Emulation to simulate low resolution atmospheric data

Technical Report · Wed Aug 01 00:00:00 EDT 2012

DOI:https://doi.org/10.2172/1051448· OSTI ID:1051448

Archibald, Richard K ^[1]; Evans, Katherine J ^[1]

---

1. ORNL

Climate simulations require significant compute power, they are complex and therefore it is time consuming to simulate them. We have developed an emulator to simulate unknown climate datasets. The emulator uses stochastic collocation and multi-dimensional in- terpolation to simulate the datasets. We have used the emulator to determine various physical quantities such as temperature, short and long wave cloud forcing, zonal winds etc. The emulation gives results which are very close to those obtained by simulations. The emulator was tested on 2 degree atmospheric datasets. The work evaluates the pros and cons of evaluating the mean first and inter- polating and vice versa. To determine the physical quantities, we have assumed them to be a function of time, longitude, latitude and a random parameter. We have looked at parameters that govern high stable clouds, low stable clouds, timescale for convection etc. The emulator is especially useful as it requires negligible compute times when compared to the simulation itself.

• Research Organization: Oak Ridge National Lab. (ORNL), Oak Ridge, TN (United States). National Center for Computational Sciences (NCCS)
• Sponsoring Organization: USDOE Office of Science (SC)
• DOE Contract Number: DE-AC05-00OR22725
• OSTI ID: 1051448
• Report Number(s): ORNL/TM-2012/317; KP1703020; ERKP746
• Country of Publication: United States
• Language: English