Tropical cyclones (TCs) are affected significantly by the climate system and can provide feedbacks. TC activities are important for weather forecasting and climate predictions. Here, we focused on the spatial distribution of accumulated cyclone energy (ACE) and its seasonal prediction. To predict the ACE distribution over the western North Pacific (WNP) in autumn, we established a physical-empirical model. Analyzing 36 years observations (1979–2014) of ACE over the WNP reveals two physically predictable modes. The sea surface temperature (SST) in the southwest Pacific and central Pacific affect the first mode through the low-level circulations. At the same time, the SST in the Gulf of Alaska and the sea-ice concentration in the Beaufort Sea affect the first mode through the circumglobal teleconnection. The development of the eastern Pacific El-Niño and anomalous SST over the North Pacific affect the second mode through the vertical wind shear and low-level circulation. The sea-ice concentration in the Greenland Sea induce an upper-level circulation anomaly over the WNP and affect the second mode. Physically meaningful predictors were selected according to the controlling mechanisms of the two modes. The cross-validated hindcast results demonstrated that the principal components of the two modes are predicted with correlation coefficients of 0.68 and 0.63. Thus, the two modes are predictable. The pattern correlation coefficient skill of the ACE spatial pattern is 0.26, which is significant at the 99% confidence level. The temporal correlation coefficient skill reaches 0.21 over major regions influenced by TCs. To validate the real-time predictability of the model, independent tests were performed on the last three years (2015, 2016 and 2017), and the results show that the pattern correlation coefficients between the observations and the predictions are 0.39, 0.70, and 0.41, respectively.

热带气旋（TC）受气候系统的影响很大，可以提供反馈。技合活动对于天气预报和气候预测很重要。在这里，我们关注于累积气旋能量（ACE）的空间分布及其季节预测。为了预测秋季ACE在北太平洋西部（WNP）上的分布，我们建立了物理经验模型。对WNP上ACE的36年观察结果（1979-2014年）进行分析，发现了两种物理上可预测的模式。西南太平洋和中太平洋的海表温度（SST）通过低空环流影响第一种模式。同时，阿拉斯加湾的海表温度和波弗特海的海冰浓度通过全球全球遥相关影响了第一种模式。东太平洋厄尔尼诺现象的发展和北太平洋上空海温异常通过垂直风切变和低空环流影响了第二种模式。格陵兰海中的海冰浓度在WNP上引起高层环流异常并影响第二种模式。根据这两种模式的控制机制选择对身体有意义的预测因子。交叉验证的后验结果表明，预测两种模式的主成分的相关系数分别为0.68和0.63。因此，这两种模式是可预测的。ACE空间模式的模式相关系数技能为0.26，在99％的置信度下非常重要。时间相关系数技能在受TC影响的主要区域达到0.21。