Frontiers in Remote Sensing (Dec 2022)

Revisiting cloud overlap with a merged dataset of liquid and ice cloud extinction from CloudSat and CALIPSO

  • Lazaros Oreopoulos,
  • Nayeong Cho,
  • Nayeong Cho,
  • Dongmin Lee,
  • Dongmin Lee

DOI
https://doi.org/10.3389/frsen.2022.1076471
Journal volume & issue
Vol. 3

Abstract

Read online

We update the parameterization capturing the variation of parameters that describe how cloud occurrence (layer cloud fraction) and layer cloud optical depth (COD) distributions overlap vertically. Our updated analysis is motivated by the availability of a new dataset constructed by combining two products describing the two-dimensional extinction properties of liquid and ice phase clouds (and their mixtures) according to active cloud observations by the CloudSat and CALIPSO satellites. As before, cloud occurrence overlap is modeled with the decorrelation length of an inverse exponential function describing the decay with separation distance of the relative likelihood that two cloudy layers are overlapped maximally versus randomly. Similarly, cloud optical depth distribution vertical overlap is described again with a decorrelation length that describes the assumed inverse exponential decay with separation distance of the rank correlation between cloud optical depth distribution members in two cloudy layers. We derive the climatological zonal variability of these two decorrelation lengths using 4 years of observations for scenes of ∼100 km scale length, a typical grid size of numerical models used for climate simulations. As previously, we find a strong latitudinal dependence reflecting systematic differences in dominant cloud types with latitude, but substantially different magnitudes of decorrelation length compared to the previous work. The previously used parameterization form is therefore updated with new parameters to describe the latitudinal dependence of decorrelation lengths and its seasonal shift. Similar zonal patterns of decorrelation length are found when the analysis is broken down by different cloud classes. When the revised parameterization is implemented in a cloud subcolumn generator, simulated column cloud properties compare to observations quite well, and so do their associated cloud radiative effects, but improvements over the earlier version of the parameterization are marginal.

Keywords