Verification of mesoscale cloud forecasts utilizing composite and fuzzy techniques

Operational cloud forecasts generated by the Coupled Ocean/Atmosphere Mesoscale Prediction System (COAMPS®) were verified over the eastern Pacific Ocean. The study focused on the accuracy of cloud forecasts associated with cyclone and convective activity during the late winter and spring of 2007. Condensed total water (liquid and solid) path was used as a proxy for cloud cover to compare the forecasts with retrievals from the Geostationary Operational Environmental Satellites (GOES). Multiple statistics were collected, ranging from simple correlations and histograms to more sophisticated fuzzy and composite statistics. The results show that synoptic scale systems were generally well predicted to at least two days, with the primary error being an over estimation of deep cloud occurrence. Smaller sub-synoptic systems were subject to spatial and timing biases in that a number of the forecasts were lagged by 3-6 hours. Despite the bias, 60-70% of the forecasts of the mesoscale phenomena displayed useful skill.