Modeling Tools

Global Earth System Model

The global Earth system model used for this project is the Community Earth System Model (CESM) (http://www.cesm.ucar.edu/) developed by the National Center for Atmospheric Research (NCAR). CESM is a leading US Earth System Model and has a long history of participating in the Intergovernmental Panel on Climate Change (IPCC) activities. Although majority of its climate simulations have been carried out at a nominal resolution of 1°, NCAR does maintain a high-resolution version of the coupled model for research purposes. The high-resolution version of CESM uses 0.25° horizontal resolution in its atmospheric and land components, while the ocean and sea-ice components share the same 0.1° horizontal resolution. It has been successfully integrated for over 100 years and its solutions show many improvements compared to those of the work-horse nominal 1° resolution CESM simulations. Small at al. (2014) summarize the major improvements of the high-resolution (HR) CESM compared to the standard version.  These include more realistic annual-mean sea surface temperature (SST) in the equatorial Pacific and El Niño-Southern Oscillation (see Figure), as well as reduced warm SST biases in all eastern boundary upwelling regions, including the notorious Benguela SST bias. Additionally, the HR-CESM simulation enables small-scale features of the climate system to be represented, such as ocean mesoscale eddies, air-sea interactions over ocean frontal zones, mesoscale systems generated by the Rockies, and tropical cyclones. We believe this HR-CESM can provide a viable framework for high-resolution global Earth System predictions.

Visualization from various earth systems computer models

ENSO composites based on warm minus cold events of greater than standard deviation of Nino3.4 time series. (a–c) For June-July-August (JJA) before the peak and (d–f) for following December-January-February (DJF). Years 1920–2011 of HadISST and NOAA Merged Land-Ocean Surface Temperature Analysis (MLOST) [Smith and Reynolds, 2005] land surface temperature, and sea level pressure (SLP) from the twentieth Century Reanalysis [Compo et al., 2011] (a & d). High-resolution CESM years 15–100 (Figures b & e), and low-resolution CESM, years 1–166 (Figures c & f). Surface temperature (°C, color) and SLP (contours—interval 2hPa). (After Small et al., 2014)

Progress and Current Status

A concerted effort has made to port and optimize the HR-CESM code to the heterogeneous HPC system of TaihuLight in Wuxi, China (https://www.top500.org/system/178764). The current version of the high-resolution CESM1.3 code has achieved a throughput rate of 3 model simulation years per day (MSYD) using approximately 25% of computing nodes of the TiahuLight HPC system. The effort is continuing to improve the performance of the HR-CESM on TaihuLight.  The goal is to achieve a throughput rate of 5 MSYD within one year.

Regional Earth System Model

The regional modeling approach has the advantage of higher accuracy and the ability to resolve smaller-scale processes such as convections in the atmosphere and submesoscale eddies in the ocean that even the high-resolution global models cannot resolve.  Therefore, it complements the high-resolution global modeling approach and can be used for regional downscaling where regional information is being provided by regional models – some coupled – that are forced by surface and boundary conditions from coarser resolution global simulations.
As a part of the collaboration between the Ocean University of China (OUC), TAMU and NCAR, a regional CESM (R-CESM) has been developed.  The model uses the Common Infrastructure for Modeling the Earth (CIME) that is the coupling infrastructure the latest version of CESM (CESM2.1.0) (http://www.cesm.ucar.edu/models/cesm2/whatsnew.html).  The current version of R-CESM consists of the Weather Research and Forecasting Model (WRF) version 3.5.1 as the atmospheric component, the Regional Ocean Modeling System (ROMS) version 3.5, and the Community Land Model (CLM), version 4.0.  It also provides support for several different configurations many of which use stub and data models that are inactive and simply report data to the coupler from pre-written files. An earlier version of this regional coupled system has been used to study interactions between ocean mesoscale eddies and atmosphere and their impact on midlatitude storm tracks (Ma et al, 2016).  R-CESM is skillful in simulating Hurricane Harvey in the Gulf of Mexico, including its rapid intensification and landfall location (Figure).
Hurricane Harvey simulated by a 3km resolution computer model

Hurricane Harvey simulated by a 3 km resolution R-CESM from August 22 to September 2, 2017.

Hurricane Harvey

Progress and Current Status

New capabilities have been added to R-CESM, including a) option to run R-CESM using CESM2 framework, i.e. using CESM2 air-sea and air-land fluxes, and optionally the CESM land model-CLM, and b) an option to embed ROMS in the global POP model.