Multiyear Ice Retrieval Using Passive Microwave Remote Sensing Radiometer AMSR-E 89GHz Data

Sea ice is an important component of the earth climate system. The albedo of sea ice is much higher than that of open water, so most of the incoming solar radiation is reflected from the sea ice surface. Also sea ice acts as a membrane between ocean and atmosphere which decreases the heat and momentum exchange. The variation of sea ice will strongly influence the atmosphere in the polar regions which will influence the global climate change. This is the reason for the recent increased public interest in the sea ice studies. There are three ways of studying sea ice: in-situ measurements, satellite remote sensing and models. Satellite remote sensing is the only means to daily observe the complete earth. Among the satellite sensors, microwave radiometers have the advantages to be independent of the day/night condition and not to be hampered by clouds. Moreover, passive microwave observations of the polar regions are available since over 30 years.In this thesis, the microwave radiometer AMSR-E 89 GHz brightness temperatures which have a high resolution of 4×6km is used to retrieve the total and multiyear ice in the Arctic. Most other sea ice retrieval algorithms use the lower frequency channels of AMSR-E which has less than half the resolution. Retrieval of total ice using AMSR-E 89 GHz data is well established, but little work has been done in multiyear ice retrieval. First, AMSR-E 89 GHz brightness temperature data from three test regions with known surface types is analyzed: first-year ice, multiyear ice and open water. According to their different brightness temperatures and polarization differences, a new algorithm is developed and daily maps and time series of the total and multiyear ice areas in the Arctic are calculated for the whole year 2007. The maximum of sea ice area is about 14 million km2, while the minimum is 4 million km2, which is consistent with other studies. The results are compared with those calculated from the NASA TEAM algorithm (total and multiyear ice), ASI algorithm (total ice) and Lomax algorithm (multiyear ice). The total ice results from the three algorithms agree quite well, the ASI algorithm gives highest ice areas though the whole year. On average, over the year 2007, the ASI algorithm produces 1.52% more ice concentration and 0.58 million km2 more ice area than the new algorithm, while the difference between the results from the new algorithm and the NASA TEAM algorithm is 8.78% in ice concentration and 0.25 million km2 in ice area. In addition, the difference between the new algorithm and the NASA TEAM algorithm is higher in summer, but quite low in winter. A possible reason would be that NASA TEAM algorithm uses lower frequency channels while the new algorithm uses high frequency channels which are more sensitive to atmospheric influence. For multiyear ice, these algorithms give quite different results. The Lomax algorithm produces highest values while NASA TEAM gives the lowest, and the results of the new algorithm are in between. Statistically, the new algorithm gives 28.27% more ice concentration and 1.38 million km2 ice area than the NASA TEAM algorithm, while 0.09% less ice concentration and 1.10 million km2 ice area than Lomax algorithm. Considering that the NASA TEAM algorithm has been developed over many years and is best validated among the considered algorithms, we take it as a reference and say the new algorithm is closer to the reference than the Lomax algorithm. The comparison of daily maps reveals that the higher multiyear ice concentrations of the new algorithm can't be attributed to specific regions, but are rather uniformly distributed over the whole Arctic. A common problem in both the new and Lomax algorithm is that the multiyear ice area greatly increases during the winter season. This is unphysical; the NASA TEAM algorithm results increase also, but much less. One possible explanation for the increase of the multiyear ice area of the new and Lomax algorithm is the variation of brightness temperature in certain winter months, which is related to the surface temperature variations. Over all, the new algorithm performs quite well in total ice retrieval, while multiyear ice retrieval should be improved, e.g. by transforming the brightness temperatures to quantities less sensitive to the physical temperature, and by combining the retrieval results with ground data.