3.4 Ensemble large-eddy simulation for the DOE Atmospheric Radiation Measurement FacilityÕs LES ARM Symbiotic Simulation and Observation (LASSO) project

Gustafson, William I., Pacific Northwest National Laboratory (PNNL), Andrew M. Vogelmann, Brookhaven National Laboratory (BNL), Heng Xiao, PNNL, Satoshi Endo, Tami Toto, BNL, Zhijin Li, Xiaoping Cheng, Jinwon Kim, University of California Los Angeles

The Department of Energy Atmospheric Radiation Measurement (ARM) Climate Research Facility is transitioning the Large-Eddy Simulation (LES) ARM Symbiotic Simulation and Observation (LASSO) Workflow from its pilot phase to operations. ARM will be performing routine LES for shallow convection days at its Southern Great Plains (SGP) site in Oklahoma. Over the last two years the pilot phase has explored various forcing options and model configurations. The overall vision is to generate an ensemble of LES for each simulated case to increase the chances of correctly reproducing the meteorology and clouds.

The overall workflow can be separated into three portions. The first is the production of an ensemble of large-scale forcings for the LES. In general, uncertainty in the forcing is the primary cause of differences in the LES output. So, we plan to run an LES ensemble based on various forcings to best capture the conditions for each given case. The second portion of the workflow is running the LES ensemble. The third is blending the LES output with ARM observations to generate a suite of diagnostics and metrics in the form of "data bundles" for user consumption. The data bundles are conceptually organized to meet a hierarchy of needs, from the metrics for doing quick case evaluation, to LES summary statistics and volumetric model output for users desiring to use ARM's LES runs directly, to the data necessary to reproduce the LES and bundle metrics for users whose needs require enhancing the LES. Efficient discovery and use of the data bundles is an important component of LASSO, so a prototype "Bundle Browser" has been developed as an addition to ARM's Data Discovery web interface.

This presentation will provide an overview of what we have learned during the pilot project, show examples of the Bundle Browser, and provide example ensembles chosen from the hundreds of simulations generated during the pilot phase.