From Optimized Random Walk to Uncertainty Quantification: Bias reduction and Process Improvement in the Next Generation Community Earth System Model Atmosphere

Abstract:

Many critical uncertainties and biases remain in Earth System Models. Biases are obvious errors, such as the excessive absorbed shortwave radiation in the Southern Ocean. Uncertainties also remain due to limitations in our observations, such as the ice water path in tropical cold clouds. These biases and uncertainties limit our ability to understand the future trajectory of climate. The development process for the next generation version of the Community Atmosphere Model is described with a focus on how some of these key uncertainties and biases have been reduced, with a particular focus on the atmosphere. The ice and mixed phase processes have been improved, as has the representation of moist turbulent motions in shallow clouds. The model formulations will be described, as well as some of the key biases addressed and the pathways for how the model was improved. Methodologies used included detailed comparisons to observations using simulators and forecast methods, explorations of critical pathways, and objective analysis systems for parameter optimization. The presentation will conclude with some thoughts on best practices for future model development.