Abstract:

Earth system models and global climate models are essential tools for predicting and projecting future climate on various timescales. However, future climate projections have large uncertainties mostly due to model forcing uncertainty, internal variability, as well as model structure errors. Model structure errors generally come from missing or erroneous representations of certain processes or their interactions in the models due to insufficient understanding of these processes. In this presentation, I will cover series of our recent studies on better understanding of atmospheric and land surface processes through observations and multi-model analysis, which can lead to better simulations of precipitation and surface air temperature over the Central United States. The simulations of temperature and precipitation over this region are known to be particularly challenging in regional or global weather forecasting and climate models. I will also demonstrate that increasing model resolutions down to cloud-resolving scales can improve certain features of precipitation and temperature, but it does not necessary resolve all the problems. Better ways to diagnosing model problems associated with temperature and precipitation will be discussed. In particular, a hierarchical modeling framework, as well as phenomena-based approach have better potential to advance our understanding of climate system especially when we are entering the era of using global storm-resolving models for climate-related applications at km scale resolutions.