印度洋海表盐度预报的线性马尔可夫模型及其改进办法

海洋盐度在水循环、海洋环流、海洋生态系统、全球天气和气候变化等方面起着至关重要的作用。然而, 受观测的限制, 以往对海洋盐度的研究相对匮乏, 对其进行预报的工作更为少见。本文采用线性马尔可夫模型对印度洋海表面盐度(sea surface salinity, SSS)开展初步的预报工作。根据混合层盐度收支方程, 选择海表面高度(sea surface height, SSH)、海表面温度 (sea surface temperature, SST)、SSS等物理量的异常值作为模型的组成部分, 对印度洋SSS开展预报工作。结果表明, 马尔可夫模型可提前9个月对印度洋SSS进行较好的预报。此外, 南太平洋海表面温度异常(sea surface temperature anomaly, SSTA), 海表面高度异常(sea surface height anomaly, SSHA)和印度洋偶极子(Indian Ocean dipole, IOD)系数等遥相关因素的加入可将线性马尔可夫预报对印度洋SSS的预报效果(相关系数)平均提高10%。利用改进的模型对印度洋SSS进行提前1~11个月的“实时”预测, 得出预报的SSS时空变化特征与观测场相吻合。综上所述, 改进的线性马尔可夫模型对印度洋SSS具有一定的预测能力, 未来可进一步完善。

A linear Markov model for the forecast of sea surface salinity in the Indian Ocean and its improvement method

Marine salinity plays a crucial role in the change of water circulation, ocean circulation, marine ecosystems, global weather and climate. However, restricted by observation, previous studies on marine salinity are relatively scarce, and the prediction involving marine salinity is even rarer. This study used the linear Markov model to forecast the Indian Ocean sea surface salinity (SSS). Based on the mixed layer salinity budget equation, the SSH (sea surface height), SST (sea surface temperature), and SSS were selected as the components of the model to forecast the Indian Ocean SSS. The Markov model makes a great prediction of the Indian Ocean SSS for nine months in advance. In addition, considering teleconnection, the addition of SSHA (sea surface height anomaly), SSTA (sea surface temperature anomaly) in the South Pacific and the Indian Ocean dipole (IOD) coefficient can improve the prediction skill by an average of 10% (correlation coefficient). Real-time prediction of the Indian Ocean SSS for 1-11 months using improved models shows that the predictions are roughly consistent with the observations. As mentioned above, the improved linear Markov model has a certain predictive skill for SSS in the Indian Ocean, and it can be further improved in the future.