17:15 〜 18:45
[ACG32-P02] ニューラルネットワークを用いた日本周辺における冬季の気温偏差の駆動機構に関する研究
キーワード：冬季気温、テレコネクション、深層学習、インド洋ダイポールモード現象
A neural network model that predicts the sign of wintertime temperature anomalies around Japan in from global sea surface temperature (SST) and 500 hPa geopotential height (Z500) anomalies in the preceding season is first constructed. Its optimal inputs reveal the most plausible patterns preceding anomalously warm/cold events around Japan. Prior to early winter warm events, a negative Western Pacific (WP)-like pattern is prominent, although the centers are shifted to the northwest of the conventional WP pattern. Also, positive SST anomalies around Japan are gradually enhanced during fall, while a positive Indian Ocean Dipole (IOD)-like pattern with the western pole more confined to the Arabian Sea appears over the Indian Ocean. In the equatorial Pacific, a weak positive El Niño/Southern Oscillation (ENSO)-like pattern is observed, but the SST anomalies are biased toward the north. Prior to the late winter warm events, the patterns have larger anomalies and different characteristics. In the anomalous Z500 fields, a negative WP-like pattern is dominant and combined with a positive Arctic Oscillation-like pattern. In addition, a positive Pacific-North American (PNA)-like pattern is seen in December and January. An increasing trend of the SST anomalies around Japan is similar to the early winter patterns, but is extended further eastward. On the other hand, the IOD-like and ENSO-like patterns are not found.
Since the influences of the IOD-like pattern on wintertime temperature around Japan, which are statistically identified by neural networks, have not been systematically examined, the dynamical mechanisms are examined in detail. During the positive IOD, anomalous divergent winds from the Arabian Sea generate a Rossby Wave source (RWS) over the Arabian Peninsula in the upper troposphere. In addition, an anomalous meridional dipole circulation between the East Asia and the Philippine Sea generates a RWS over the East Asia. A wave activity flux analysis shows that the Rossby waves propagate from the Arabian Peninsula, intensify over the East Asia, and reach the North Pacific during the positive IOD. Accordingly, geopotential height anomalies at 200 hPa during the positive IOD show a negative extreme over southern China and a positive extreme over the northwestern Pacific. The northern edge of the positive center in the northwestern Pacific strengthens westerly winds over the Sea of Okhotsk in the upper troposphere during the positive IOD, providing an unfavorable condition for cold air intrusion into Japan. Furthermore, weakened seasonal northwesterly winds lead to warm winters around Japan with an anomalous cyclone in the North Pacific at the lower level.
Since the influences of the IOD-like pattern on wintertime temperature around Japan, which are statistically identified by neural networks, have not been systematically examined, the dynamical mechanisms are examined in detail. During the positive IOD, anomalous divergent winds from the Arabian Sea generate a Rossby Wave source (RWS) over the Arabian Peninsula in the upper troposphere. In addition, an anomalous meridional dipole circulation between the East Asia and the Philippine Sea generates a RWS over the East Asia. A wave activity flux analysis shows that the Rossby waves propagate from the Arabian Peninsula, intensify over the East Asia, and reach the North Pacific during the positive IOD. Accordingly, geopotential height anomalies at 200 hPa during the positive IOD show a negative extreme over southern China and a positive extreme over the northwestern Pacific. The northern edge of the positive center in the northwestern Pacific strengthens westerly winds over the Sea of Okhotsk in the upper troposphere during the positive IOD, providing an unfavorable condition for cold air intrusion into Japan. Furthermore, weakened seasonal northwesterly winds lead to warm winters around Japan with an anomalous cyclone in the North Pacific at the lower level.