Huge plumes of smoke are episodically transported westward over the South Atlantic Ocean during the savannah-burning season in southwestern Africa (southern hemisphere winter). A vertical cross section through one of these events is depicted in quick look imagery from the CALIOP lidar instrument on the Calipso satellite (top image with ground track indicated in the inset). A thick layer of heavy smoke extends up to 5 km altitude with a highly reflective layer of stratocumulus cloud beneath. In between is a layer of relatively clear air.

Observations of cloud properties (optical thickness and cloud drop radius) from the Moderate Resolution Imaging Spectrometer (MODIS) instrument on the Aqua and Terra satellites make use of the measured reflectance of visible and near-infrared light from the sun. When a thick layer of smoke resides between the satellite and the cloud, the smoke absorbs a portion of the sunlight before it reaches the satellite detector. This makes the cloud appear darker than it would in the absence of the smoke. This darkening of the cloud results in a low bias in the retrieval of cloud optical thickness reported by MODIS. Cloud optical thickness measurements are used routinely to infer the amount of water in the cloud (a quantity called the liquid water path [LWP]), which, in turn, is useful for understanding the role of the cloud deck in the regional climate.

The magnitude of the bias in MODIS observations owing to the presence of an overlaying layer of smoke can be estimated by comparing two independent observations of LWP: one inferred from the MODIS observations of cloud optical thickness and drop radius, and one inferred from the emission of microwave radiation using the Advanced Microwave Scanning Radiometer - EOS (AMSR-E) instrument on the Aqua satellite. Smoke aerosols are transparent to microwave radiation, therefore systematic differences between the two measures of LWP should reveal the effect of the smoke on the MODIS data. That difference is shown in the bottom-left panel for all passes of the Aqua satellite over the Southeast Atlantic Ocean during August and September 2005 and 2006. The difference in measures of LWP is plotted against the OMI aerosol index, a semi-quantitative measure of the amount of smoke aerosol above the clouds derived from Ozone Monitoring Instrument measurements of reflected ultraviolet radiation. Higher numbers indicate heavier plumes of smoke. The AMSR-E microwave-based estimate of LWP is increasingly greater than the MODIS estimate for greater amounts of smoke because of the influence of the smoke on the visible light reaching the MODIS detector. The bias in the MODIS retrieval of cloud optical thickness implied by the LWP difference is shown in the bottom-right panel. Cloud optical thickness is underestimated by as much as 5 for the smokiest periods. However, uncertainties in measurements of cloud optical thickness and LWP are significant. This is revealed in the scatter of individual samples in the two plots (green symbols). The bias in cloud optical thickness attributable to the presence of smoke only exceeds the magnitude of other uncertainties in the measurement for scenes with OMI aerosol index greater than 2, which occurs only infrequently.

Reference: Eric Wilcox, Harshvardhan, and Steven Platnick, “An estimate of the impact of absorbing aerosol over cloud on the MODIS retrievals of cloud optical thickness and effective radius using two independent retrievals of liquid water path”, Journal of Geophysical Research - Atmospheres, in press (2009) Full Text