Retrieval of aerosol optical thickness in Arctic region using dual-view AATSR observations
Larysa Istomina(1), Wolfgang von Hoyningen-Huene(1), Alexander Kokhanovsky(1), Mathias Schreier(1), Vladimir Rozanov(1) and John Burrows(1)
(1) University of Bremen, O. Hahn Allee 1, D-28334 Bremen, Germany
Aerosols affect climate in two ways by direct and indirect climate forcing. In Arctic aerosols can also affect the albedo of the surface by deposition on snow and ice. These effects result in changing of atmospheric circulations and melting of snow and ice coverage, and this has a lot of consequences on both local and global scales. To understand the scale of these consequences one should know the amount of aerosols and details of their transport in Arctic. Remote sensing can provide necessary data, but aerosol retrival over very bright surface is a difficult task, therefore new development of retrievals are necessary.
In this work a dual-view algorithm to retrieve the amount of aerosol in Artic region has been established and validated. The algorithm consists of two main steps: 1. cloud/snow/ocean determination 2. AOT retrieval over snow, ice or open water. The first step is performed using spectral analysis of data pixels in 7 wavelengths from VIS to TIR, using the difference in spectral behaviour of snow, clouds and ocean in these channels. The second step solves the equation for top of atmosphere radiance in two different ways, depending on the type of the surface. For ocean the retrieval is made by interpolation of LUTs, precalculated with forward RT model. For snow and ice the advantage of AATSR observations, the dual-viewing geometry (nadir view 0° and forward view 55°), is used. The albedo of snow is highly variable, it depends on grain size of snow, soot concentration, humidity, age and can change in time and in space. Using the two views simulatiously makes it possible to exclude the snow albedo from the retrieval, and only account for snow bidirectional reflection function (BRDF) shape. In order to take the most realistic shape of snow BDRF, we use measured radiances of the whole system „surface + atmosphere“ to weight a preassumed snow model and bring it closer to reality. This also lets us diminish the contamination of the retrieval by the relief of the surface.
The retrieval has been applied to sequences of data for various Arctic haze events at some Artic sites such as Spitsbergen and Alaska. The results of AOT retrieval over snow and over ocean show good correspondence to each other. The retrieved AOT has also been compared to ground-based measurements. The developed cloudscreening method has been validated against Micro-Pulse Lidar Network data.