Multi-scale modeling

Much of my climate simulation research applies cloud “superparameterization (SP)” [3-6,9-11], which is a multi-scale atmospheric modeling approach that embeds thousands of moist turbulence-permitting cloud-resolving models (CRMs) within global climate models, to replace the traditional error-prone approximations (or “parameterizations”) of unresolved convection and clouds. SP is not without its own limiting idealizations so one focus of my work at UCI has attempted to clarify its trade-offs for practical issues of weather and climate simulation. For example, in [14] I revealed that an interesting trade-off of using limited extent CRMs is to artificially limit the efficiency of vertical mixing in tropical deep convection regions, controlled by the room available for compensating subsidence to balance deep updrafts. In [19], together with my first PhD student S. Yu, I revealed a surprising sensitivity to increasing the frequency of communication between the cloud-resolving and planetary-resolving scale regimes — stiffer communication drives a lower tropical gross moist stability and thus juicier precipitation extremes. This has raised new questions about the emergent dynamics of superparameterized climate models, worthy of future work, and meanwhile has become a useful new tuning knob for developers. Meanwhile, together with my first postdoc G. Kooperman, I published the first comprehensive multi-model analysis of SP’s effects on the simulated rainfall distribution and its climate sensitivity, including moderate vs. extreme rainfall statistics, and their geographic structure [21,25]. In follow-on work honing in on some promising effects of SP on the Intertropical Convergence Zone (ITCZ) rainfall band in the tropics, my student S. Yu helped discover that the neglect of convective momentum transport in the traditional SP algorithm is actually a source of compensating bias supporting the ITCZ [27].


Contact

164 Rowland Hall
University of California, Irvine
Irvine, CA 92697-4675