Insights from a refined decomposition of cloud feedbacks

Decomposing cloud feedback into components due to changes in several gross cloud properties provides valuable insights into its physical causes. Here we present a refined decomposition that separately considers changes in free tropospheric and low cloud properties, better connecting feedbacks to individual governing processes and avoiding ambiguities present in a commonly used decomposition. It reveals that three net cloud feedback components are robustly nonzero: positive feedbacks from increasing free tropospheric cloud altitude and decreasing low cloud cover and a negative feedback from increasing low cloud optical depth. Low cloud amount feedback is the dominant contributor to spread in net cloud feedback but its anticorrelation with other components damps overall spread. The ensemble mean free tropospheric cloud altitude feedback is roughly 60% as large as the standard cloud altitude feedback because it avoids aliasing in low cloud reductions. Implications for the “null hypothesis” climate sensitivity from well‐understood and robustly simulated feedbacks are discussed. Key Points Three robust cloud feedbacks in GCMs: low amount, low optical depth, and non‐low altitude Positive feedback from rising non‐low clouds is smaller and better constrained than commonly quoted Low cloud amount feedback is single largest contributor to spread in net cloud feedback