Satellite measurements and climate modelling of water in the tropical upper troposphere

Licentiate thesis, 2011

The International Panel on Climate Change (2007) has identified key areas of uncertainty with regards to future climate change, e.g., the upper troposphere. A portion of the uncertain- ties come from climate models and this makes model accuracy very important. Improvement in models can be achieved through evaluations that highlight problem areas. Satellites provide quality observations that can increase the level of scientific understanding of processes that are, even now, poorly understood. But satellites suffer from errors and therefore also need to be assessed. An introduction to the basics of the atmospheric component of a climate model and a cursive overview of satellite remote sensing, are presented. The model, EC Earth, is introduced where emphasis is placed on model parametrizations and cloud microphysics. The representation of tropical upper tropospheric (500–100hPa) humidity and clouds is evaluated in the climate model EC Earth. This is done with the aid of several satellite datasets using a model-to-retrieval method and involves a novel use of multiple measurement techniques to measure upper tropospheric variables. The effects of these variables on the outgoing long- wave radiation are also compared with satellite observations. The study shows that relative humidity in the model agrees well with observations from the Microwave Limb Sounder and the Atmospheric Infrared Sounder satellites, though the observations contain large uncertain- ties. EC Earth produces too much cloud above 200 hPa with often twice as much as the best estimates of cloud fraction (CALIPSO). The model’s precipitating ice and suspended ice are combined and compared to CloudSat’s ice water content (IWC). Above 300hPa, the model IWC is lower than the observations while below the model overestimates IWC. The impact on the outgoing longwave radiation, over the Tropics, by the model cirrus is underestimated by ∼9 W m−2 while, regionally, this bias can be ∼20 W m−2.