10:45 〜 11:00
[ACG43-07] Optimal selection of sea ice motion vectors from AMSR2 89-GHz imageries in the Arctic Ocean
キーワード：海氷移動ベクトル、北極海、衛星リモートセンシング
Sea ice motion vector (SIMV) retrievals have been developed on the basis of the maximum cross-correlation (MCC) method which searches matched spatial patterns in a sequence of imageries obtained by passive microwave radiometer observations that are capable of monitoring across the Arctic/Antarctic Ocean on daily basis without severe cloud contaminations. Since a theoretical motion precision in the MCC method is decreased as an image pixel is upsized, the Advanced Microwave Scanning Radiometer 2 (AMSR2) 89-GHz channel, which offers 3- and 5-km footprint resolutions in cross and across orbit directions, has a potential to increase the precision. On the other hand, this channel also has a drawback that its higher sensitivity to atmospheric moisture compared with lower frequency channels increase spurious vectors in retrieved SIMV fields. In such an erroneous case, however, valid sea ice vectors are often detected by not the maximum (i.e., first) cross correlation peak but second or third ones which are dropped in the MCC scheme.
In this study, we present an algorithm to select optimal vectors from multiple candidates nominated by several cross-correlation peaks. The basic idea of this optimal selection is that a vector closest to a reference vector is selected from multiple candidates including false ones. The reference vector field is obtained by merging retrievals from not only 89-GHz data but also lower frequency (18- and 36-GHz) data after filtering processes of questionable motion estimates, to improve a reliability of the reference in areas where suffer atmospheric effects. In the presentation, we will demonstrate performances of this algorithm to retrieve SIMVs using AMSR2 89-GHz data in the Arctic Ocean.
In this study, we present an algorithm to select optimal vectors from multiple candidates nominated by several cross-correlation peaks. The basic idea of this optimal selection is that a vector closest to a reference vector is selected from multiple candidates including false ones. The reference vector field is obtained by merging retrievals from not only 89-GHz data but also lower frequency (18- and 36-GHz) data after filtering processes of questionable motion estimates, to improve a reliability of the reference in areas where suffer atmospheric effects. In the presentation, we will demonstrate performances of this algorithm to retrieve SIMVs using AMSR2 89-GHz data in the Arctic Ocean.