Top-Down and Bottom-Up Development of Climate Models

Given their impact on Earth's energy and water budgets, cloud and thunderstorm representations are among the most crucial components of climate models.  Thus, this seminar will begin with an overview of how clouds and thunderstorms are broadly represented in climate models (with a focus on the NASA Goddard Institute for Space Studies [GISS] Earth System Model [ESM]).  Satellites provide global 'top-down' observations of the distributions of clouds and thunderstorms, as well as information on the environments in which they are developing; thus, these data are important sources for determining whether ESMs are representing the various clouds and thunderstorms correctly.  Typically, satellite observations are averaged across long spatiotemporal timescales (in part to minimize random errors in the data), and then, the “averaged” datasets are used in ESM evaluation and development.  These ESM-observation assessment approaches are necessary, but come at a cost: difficulty in determining if ESMs are improving for the right reasons (i.e., improvement because clouds and thunderstorms are better represented).  This motivates a need to iteratively evaluate and develop ESMs using benchmarks at higher-resolution scales.  I will discuss various approaches to use observations and higher-resolution model simulations in new ways to inform future ESM development and to ensure that recently-improved ESM climatologies are not the result of errors cancelling at the scales on which clouds and thunderstorms occur.