To determine whether these conditions exist in nature, a new satellite remote sensing method was developed that is based on wave resonance absorption, which is sensitive to small ice crystals. The retrieval utilizes radiances from the infrared imaging radiometer and backscatter from the CALIPSO lidar. Retrieved single-layer cirrus cloud (T < 235 K) properties include the mid-cloud temperature T, ice particle number concentration Ni, effective diameter De, ice water content (IWC) and visible optical depth (OD), where cirrus cloud OD ranges from 0.3 to 3.0. Knowing Ni, conservative estimates of the cirrus fraction formed through hom can be made. This is because only hom can account for Ni exceeding ~ 250 liter-1 in regions typically having relatively low concentrations of ice nuclei. It was surprising to discover that the above conditions required for CCT do apparently occur in nature, and that the cooling potential of CCT appears substantial, as described below.
Global retrievals of cirrus cloud De and T were used to make the cirrus clouds simulated in CAM5 conform with the retrieved De, with the number- and mass-weighted ice fall speeds in CAM5 calculated from the retrieved De. This was done by developing De-T relationships for six latitude zones (0-30, 30-60 and 60-90 for both hemispheres). Within each latitude zone, seasonal De-T relationships were developed for cirrus over land and for cirrus over ocean (making 48 De-T relationships in total). Retrieved De (N) is largest (lowest) between 30S and 30N latitude; a region dominated by anvil cirrus (a type of “liquid origin cirrus”) where pre-existing ice strongly favors heterogeneous ice nucleation (henceforth het). Therefore the De-T relations for this region are considered representative for cirrus formed via het. Outside this region, retrieved De (N) tended to be considerably smaller (higher), presumably due to a combination of hom and het.
Two 10-year CAM5 simulations were performed; one where cirrus cloud De is based on the CALIPSO retrievals (henceforth CALIPSO), and one where the De-T relationship for het cirrus is applied globally (henceforth Het). Differences in net cloud radiative forcing (i.e. net CRF) between simulations are believed due to differences in cirrus formation mechanism; hom vs. het. Zonal mean differences outside the ± 30 °latitude zone are typically ~ 1.3 W m-2 in the mid-to-high latitudes in the N. Hemisphere excepting summer, and are typically ~ 1 to 3 W m-2 in the mid-to-high latitudes of the S. Hemisphere excepting summer. These are estimates of CCT’s cooling potential, assuming all hom cirrus convert to het cirrus, based on the CALIPSO retrievals.
Regarding the figure, zonal mean differences in net CRF between simulations (CALIPSO minus Het) are shown for each season. The figure shows the warming effect of observed cirrus clouds relative to cirrus clouds formed via het. If the sign of these curves were reversed, they would show the potential cooling effect of CCT. The net CRF at high latitudes depends strongly on solar zenith angle (or lack of sunshine during winter), with longwave cloud forcing (LWCF) strongly dominating during winter. During early summer, the sun is shining all day, making the SWCF much stronger than the LWCF. Since there is considerable ice-free surface (e.g. open ocean) in the Arctic, SWCF dominates resulting in a strong pike in net cooling (~ -6 W/m2) near the North Pole. This net cooling is less evident over Antarctica during DJF where the surface is always ice, making SWCF less important.