Investigation of smoke-cloud mixed scenes with A-Train multi-sensor data during the boreal wild fires in summer of 2007

We will present the results of the analysis of smoke-cloud mixed scenes with the goal to establish empirical evidence for changes in cloud properties caused by aerosol originating from boreal wild fires. We analyzed near-instantaneous, spatially collocated data from the NASA A-Train satellite constellation, including aerosol and cloud products from MODIS, CALIPSO lidar, and CloudSat. Smoke plume altitudes were constrained with the CALIPSO lidar, while MODIS aerosol products were used to determine aerosol optical depth and verify the spatial extent of the smoke plumes relative to the clouds. In our study, we examined aerosol optical depth, effective droplet radius (Re), liquid water/ice path (LWP, IWP), and water content of smoke-polluted clouds. We analyzed in detail the four cases of mesoscale cloud systems that were most likely affected by smoke plumes originating from the boreal wild fires that occurred in Canada and Alaska in the summer of 2007.

Our preliminary results show complex behavior and significant variations of Re and LWP, IWP values in analyzed data. We found significant differences between MODIS- and CloudSat- retrieved cloud products and between their behavior for both polluted and clean clouds. For both MODIS and CloudSat, more coherent (while still different) Re vs. LWP dependencies were found when data were stratified by cloud types and/or by cloud phase.

In MODIS retrievals, we observed a narrower distribution and a shift of the Re vs. LWP dependencies toward larger sizes (17…35 µm) for clean deep convective clouds, while for polluted deep convective clouds the distributions were broader, with many more Re values within 6…15 µm. Similarly in CloudSat retrievals, we observed larger Re values (up to 2 µm difference) for fixed LWP in clean altostratus clouds than in polluted ones, which is in agreement with the classical Twomey hypothesis. At the same time, opposite tendencies or the absence of obvious Re vs. LWP relationships were observed for the other cases (e.g. altocumulus clouds in CloudSat retrievals). A maximum of the Re distribution for the whole data set retrieved by MODIS was found to be at 8 µm for polluted clouds and close to 10 µm for clean clouds. The differences between “polluted” and “clean” distributions were much less pronounced in the CloudSat data.