hide
Free keywords:
-
Abstract:
An accurate determination of location and amount of liquid water in clouds is crucial for precipitation formation, cloud lifetime, and cloud radiative effects. Most remote-sensing retrievals, such as Cloudnet use lidar measurements to infer the location of liquid cloud droplets from measurements. However, lidar observations are of very limited use for optically thick or multilayer mixed-phase clouds (MPC) where they usually underestimate the presence of liquid water due to full signal attenuation, leading to large biases in simulated radiative fluxes. At the same time, general circulation models largely overestimate the downwelling shortwave radiation at the bottom of the atmosphere especially in the Southern Ocean regions. We argue that, in order to reduce this shortwave radiation bias in models, we first need better observational-based retrievals for supercooled-liquid detection that can be used for model validation. For this purpose, the machine-learning-based retrieval VOODOO is used to capture the extent of liquid layers over the complete vertical range of the clouds.To conceptualize the latter, a case study from the DACAPO-PESO campaign in Punta Arenas, Chili (53.13° S, 70.88° W) was investigated in detail by performing a radiative closures study. The shortwave cloud radiative effects of multilayer non-precipitating stratiform MPC - with liquid water layers detected by Cloudnet and VOODOO - was determined using a 1-D radiative transfer simulator and validated with downwelling pyranometer observations. The shortwave radiation bias was reduced by a factor of two suggesting that improved liquid-layer detection helps to decrease the shortwave radiation bias in radiative transfer simulations.