When related to the visible extinction coefficient αext, De exhibits a maximum. Moreover, Tr – Ttop related to αext (Tr = radiative temperature; Ttop = temperature at cloud top), also exhibits a maximum at or near this De maximum. Note that Tr – Ttop is a proxy for cirrus geometrical thickness because Tr is typically near the cloud center. This maximum De is interpreted as a transition region in terms of extinction which separates cirrus clouds formed through heterogeneous ice nucleation only (henceforth het and het cirrus) and through het and homogeneous ice nucleation (hom) combined (henceforth hom cirrus). Hom cirrus have higher Ni and IWC relative to het cirrus.
Thus, αext is sensitive to changes resulting from hom that separates the het and hom cirrus regimes. This led to a conceptual model of cirrus cloud formation that integrates these two types of cirrus cloud, which is partially confirmed by a simple model based on hom theory and is also supported by other published studies on cirrus clouds.
The above analysis allowed us to globally map the fraction of hom cirrus by season. While in the tropics (± 30° latitude) this fraction is typically < 15%, outside the tropics this fraction varies by ~ 20% to 35% during the winter. Estimating the cloud radiative effect (CRE) as the product of cloud frequency of occurrence and cloud optical depth, the cirrus CRE over ocean outside the tropics during winter is dominated by the hom cirrus fraction.
Based on theoretical expectations, the balance of het cirrus and hom cirrus may partly depend on the concentration of mineral dust at cirrus levels, with a lower hom fraction when dust levels are relatively high. The fraction of hom cirrus has a strong seasonal dependence far downwind of the main Asian dust sources (Taklimakan and Gobi deserts) and throughout the Southern Ocean; higher in winter relative to summer. This inversely coincides with dust loading and dust aerosol optical depth in these regions at cirrus levels, suggesting that seasonal changes in mineral dust concentration may alter the hom cirrus fraction. This could have a significant impact on the cirrus CRE as shown by numerous studies. In addition, there is clear evidence of a Twomey effect for het cirrus clouds, assuming there are more ice nuclei over land relative to oceans.

