Cloud Changes in Climate Models: Response to Solar and CO2 Forcing and the Relationship between Model Bias and Feedbacks

Abstract

How clouds respond to climatic forcing agents (such as CO2 and other heat-trapping gases or aerosols) creates feedbacks which are currently the greatest source of spread in estimates of future warming. The cloud response to climate change can be decomposed into two components; there are the ways that clouds change in response to global mean surface temperature change, called temperature mediated changes, and there are the ways that clouds adjust to forcing agents, and is nominally independent of global surface temperature change. Clouds can impact the top-of-atmosphere radiative anomaly created by a forcing agent, such that the temperature mediated cloud changes feedback on the temperature change (called cloud feedbacks), and the cloud adjustments can contribute to or somewhat counter-act the original forcing. The first portion of this dissertation is on temperature mediated cloud changes (and the associated radiative feedbacks) from model simulations of warming and cooling induced by solar and CO2 forcing. This research shows that temperature mediated cloud changes are quite similar from solar and CO2 forcing, but there are considerable differences when comparing cloud changes from warming and cooling. This is especially notable in the pattern of tropical high cloud changes, and the different latitudes where cloud liquification or glaciation (which impacts cloud optical depth) occur from global warming and cooling respectively. Regarding the cloud adjustments to forcing, a new method to calculate the cloud adjustment in coupled model simulations is derived, followed by an analysis of the cloud adjustment to solar and CO2 forcing. There are some similarities in the pattern of adjustment to both forcing mechanisms, however there are also notable differences, especially in the amount and cloud-top height of low and mid-level clouds, and in the optical depth of high clouds. The final piece of research included in this dissertation is a study on the relationships between models’ simulations of cloud in present-day climate conditions, and their simulation of cloud feedbacks with warming. The ways a model deviates from observations in simulations of the present-day climate can impact how the they predict the climate will change in the future. Radiative feedbacks associated with changes in tropical anvil clouds, marine midlatitude clouds, high-cloud altitude, and tropical marine low clouds all exhibit statistically significant correlations with spread in cloud attributes during present-day climate simulations. These relationships are used to further understand the mechanisms which mediate the inter-model spread of cloud feedbacks. For example, the strength of the tropical anvil cloud area feedback in models depends strongly on the optical depth of (and model bias in) tropical high clouds, and using such a relationship as an emergent constraint predicts a feedback parameter which is less negative than previous estimates.