Using SST as a proxy for cloud phase biases over the Southern Ocean

The 50˚–65˚ latitude band exhibits the largest hemispheric asymmetry of cloud albedo over the oceans as well as the largest negative Southern Ocean (SO) cloud albedo biases in CMIP models. In this study, we show that cloud albedo regressed against sea-surface temperatures (SSTs) highlights essential differences between the observed Northern and Southern hemisphere climatologies, and between the SO’s simulated and observed albedos. The threshold 4˚–5˚C stands out as a regime separator in both comparisons.

By linking our empirical findings with the extensive evidence that model errors are related to the unique microphysical characteristics of the SO environment, we hypothesize that cloud albedo as a function of SST may act as a predictor of the presence/absence of supercooled liquid water cloud content.

Using satellite-retrieved cloud optical thickness (COT) and cloud top temperature (CTT), we verify that a regime separation of COT as a function of CTT exists between the Northern and Southern hemispheres (for CTT< -12˚C), which becomes more noticeable under midlevel subsidence conditions (i.e., low, boundary layer clouds).

Our simple and straightforward method using macrophysical variables can be easily applied in model evaluation with an insight in microphysics performance, especially given the scarcity of archived cloud-specific variables by the participating CMIP models. For example, it is well known that models tend to produce glaciated rather than supercooled liquid water clouds, and we show that in many cases models are simulating Northern Hemisphere clouds for the SO. We also detect that some of the CMIP models produce the right climatological cloud albedo over the SO but for the wrong reasons.