Arctic Cloud Surface Response Experiment

Arctic sea ice conditions are changing rapidly. The summer of 2012 saw the late summer minimum extent record of 2007 shattered, and every year since 2007 has seen a minimum below any point in the satellite record prior to 2007. The large area of melt results in a greater fraction of first year ice the following winter; this is more easily fragmented during the following melt season. Climate models have failed to predict, of for the most part reproduce, the extent of recent minima. The reasons for the rapid decline in Arctic sea ice are not well understood, but are believed to relate to several strong feedback processes in the Arctic. Of interest here is a cloud feedback.

At low and mid-latitudes low level clouds act to reduce warming at the surface by reflecting solar radiation; however, the combination of low solar radiation and a highly reflective surface (sea ice and snow) mean that the impact of low cloud on the solar radiation budget is small. The impact on the infra red (longwave) radiative budget is more significant since the clouds radiate at almost the same temperature as the surface (much warmer than a clear sky) and greatly reduce longwave surface cooling. The net effect is that, except for a short period at the height of summer, low level Arctic clouds act to warm the surface. Satellite retrievals of cloud cover show that in the spring and autumn cloud cover increases as sea ice fraction decreases. As the melt season starts earlier and ends later with increased warming, the accompanying increase in spring and autumn cloud cover acts as a positive feedback on warming.

Clouds are one of the largest sources of uncertainty within climate models; particularly so in the Arctic where they are the single largest factor controlling the surface energy balance. Clouds are sub-gridscale processes in climate models, so must be parameterised in terms of resolved variables. Such parameterizations must ultimately be based upon measurements. The Arctic is remote and a harsh environment in which to work, consequently measurements are very sparse and cloud parameterisations are largely derived from - and models tested against - measurements at lower latitudes where conditions differ substantially from those in the Arctic. Recent evaluations of a variety of models show that their representation of Arctic clouds suffer biases in prevalence, vertical extent, water content, microphysical and radiative properties. This introduces significant biases into the modelled surface energy budget, and hence modelled ice melt/growth.

Improvements in cloud representation for the Arctic depend upon in situ measurements. This project will make measurements of cloud properties, boundary layer structure (to which the clouds are closely coupled), surface exchanges of heat and moisture, and the sea ice conditions. Although it is clear that mean cloud cover is correlated with ice fraction, it is not known to what extent the properties of the cloud depend upon ice fraction. Our observations will span a range of ice conditions from dense pack ice, through the marginal ice zone into open water. The 3-month duration of the measurement programme, through the peak of the melt season and into the early autumn freeze up, ensures that we will obtain a large volume of data over a wide range of conditions, allowing statistical relationships between cloud and sea ice fraction to be determined. Measurements will include Doppler cloud radar (cloud properties, dynamics, and turbulence), scanning and vertically pointing microwave radiometers (temperature and humidity profiles, cloud water content), Doppler lidar (wind profiles and turbulence, cloud particle shape), surface turbulent fluxes of heat, moisture and momentum, downwelling long and shortwave radiation, ice fraction, and surface temperature. We will determine how cloud properties vary with surface conditions, and how they relate to the boundary layer turbulent structure which links them to the surface.

This study has potentially wide impacts throughout the climate modelling community, both within academia and government agencies. Our ongoing collaboration with the Met Office provides a direct route to ensure that all our results will be available to the Met Office and Hadley Centre.

Improvements to parameterizations schemes for boundary layer turbulent processes, low level cloud representation, and cloud radiative properties for large scale models will:
- Improve fidelity of climate predictions - this is essential if an accurate assessment of future climate change is to be achieved. This need is particularly pressing for the Arctic regions due to the rapid rate of observed change, and the expected continuation (perhaps acceleration) of this change, but also impacts on predicted climate for the rest of the world.
- Improved performance of numerical weather prediction for mid-to-high latitude regions (including the UK). There is increasing evidence that changes in Arctic climate have impacts on mid-latitude weather via changes in large scale pressure fields and the motion of weather systems. Improved representation of Arctic boundary layer and cloud processes will lead to improvements in prediction of ice cover, and to both seasonal forecasts and climate projections.
Improvements to predictive capabilities in the Arctic have impacts for climate prediction around the world. Reducing uncertainty in climate prediction is essential if policy makers are to be able to implement effective plans for limiting the human impacts of climate change.