Constraining Tropical Low Cloud Feedbacks Using Observations of the Fast Cloud Response

热带和亚热带地区低云的特性因全球变暖而发生的变化，已被确定为气候模式对全球变暖预测存在很大不确定性的一个关键原因。 在某些模式中，变暖会产生更普遍或反射性更强的低云，从而将更多的阳光反射回太空，并部分缓解温室气体的变暖效应（对强加的变暖的负云反馈），而在其他模式中，云变得不那么反射或不那么普遍，地表的阳光增强会加剧全球变暖（正云反馈）。 本研究试图结合卫星观测、再分析产品和气候模式输出，对真实的低云反馈进行观测约束估计。 该方法首先使用神经网络建立经验关系低云反射率和其他云的属性（从卫星观测确定）和大尺度变量，包括海面温度和大气温度和水蒸气。这些关系是利用云对其环境的快速（几小时到一天的时间尺度）响应建立的。 一旦建立了这些关系，云反馈将通过应用由神经网络代表的经验关系，结合气候模型中大尺度变量的变化来估计，这些变化是模拟全球变暖的结果。 除了对全球变暖的低云反馈的估计外，该方法还将为这些估计提供不确定性界限，并可用于诊断低云特性的模式参数化中的错误，由于热带低云反馈在产生气候变化模式预测中的大不确定性方面的关键作用，这项工作具有更广泛的影响。 对低云反馈进行可靠的、基于观测的估计有助于减少这种不确定性，并为决策者提供有关温室气体引起的全球变暖的可能程度和物理后果的更好信息。

Changes in the properties of low clouds in the tropics and subtropics in response to global warming have been identified as a key cause of the large uncertainty in global warming projections from climate models. In some models warming produces more prevalent or more reflective low clouds, thus reflecting more sunlight back to space and partially mitigating the warming effects of greenhouse gases (a negative cloud feedback to the imposed warming), while in other models the clouds become less reflective or less prevalent and enhanced sunlight at the surface exacerbates global warming (a positive cloud feedback). This research attempts to produce observationally-constrained estimates of the true low cloud feedback using a combination of satellite observations, reanalysis products, and climate model outputs. The method first uses a neural network to establish empirical relationships between low cloud reflectivity and other cloud properties (determined from satellite observations) and large-scale variables including sea surface temperature and atmospheric temperature and water vapor. These relationships are established using the fast (timescales of hours to a day) response of clouds to their environment. Once these relationships are established, cloud feedback will estimated by applying the empirical relationships represented by the neural network in combination with changes in the large-scale variables in the climate models which occur as a result of simulated global warming. In addition to estimates of the low cloud feedback to global warming, the method will provide uncertainty bounds for those estimates, and can be used to diagnose errors in model parameterizations of low cloud properties.The work has broader impacts due to the key role of tropical low cloud feedback in generating the large uncertainty in model projections of climate change. A reliable, observationally-based estimate of the low cloud feedback could help to reduce this uncertainty and provide better information to decision makers regarding the likely extent and physical consequences of greenhouse-gas induced global warming.