Robert Wood

Journal of the Atmospheric Sciences, in press

Detailed observations of stratiform boundary layer clouds on twelve days are examined with specific reference to drizzle formation processes. The clouds differ considerably in mean thickness, liquid water path ($LWP$) and droplet concentration. Cloud base precipitation rates differ by a factor of 20 between cases. The lowest precipitation rate is found in the case with the highest droplet concentration even though this case had by far the highest $LWP$, suggesting that drizzle can be severely suppressed in polluted clouds.

The vertical and horizontal structure of cloud and drizzle liquid water and bulk microphysical parameters are examined in detail. In general, the highest concentration of $r>20 \mu$m drizzle drops is found towards the top of the cloud and the mean volume radius of the drizzle drops increases monotonically from cloud top to base. The resulting precipitation rates are largest at the cloud base but decrease markedly only in the upper third of the cloud. Below cloud, precipitation rates decrease markedly with distance below base due to evaporation, and are broadly consistent in most cases with the results from a simple sedimentation-evaporation model. Evidence is presented that suggests evaporating drizzle is cooling regions of the subcloud layer which could result in dynamical feedbacks. A composite power spectrum of the horizontal spatial series of precipitation rate is found to exhibit a power-law scaling from the smallest observable scales to close to the maximum observable scale ($\sim$~30~km). The exponent is considerably lower (1.1-1.2) than corresponding exponents for $LWP$ variability obtained in other studies ($\sim$~1.5-2) demonstrating that there is relatively more variability of drizzle on small scales. Singular measures analysis shows that drizzle fields are much more intermittent than the cloud liquid water content fields, consistent with a drizzle production process that depends strongly upon liquid water content. The adiabaticity of the clouds, which can be modeled as a simple balance between drizzle loss and turbulent replenishment, is found to decrease if the timescale for drizzle loss is shorter than roughly 5-10 eddy turnover timescales. Finally, the data are compared with three simple scalings derived from recent observations of drizzle in subtropical stratocumulus clouds.