Presentations featuring:

Mark Zelinka

Atmospheric Sciences, University of Washington

Lis Cohen

Meteorology, University of Utah

Neil Gordon

Scripps Institution of Oceanography

Jasper Kok

Applied Physics, University of Michigan

    Clouds, aerosols, and water vapor assert their influence on the climate system by way of the radiation budget: All three entities regulate the amount of solar radiation that is absorbed at the surface and in the atmosphere and how much is reflected back out to space, as well as how much outgoing longwave radiation (OLR) is emitted to space.  In addition, they interact with each other, resulting in compounding effects: clouds form under supersaturated conditions, provided that there are aerosols onto which water vapor molecules can condense.  Given a cloud with a fixed liquid water content, the addition of aerosols results in smaller, more numerous cloud droplets, which has the effect of increasing cloud albedo and inhibiting precipitation, resulting in more reflective, longer-lasting clouds.  These indirect aerosol effects represent a strong but highly uncertain negative radiative forcing on the climate system that rivals the direct forcing due to reflecting aerosols. 

    Whereas the largest sources of the global aerosol load are natural (oceans, deserts, volcanoes), anthropogenic sources (burning of biomass and fossil fuel) are substantial, especially regionally.  Jasper Kok studied the mechanism by which dust particles are emitted from the surface to the atmosphere in a process known as saltation.  He showed that assumptions in the conventional theory of saltation, in which upward dust flux scales with the cube of the wind speed, are inconsistent with measurements.  With an improved physical-based model of saltation, he found much better agreement with observations.  Thus, climate model parameterizations of the dust emission process using the convectional method must be updated to avoid over-prediction of dust emissions.

    Cloud feedbacks, especially the changes in fractional coverage of low clouds, represent the most uncertain feedback in the climate system.  An important concept is that of cloud radiative forcing (CRF), which compares the net (downward minus upward) flux at the top of the atmosphere in clear-sky conditions to that in cloudy conditions.  The forcing can be separated into longwave and shortwave components, with net CRF being the sum of these components.  Longwave CRF tends to be positive because clouds are cooler than the surface and so emission to space is reduced and more heat is retained in the system.  Shortwave CRF tends to be negative because clouds have higher albedo than the surface, so more sunlight is reflected out to space.  Models that simulate an increase in low cloud amount tend to be less sensitive to a doubling of CO2 than those that simulate a decrease in low cloud amount.  This is because low clouds are highly ubiquitous and are characterized by strongly negative net CRF due to their high albedo and warm temperature.  Conversely, high thin cirrus clouds which are ubiquitous in the convectively active regions of the planet, tend to have a positive net CRF because of their low reflectivity and very cold temperature. 

    Neil Gordon attempted to diagnose cloud feedbacks using the response of clouds to short-term variability.  Employing a clustering algorithm on the International Satellite Cloud Climatology Project (ISCCP) dataset, he found 7 distinct cloud types (which are found in similar dynamical regimes) and compared their cloud radiative forcing due to temperature changes alone.  He calculated a weak negative cloud feedback due to increasing temperature because increases in both cloud-top emission and cloud albedo dominate over reductions in cloud coverage.

    Lis Cohen demonstrated using a tracking algorithm and aircraft measurements that the properties of cirrus clouds detrained from deep convection are different in long-lived cirrus clouds compared to those that sublimate more rapidly.  In cirrus clouds that decayed rapidly, a broader ice crystal size distribution and a higher total ice water content was observed compared to the long-lived cirrus clouds, which had a narrow peak in the distribution at small crystal sizes.  Future work will investigate what causes these differences in crystal size and what processes maintain cirrus clouds after they detrain from deep convection.

    The processes by which cirrus clouds are maintained following deep convection is likely related to those which maintain an anomalously moist upper troposphere following deep convection.  Using a compositing technique centered on regions of tropical deep convection, Mark Zelinka explored the evolution of high clouds, water vapor, vertical motion, and clear sky OLR.  While convective core cloud fraction, rain rate, and upper tropospheric vertical motion peak in phase, anvil cloud fraction peaks 3 hours afterward, and upper tropospheric humidity is most moist between 9 and 12 hours afterward.  Thin cirrus clouds are ubiquitous in the convective region throughout the period.  Radiative transfer calculations show that the clear sky OLR is most sensitive to relative humidity changes in the upper troposphere; indeed clear sky OLR is observed to remain anomalously low for several hours following deep convection, consistent with a sustained upper tropospheric moist anomaly over that period.

    The community remains uncertain about the role of high clouds in the humidification of the upper troposphere.  It is unclear whether sublimation of cirrus ice crystals provide a source of moisture to the atmosphere, if cirrus clouds spread away from deep convection by "feeding off" upper tropospheric moisture, or if radiative heating in thin cirrus clouds results in upward motion that maintains the moist anomaly following convection.  Recent studies suggest that the free tropospheric humidity distribution can be explained simply by tracking air motions and assuming that the relative humidity never exceeds 100%.  If this is the case, we can be more confident in modeled changes in humidity, and thus in estimates of water vapor feedback because air motions are considerably easier to model than cloud microphysical processes.  Nonetheless, physical knowledge of cloud processes is a necessary requirement for progress to be made in modeling clouds and therefore in our ability to reduce uncertainty in climate change simulations.