Interannual Cloud Feedbacks: Observations vs. Long-Term Hindcasts vs. AMIP Simulations

Global climate model predictions of future warming vary substantially due to uncertainties in how clouds respond, through their impact on Earth’s energy budget. Because cloud feedbacks on inter-annual timescales are highly correlated with those on climate change timescales, evaluating models’ cloud feedbacks on short timescales may help constrain climate sensitivity. Here, a novel cloud radiative kernel technique was used to detail the short-term cloud feedback in a single climate model and in a suite of satellite cloud observations. Whereas past studies indicated that models’ short-term tropical cloud feedbacks tend to be too positive, we found that the model closely matched observations. However, in agreement with previous work, we found that the tropical high cloud amount feedback is too large in models. Constraining meteorology using hindcast simulations improves regional aspects of the simulated feedbacks.