Aerosol, surface and cloud retrieval using passive remote sensing over the Arctic

The lack knowledge of aerosol optical properties is one of the sources of uncertainty in assessment and projections of the evolution of climate change and the phenomenon of Arctic Amplification. The spatial and temporal change of microphysical, chemical and optical properties of aerosols in the Arctic and the induced effects through direct and indirect radiative forcing of aerosols remain an open question. The cause of this gap in our understanding and therefore in the global aerosol optical thickness (AOT) maps is associated with the difficulty of retrieving aerosol properties over bright surfaces covered with snow and ice. Decoupling a strong surface signal from that of aerosol in the measured top-of-atmosphere reflectance is challenging and still hampered due to remaining unresolved issues in state-of-the-art algorithms. Despite the promising performance of previously-developed methods and ongoing research, there is no published long-term AOT product over polar regions (over land and ocean) to be used for climate studies. In this work, to extend our knowledge about the open issues and improve the existing algorithms, first we focus on the two major obstacles in the retrieval of AOT over snow/ice surfaces: i) cloud identification, and ii) surface properties; Second, we apply the outcome of studying the two mentioned prerequisites to improve the previously-developed aerosol retrieval algorithm called AEROSNOW and create a long-term data record for aerosol optical thickness over the Arctic circle. In the framework of this work, a new cloud identification algorithm called the AATSR/SLSTR Cloud Identification Algorithm (ASCIA) has been developed to screen cloudy scenes in observations of Advanced Along-Track Scanning Radiometer (AATSR) on-board ENVISAT as well as its successor Sea and Land Surface Temperature Radiometer (SLSTR) on-board Sentinel-3. The cloud detection results are verified by comparing them with available cloud products over the Arctic. Furthermore, the cloud product from ASCIA is validated using the ground-based measurements SYNOP, resulting in a promising agreement. In general, ASCIA shows an improved performance in comparison with other algorithms applied to AATSR measurements over snow/ice. For the study of snow surface properties, the reflectance is simulated in a snow–atmosphere system, using the SCIATRAN radiative transfer model, and the results are compared with those of airborne measurements. A sensitivity study is conducted to highlight the importance of having a priori knowledge about snow morphology (size and shape) and atmospheric parameters to minimise the difference between simulated and real world reflectance. The absolute difference between the modelled results and measurements in off-glint regions is generally small and promising. In the final step, we apply the outcome of previous steps in the AEROSNOW algorithm as far as possible within the scope of this work and retrieve AOT over the Arctic circle for the 2002-2012 period with the spatial resolution of 1 km2. The retrieved AOT is validated using ground-based measurements AErosol RObotic NETwork (AERONET). The results of validation are promising and show the successful performance of the algorithm especially during haze episodes. However, in some cases large differences exist between the retrieved AOT and AERONET measurements for which more statistical and physical analysis are necessary to better understand the cause. Nevertheless, the long-term data record and validation produced hold significant value as are the first attempt to better understand the role of aerosols in the Arctic Amplification over land and ocean on the full Arctic scale.