Bridging Scales in Climate System Modeling

Todd Ringler (LANL)

Abstract

In the coming decade and beyond, the climate modeling community will be
challenged to resolve scales and processes that are far beyond its current
reach. The challenge to resolve these small scales and new processes
arises primarily for two reasons. The first reason is that there are likely to
be currently unresolved processes that have a significant influence on the
global climate system. Examples of this might include ice streams (O(km))
within large-scale ice sheets, ice shelves (O(km)) collapsing due to
interaction with ocean processes, cloud process (O(km)) with the
atmosphere and the dependence of ocean biochemistry on ocean eddy
(O(10 km)) activity. The second reason for this challenge is the pressing
need to quantitatively characterize the regional-scale signature of
anthropogenic climate change in order to assess its impacts to socioeconomic
activity.

The ability to resolve the scales noted above throughout the relevant
climate system component is beyond the current computing capacity of
even the most powerful computer available today. Basic scaling arguments
of computer resources available into the future indicate that this will remain
true for at least the next two decades. In this talk I will outline the
approaches we might take to accommodate these relevant processes, yet
retain a computationally-tractable climate system model. Within the context
of the shallow-water system, I will demonstrate a prototype of the approach
we are developing in order to produce model simulations across scale. I will
also outline our approach for targeting the emerging class of hybrid cpugpu
architectures.

back to agenda