Horizontal structure of marine boundary layer clouds from centimeter to kilometer scales

Abstract

Horizontal transects of cloud liquid water content (LWC) measured at unprecedented 4‐cm resolution are statistically analyzed scale‐by‐scale. The data were collected with a Particulate Volume Monitor (PVM) probe during the winter Southern Ocean Cloud Experiment (SOCEX) on July 26, 1993, in a broken‐stratocumulus/towering‐cumulus cloud complex. Two scaling regimes are found in the sense that two distinct power laws, k−β are needed to represent the wavenumber spectrum E(k) over the full range of scales r ≈ 1/k. Detailed numerical simulations show that the scale break at 2–5 m is not traceable to the normal variability of LWC in the PVM's instantaneous sampling volume (1.25 cm3) driven by Poissonian fluctuations of droplet number and size. The two regimes therefore differ physically. The non‐Poissonian character of the small‐scale LWC variability is consistent with a similar finding by Baker [1992] for droplet number concentration obtained from Forward Scattering Spectrometer Probe (FSSP) data: at scales of a few centimeters, spatial droplet distributions do not always follow a uniform Poisson law. With β = 0.9 ± 0.1, the small‐scale (8–12 cm ≲ r ≲ 2–5 m) regime is stationary: jumps in LWC are highly variable in size and rapidly cancel each other, leading to short‐range correlations. By contrast, the large‐scale (5 m ≲ r ≲ 2 km) variability with β = 1.6 ± 0.1 is nonstationary: jumps are generally quite small, conveying a degree of pixel‐to‐pixel continuity and thus building up long‐range correlations in the low‐pass filtered signal. The large‐scale structure of the complex SOCEX cloud system proves to be multifractal, meaning that large jumps do occur on an intermittent basis, that is, on a sparse fractal subset of space. Low‐order, hence more robust, multifractal properties of the SOCEX clouds are remarkably similar to those of their First ISCCP Regional Experiment (FIRE) and Atlantic Stratocumulus Transition EXperiment (ASTEX) counterparts, and also to those of passive scalars in fully developed turbulence. This is indicative of a remarkable similarity in the micro‐physical and macrophysical processes that determine cloud structure in the marine boundary layer at very remote locales, especially since the particular SOCEX cloud system investigated here was rather atypical. Interesting differences are also found: in the scaling ranges on the one hand, and in higher‐order moments on the other hand. Finally, we discuss cloud‐radiative effects of the large‐ and small‐scale variabilities.