P24 The Smart Management of WRF Simulation (SMWRFS) system for the smarter operations of the WRF model for regional climate simulations

Yoo, Jinwoong, Jeffrey S. Dukes, Michael Baldwin, Matthew Huber, Purdue University, and Chanh Kieu, Indiana University Bloomington

A prototype of the Smart Management of WRF Simulation (SMWRFS, pronounced as "Smurfs") system was developed and employed for the efficient and seamless operations of the parallel executions of WRF regional climate simulations for Indiana climate change impact assessment project which is a state level effort to understand climate change and its multifaceted risks and impacts on the state. The main usage of the SMWRFS is to run the WRF simulations of any numbers in an automated fashion, avoiding time-consuming manual managements of namelist files, executions of simulations, and output files on the production scratch directories during restarts of simulations, which can be overwhelming in a long-term climate simulation with many time slices. Key tasks of the SMWRFS include: 1) to restart the WRF climate simulation from the point where the previous WRF simulation is left over; 2) to handle the simulation outputs professionally by storing the WRF model outputs (if available) to HPSS tape archive storage using HTAR method, managing the scratch disc space of the cluster computer efficiently. The SMWRFS was successfully used on four Linux clusters at Purdue University and one Cray cluster at Indiana University, storing the WRF simulation outputs on a HPSS simultaneously and collaboratively.