Do Emergent Constraints Help Predict Future Climate?
Scientists at Lawrence Livermore National Laboratory and the University of California at Los Angeles reviewed a number of recently found empirical relationships between aspects of the current and future climate found in climate model simulations. While most of these so-called “emergent constraints” lack a physical basis and thus shouldn’t be trusted, there are some credible constraints which predict a more sensitive climate and identify where model improvements would lead to reduced spread in climate projections.
Several studies have found that climate models that better simulate present-day clouds suggest that the climate is more sensitive to greenhouse gas warming. The authors reviewed these studies and find that there is a physical basis to some of the empirical relationships, suggesting that there is good reason to believe climate sensitivity is in the upper half of model estimates.
By examining the diversity of climate model simulations, scientists have recently uncovered a number of empirical relationships between observable aspects of the current climate and aspects of future climate that we want to know. These empirical relationships, called “emergent constraints”, are sometimes used to predict the future when observations can tell us which models correctly simulate the current climate. Scientists at Lawrence Livermore National Laboratory and the University of California at Los Angeles reviewed a number of these constraints to determine their trustworthiness. They concluded that unless a strong and identifiable physical basis can be found for the relationship, that the relationship should not be trusted. The relationships reviewed involve cloud feedbacks, namely how clouds will respond to climate warming, which is the leading source of uncertainty in model-predictions of climate sensitivity. The more credible constraints involve the behavior of low-level clouds, a leading source of diversity in cloud feedbacks. Because emergent constraints identify a source of model error that projects onto climate predictions, they deserve extra attention from those developing climate models and climate observations. While a systematic bias cannot be ruled out, it is noteworthy that the most promising emergent constraints suggest larger cloud feedback and hence climate sensitivity.