Insights into Cloud Radiative Variability using Cloud Classifications

The interannual variations of cloudiness contributions to the planet's radiation budget may be revealing about cloud radiative feedbacks and can be gleaned by studying the interannual variability of the Cloud Radiative Effect (CRE). The interpretation of the CRE variability, however, requires association with cloud changes. Such an association can be facilitated with a record that links CRE with various cloud classes and continuously tracks the breakdown of CRE by cloud class. This is presumably the motivation behind the creation of the new CERES "Flux-by-Cloud-Type" product. Another approach is to accept that at scales of ~100km even on a daily basis, clouds are mixtures of cloud types where one or two types may be dominant. It therefore makes sense to identify and classify not cloud types per se, but the most common of these mixtures in a way that will allow the dominant cloud types to emerge from the classification as well. This is the main idea behind the MODIS Cloud Regimes (CRs) (and previously the ISCCP Weather States), both representing the most common cloud mixtures where one cloud type usually dominates, and derived by applying k-means clustering on joint cloud optical thickness-cloud top pressure histograms, more recently those in MODIS Collection 6.1 (C6.1). By combining knowledge from MODIS of what CR occurs in a one-degree grid cell during a Terra and Aqua overpass and what the associated top-of-the atmosphere (TOA) shortwave (SW) and longwave (LW) CRE was observed by CERES, as transformed into a daily value in the SSF1deg-daily product, we can build a climatology of TOA CRE by CR. Note that this is not possible with monthly CERES EBAF data, and not advisable to construct with CERES SYN1deg fluxes which are not temporally uniform due to changes in geostationary satellite input. Our goal in this presentation is to show initially the climatological CERES CRE breakdown by the newly-derived MODIS C6.1 CRs, but to also examine subsequently whether the breakdown exhibits interannual variability, namely to examine the degree to which individual CR contributions to the CRE are stable during a time period spanning almost two decades. Digging further, we are interested in finding whether the change, or lack thereof, of CR-specific contributions to global CRE is due to changes in the frequency of occurrence or the properties of the CR, or whether compensations between these two factors occur for CRs that are relatively stable in their CRE contributions.