Could Machine Learning Break the Convection Parameterization Deadlock?
Could Machine Learning Break the Convection Parameterization Deadlock?
复制标题
DOI：
10.1029/2018gl078202
复制
发表时间：
2018-06-16
影响因子：
5.2
通讯作者：
Yacalis, G.
中科院分区：
文献类型：
--
作者：
Gentine, P.;Pritchard, M.;Yacalis, G.
在大尺度气候模式中表示未解决的湿对流仍然是当前气候模拟的主要瓶颈之一。当显式地解决对流时(即，在大约一公里左右的高空间分辨率的云解析模式中)，存在的许多与参数化对流有关的偏差被强烈地减小。在这里，我们提出了一种基于机器学习的对流参数化的新方法，使用具有指定海面温度的水行星作为概念验证。深度神经网络用超参数版本的气候模式训练，其中对流由数千个嵌入的二维云解析模式来解决。对流的机器学习表示，我们称为云脑(CBRAIN)，可以巧妙地预测超参数化的许多对流加热、增湿和辐射特征，这些特征对气候模拟最重要，尽管一个意外的副作用是减少一些超参数化的内在变化。由于短短三个月的高频全球训练数据足以提供这项技能，本文提出的方法为未来在自上而下建立的气候模式中的对流参数化开辟了新的可能性，即通过从异常显式的模拟中学习对流的显著特征。简单语言摘要云辐射效应和湿对流引起的大气加热和增湿的表示仍然是当代气候模式的主要挑战，导致气候预测的广泛传播。在这里，我们表明，在一个高分辨率模式上训练的神经网络可以解决湿对流问题，这是一种吸引人的技术，可以更好地描述粗分辨率气候模式中的湿对流。
Representing unresolved moist convection in coarse-scale climate models remains one of the main bottlenecks of current climate simulations. Many of the biases present with parameterized convection are strongly reduced when convection is explicitly resolved (i.e., in cloud resolving models at high spatial resolution approximately a kilometer or so). We here present a novel approach to convective parameterization based on machine learning, using an aquaplanet with prescribed sea surface temperatures as a proof of concept. A deep neural network is trained with a superparameterized version of a climate model in which convection is resolved by thousands of embedded 2-D cloud resolving models. The machine learning representation of convection, which we call the Cloud Brain (CBRAIN), can skillfully predict many of the convective heating, moistening, and radiative features of superparameterization that are most important to climate simulation, although an unintended side effect is to reduce some of the superparameterization's inherent variance. Since as few as three months' high-frequency global training data prove sufficient to provide this skill, the approach presented here opens up a new possibility for a future class of convection parameterizations in climate models that are built top-down, that is, by learning salient features of convection from unusually explicit simulations.Plain Language Summary The representation of cloud radiative effects and the atmospheric heating and moistening due to moist convection remains a major challenge in current generation climate models, leading to a large spread in climate prediction. Here we show that neural networks trained on a high-resolution model in which moist convection is resolved can be an appealing technique to tackle and better represent moist convection in coarse resolution climate models.