A11F-0109
Impacts of Lateral-Boundary-Condition Errors on Regional Climate Downscaling: Lessons Learned from the NASA Downscaling Project

Monday, 14 December 2015
Poster Hall (Moscone South)
Weile Wang1, Takamichi Iguchi2, Jonathan Case3, Eric M Kemp4, William Putman4, Di Wu4, Robert Ferraro5, Christa D Peters-Lidard4 and Ramakrishna R Nemani6, (1)CSUMB & NASA/AMES, Seaside, CA, United States, (2)University of Maryland College Park, College Park, MD, United States, (3)ENSCO, Inc./NASA Marshall Space Flight Center, Huntsville, AL, United States, (4)NASA Goddard Space Flight Center, Greenbelt, MD, United States, (5)NASA Jet Propulsion Laboratory, Pasadena, CA, United States, (6)NASA Ames Research Center, Moffett Field, CA, United States
Abstract:
The ongoing NASA dynamical climate downscaling project runs the NASA Unified WRF (NU-WRF) model to produce high-resolution climate information with prescribed lateral boundary conditions (LBCs) from the agency’s latest global reanalysis product MERRA2. The domain of the project is set over a vast area including the continental US and the adjacent oceans, so that a variety of important meteorological phenomena (e.g., atmospheric river, mesoscale convective system, and the northeastern winter storms) can be simultaneously simulated. The regional modeling experiments include combinations of three spatial-resolution configurations (i.e., 24km, 12km, and 4km) and three spectral-nudging configurations (i.e., “no nudging”, “nudging over 2000km scales”, and “nudging over 600km scales”). In addition to the regional downscaling efforts, MERRA2 data are also being downscaled globally to 12km resolution with the NASA GEOS-5 general circulation model. Outputs from these modeling experiments provide a unique opportunity for the science community to study many important questions in the field of climate downscaling.

This study reports initial results regarding the impacts of LBC errors on the model experiments. Because in the regional modeling experiments information flows only one-way from MERRA2 to NUWRF, over time integration of the latter inevitably diverges from the former at the lateral boundaries as well as the interior of the domain. The differences between the un-nudged NUWRF runs and the MERRA2 data become distinguishable within a month or two of integration. Spectral nudging helps constrain the model differences between the downscaled and the original coarse-resolution climate fields. Specifically, nudging with more spatial signals (e.g., “600km”) seems to generate higher fidelity than between the NUWRF outputs and the MERRA2 forcing data. However, stronger nudging does not necessarily enhance the comparison results between the simulated climate fields and other independent observational datasets. No easy criterion is found yet with regard to determining the optimal strength of spectral nudging or, equivalently, the degree of freedom that should be given to the RCM (NUWRF) in the simulations, though more analysis efforts along this line are in progress.

<< Previous Abstract | Next Abstract >>