Convolutional neural networks (CNNs) have become one of the state-of-the-art techniques for downscaling climate projections. They are being applied under Perfect-Prognosis (trained in a historical period with observations) and hybrid approaches (as Regional Climate Models (RCMs) emulators), with satisfactory results. Nevertheless, two important aspects have not been, to our knowledge, properly assessed yet: (1) their performance as emulators for other Earth System Models (ESMs) different to the one used for training, and (2) their performance under extrapolation, that is, when applied outside of their calibration range. In this study, we use UNET, a popular CNN, to assess these two aspects through two pseudo-reality experiments, and we compare it with simpler emulators: an interpolation and a linear regression. The RCA4 regional model, with 0.11° resolution over a complex domain centered in the Pyrenees, and driven by the CNRM-CM5 global model is used to train the emulators. Two frameworks are followed for the training: predictors are taken (1) from the upscaled RCM and (2) from the ESM. In both frameworks, the performance of the UNET when applied for other ESMs different to the one used for training is considerably worse, indicating poor generalization. For the linear method a similar deterioration is seen, so this limitation does not seem method specific but inherent to the task. For the second experiment, the emulators are trained in present and evaluated in future, under extrapolation. While averaged aspects such as the mean values are well simulated in future, significant biases (up to 5°C) appear when assessing warm extremes. These biases are larger by UNET than those produced by the linear method. This limitation suggests that, for variables such as temperature, with a marked signal of change and a strong linear relationship with predictors, simple linear methods might be more appropriate than the sophisticated deep learning techniques.

中文翻译：

---

关于深度学习对气候变化预测统计降尺度的局限性：可转移性和外推问题

卷积神经网络（CNN）已成为缩小气候预测规模的最先进技术之一。它们正在完美预测（在历史时期进行观测训练）和混合方法（作为区域气候模型（RCM）模拟器）下应用，并取得了令人满意的结果。然而，据我们所知，两个重要方面尚未得到适当评估：（1）它们作为其他地球系统模型（ESM）模拟器的性能与用于训练的模型不同，以及（2）它们在外推下的性能，即即，当应用在其校准范围之外时。在本研究中，我们使用流行的 CNN UNET 通过两个伪现实实验来评估这两个方面，并将其与更简单的模拟器进行比较：插值和线性回归。 RCA4 区域模型在以比利牛斯山脉为中心的复杂域上具有 0.11° 分辨率，并由 CNRM-CM5 全局模型驱动，用于训练模拟器。训练遵循两个框架：（1）从升级的 RCM 中获取预测变量，（2）从 ESM 中获取预测变量。在这两种框架中，当 UNET 应用于与训练所用的不同的其他 ESM 时，其性能要差得多，这表明泛化能力较差。对于线性方法，可以看到类似的恶化，因此这种限制似乎不是特定于方法的，而是任务固有的。对于第二个实验，模拟器在当前进行训练并在未来进行外推评估。虽然未来可以很好地模拟平均值等平均方面，但在评估极端温暖情况时会出现显着偏差（高达 5°C）。 UNET 产生的这些偏差比线性方法产生的偏差更大。这一限制表明，对于温度等具有明显变化信号且与预测变量具有很强线性关系的变量，简单的线性方法可能比复杂的深度学习技术更合适。