Simplified ISCCP cloud regimes for evaluating cloudiness in CMIP5 models

摘要: We take advantage of ISCCP simulator data available for many models that participated in CMIP5, order to introduce a framework comparing model cloud output with corresponding observations based on the regime (CR) concept. Simplified global CRs are employed derived from co-variations three variables, namely optical thickness, top pressure and fraction (τ, p c , CF). Following evaluation criteria established companion paper ours (Jin et al. 2016), we assess simulation performance how well simplified simulated terms similarity centroids, values map correlations relative-frequency-of-occurrence, long-term total amounts. Mirroring prior results, modeled clouds tend be too optically thick not as extensive observations. high-altitude storm activity here compared previous study, but other regimes containing near-overcast low show improvement. Models have performed against defined by joint τ–p histograms distinguish themselves again here, improvements previously underperforming also seen. Averaging across does yield drastically better picture, except geographical locations. Cloud seems thus more forgiving than using histogram-based while still strict enough reveal weaknesses.