Those making major decisions on how to address climate change, against a backdrop of global economic instability, require reliable estimates of how the climate may change in the future. This information is mainly provided by Earth System Models (ESMs) which describe the important physical, dynamical and biogeochemical processes in the climate system. However, despite their complexity, and great advances over recent decades, they remain simplified representations of the real earth system and are associated with a number of errors that reduce the reliability of future climate projections. Therefore improving the realism of European ESMs is vital in order to provide the most reliable estimates of future climate change to decision makers allowing them to make informed policy decisions at national, European and international levels.
Highlights
Cloud cover influences the solar radiation flux received by the surface of the earth; less cloud generally leads to more solar radiation reaching the Earth's surface and conversely more cloud equals less radiation. The surface solar radiation flux is one of the key processes controlling ocean surface temperatures and thereby upper ocean stability and mixing which are key controls on ocean uptake of heat and carbon; small changes to either, or both, of these uptakes can have profound effects on the global climate. This is particularly important in the Southern Ocean as it is the region of the earth with the greatest heat and carbon uptake.
EMBRACE scientists based at SMHI in Sweden have been testing a new coupled version of the EC-Earth Earth System Model (ESM). One interesting outcome of these experiments is some seasonal (December-February) differences in cloud cover at different levels between the model estimate and real world observational estimates over the Southern Ocean. The total cloud amount is relatively well represented by the model but it is the so called ‘vertical structure’ which is less realistic; there is more upper-atmosphere cloud (a positive bias) and less mid-atmosphere cloud than expected (a negative bias) from the observational estimates. A highly simplified diagram below illustrates this; If total cloud cover is a reasonable estimate does the vertical structure matter? The SMHI team has found that the model provides a reasonable estimate of the total cloud cover but it is the vertical structure where biases (either too much or too little cloud at each level) have been found; so why is this important? It is important because different types of clouds form at different levels of the atmosphere and each cloud type transmits or traps radiation to a different extent. For example high-level cirrus clouds transmit nearly all the incoming shortwave radiation from the sun and trap and re-emit long wave radiation from the surface leading to surface warming effect. Mid-atmosphere clouds contain more water than upper-atmosphere clouds, with a larger fraction of this water in liquid, rather than frozen, form. As a result mid-atmosphere clouds transmit less solar radiation than upper-atmosphere clouds. If a model simulates more of upper-atmosphere clouds than observational estimates and too few mid-atmosphere clouds then the surface will receive too much solar radiation and be warmer than in reality. Indeed many ESMs show a positive bias in surface temperature in the Southern Ocean. How will the SMHI team work to improve the cloud representation? The first step in improving the vertical structure of the EC-Earth clouds is to fully understand which processes in the model (there may be a number) are incorrectly representing reality and are thus the underlying cause of the erroneous cloud structure seen. To do this we need to first uncouple the model ocean from the atmosphere and run the atmosphere-only version of EC Earth which is forced by observed ocean surface conditions. This helps to constrain the model to simulate a specific time period for which there are detailed observations to help in identifying shortcomings in the model. Direct cloud observations over the Southern Ocean are limited so satellite dat is relied on to establish what is going wrong with the model. We use a relatively new satellite series called A-Train which includes two satellites (CloudSAT and CALIPSO); these provide detailed observations of the vertical structure of cloud amount but also how they clouds are constituted with estimates of the amount of liquid and frozen water within a cloud. Using this satellite data, together with other observational estimates of the atmospheric structure over the Southern Ocean, we will try to identify typical and repeating situation and the vertical locations where the model clouds differ from the satellite reality. Based on these findings, backed up by our theoretical understanding of cloud physics, we will test a range of targeted modifications to the EC-Earth model. By performing specially designed simulations for the time period for which the satellite data is available (2006-present) we will be able to see whether the targeted changes made to the model result in a better match between the clouds simulated by EC-Earth and those in reality. Once an acceptable agreement is seen between the model and observed clouds the EC Earth atmosphere model can then be coupled to the ocean model once more. Then a series of long climate runs can be made to see if the cloud improvements lead to a better simulation of the surface radiation budget and upper ocean temperatures. 
News
