图2 各同化方法的时空平均RMSE随集合数目的变化。重建状态量是(a)地表温度(b)降水(c)表层土壤湿度(d)深层土壤湿度。黑色水平线表示使用集合数目100的OEnKF进行100次蒙特卡洛实现的RMSE,AOEnKF-B与OEnKF-B表示使用静态B更新 “类比”与静态先验集合。
(1) Sun, H., L. Lei, Z. Liu, L. Ning, and Z. Tan, 2023: A Hybrid Gain Analog Offline EnKF for Paleoclimate Data Assimilation. Journal of Advances in Modeling Earth Systems 16, e2022MS003414. https://doi.org/10.1029/2022MS003414
(2) Sun, H., L. Lei, Z. Liu, L. Ning, and Z. Tan, 2022: An Analog Offline EnKF for Paleoclimate Data Assimilation. Journal of Advances in Modeling Earth Systems, 14, e2021MS002674. https://doi.org/10.1029/2021MS002674
进展:基于卷积神经网络的集合Kalman滤波适应性局地化
编辑:侯梦瑶