In this study, two machine learning (ML) models (support vector regression (SVR) and extreme gradient boosting (XGBoost)) are developed to perform seasonal forecasts of the surface air temperature (SAT) in winter (December‐January‐February, DJF) in North America (NA). The seasonal forecast skills of the two ML models are evaluated via cross validation. The forecast results from one linear regression (LR) model, and two dynamic climate models are used for comparison. In the take‐one‐out hindcast experiment, the two ML models and the LR model show reasonable seasonal forecast skills for winter SAT in NA. Compared to the two dynamic models, the two ML models and the LR model have better forecast skill for the winter SAT over central NA, which is mainly derived from a skillful forecast of the second empirical orthogonal function (EOF) mode of winter SAT over NA. In general, the SVR model and XGBoost model hindcasts show better forecast performances than the LR model. However, the LR model shows less dependence on the size of the training data set than the SVR and XGBoost models. In the real forecast experiments during the period of 2011–2017, the two ML models exhibit better forecasting skills for the winter SAT over northern and central NA than do the two dynamic models. The results of this study suggest that the ML models may provide improved forecasting skill for seasonal forecasts of the winter climate in NA.

中文翻译：

---

北美冬季地面气温季节预报的机器学习模型

在这项研究中，开发了两个机器学习（ML）模型（支持向量回归（SVR）和极端梯度增强（XGBoost））以对冬季（12月至1月至2月，DJF）执行地面气温（SAT）的季节性预测）在北美（NA）。这两个机器学习模型的季节预报技能通过交叉验证进行评估。一个线性回归（LR）模型的预测结果和两个动态气候模型用于比较。在外带后验实验中，两个ML模型和LR模型显示了北美冬季SAT的合理季节预报技巧。与两个动态模型相比，两个ML模型和LR模型对NA中部地区冬季SAT的预报技巧更高，这主要来自对北美地区冬季SAT的第二经验正交函数（EOF）模式的熟练预测。通常，SVR模型和XGBoost模型的后播比LR模型显示出更好的预测性能。但是，与SVR和XGBoost模型相比，LR模型显示出对训练数据集大小的依赖性较小。在2011-2017年期间的真实预报实验中，与两个动态模型相比，这两个ML模型对北部和中部NA的冬季SAT表现出更好的预报技巧。这项研究的结果表明，ML模型可以为北美地区冬季气候的季节预报提供改进的预报技巧。与SVR和XGBoost模型相比，LR模型显示出对训练数据集大小的依赖性更小。在2011-2017年期间的真实预报实验中，与两个动态模型相比，这两个ML模型对北部和中部NA的冬季SAT表现出更好的预报技巧。这项研究的结果表明，ML模型可以为北美地区冬季气候的季节预报提供改进的预报技巧。与SVR和XGBoost模型相比，LR模型显示出对训练数据集大小的依赖性更小。在2011-2017年期间的真实预报实验中，与两个动态模型相比，这两个ML模型对北部和中部NA的冬季SAT表现出更好的预报技巧。这项研究的结果表明，ML模型可以为北美地区冬季气候的季节预报提供改进的预报技巧。