CloudSat-CALIPSO observations

Evaluation of clouds and convection in a limited-area numerical weather prediction system using near real-time CloudSat-CALIPSO observations

Abstract

Near real-time measurements from the CloudSat and Cloud-Aerosol Lidar and Infrared Pathfinder Satellite Observation (CALIPSO) mission are used to evaluate the representation of the hydrometeor frequency of occurrence (HFO) in the limited-area version of the Australian Community Climate and Earth-System Simulator (ACCESS-A), using one year of collocated satellite data and model forecasts.  We show that the use of short-term simulations and sensitivity tests of model parameterizations evaluated in near real-time by CloudSat-CALIPSO observations should be an appropriate strategy to speed up model improvement. ACCESS-A is found to over-predict the HFO below 12 km height (primarily over the Southern Ocean), and largely under-predict the HFO above 12 km height (primarily in the Tropics. The largest over-prediction of HFO was found over the Maritime Continent islands and over the Southern Ocean, while the largest under-prediction of HFO was found in the 10°S-20°S latitude band. The seasonal variability of these biases was found to be small suggesting that these model problems can be investigated with short-term simulations.

Selected skill scores were then analysed as a function of lead time, hydrometeor height in the troposphere, and season. The highest forecast skill was found in the subtropics, mostly owing to a low incidence of false positives. Overall the ACCESS-A forecast skill at the Southern Hemisphere mid-latitudes is comparable to that of the North-Atlantic / European version of the UK Met Office Unified Model at Northern Hemisphere mid-latitudes. It is also found that mid-latitude low-level hydrometeors and tropical low-level and high-level hydrometeors during the Southern Hemisphere summer are clearly the most challenging hydrometeors to simulate in the domain. The mid-latitude low-level hydrometeors are overall well predicted when observed, but over-predicted when not observed. The tropical high-level hydrometeors are not predicted often enough when observed, but most of the time correctly not predicted when not observed.