Humidity variability revealed by a sounding array and its implications for cloud representation in GCMs

Abstract

AbstractSubgrid‐scale humidity variability is critical for statistical cloud schemes. The Atmospheric Radiation Measurement (ARM) sounding array, covering a comparable size to a grid box of current global climate models (GCMs), provides estimates of subgrid‐scale humidity variability. Using radiosonde measurements from ARM sounding arrays, humidity variabilities at a tropical and a midlatitude site are analyzed. Normalized moisture variance is small in the lower troposphere but tends to be large in the middle troposphere. Moisture variability dominates relative humidity (RH) variability at the tropical site, but the contribution by temperature variability cannot be ignored at the midlatitude site. Based on the analysis, a simple linear relationship between RH variance and mean is proposed. Cloud schemes assuming Gaussian distributions give comparable cloudiness as those assuming a beta distribution considering both moisture variance and skewness. A probability distribution function scheme assuming a Gaussian distribution of RH has more flexibility than a RH threshold scheme, which tends to overestimate cloudiness in the moist boundary layer over the tropics. Supersaturation with respect to ice needs to be allowed in the upper troposphere over the tropics to reduce the persistent overestimated cloudiness there. Similar comparison results are found from analysis of cloud resolving model simulations. The study suggests that measurements from sounding arrays would be valuable for statistical cloud scheme evaluation and development.