The enhancements prove to be relatively small, approximately 20%. If cloud liquid water is held fixed to estimate the forcing, the biases in threshold‐derived droplet radii and in the sensitivity of droplet radii to changes in aerosol nearly cancel so that the estimates are almost the same for threshold and partly cloudy pixel retrievals.