Dynamical Downscaling of NASAGISS Model E Continuous MultiYear

Dynamical Downscaling of NASA/GISS Model. E: Continuous, Multi-Year WRF Runs Tanya L. Otte 1, Jared H. Bowden 1, Christopher G. Nolte 1, Martin J. Otte 1, Jonathan E. Pleim 1, Jerold A. Herwehe 1, Greg Faluvegi 2, and Drew T. Shindell 2 1 U. S. EPA, Research Triangle Park, North Carolina Institute for Space Studies, New York 2 NASA/Goddard 9 th Annual CMAS Conference, Chapel Hill, NC 11 October 2010 Office of Research and Development National Exposure Research Laboratory, Atmospheric Modeling and Analysis Division

Why are we performing this research? Regional Impacts of Climate Change Global Climate - Comprehensive science - Emissions scenarios - Multi-century data - Coarse resolution Large, well-established programs 2 Regional Climate Modeling a. k. a. Downscaling - Effects on air quality and environment - Assessment, mitigation - Detailed local land cover, topography, etc. Our interests

What is “dynamical downscaling”? Global climate model (GCM) creates coarse gridded future climate with world-wide coverage. GCM Regional climate model (RCM) generates gridded higher-resolution climate predictions over focal area. 3 More detail in local effects from: - scale-appropriate physics - topography & land/water interfaces - urban areas (population centers) - precipitation patterns

Our Research Problem…Simplified Freedom of RCM to develop smaller-scale processes Constraint of RCM toward GCM delicate balance Keeps RCM climate consistent with GCM Allows RCM climate to deviate from GCM Decreases variability Increases variability More constraint toward GCM 4 Less constraint toward GCM RCM to be constrained to GCM using nudging.

Downscaling Configuration • NASA/GISS Model. E (6 years, ca. 2000) – 1 of 3 U. S. GCMs in IPCC AR 5 – Coupled atmosphere-ocean model creates GCM forecast – 2. 0° x 2. 5°, 6 -h fields, 40 s-p layers up to 0. 1 h. Pa – Use native vertical layers in downscaling – Validate WRF vs. Model. E • WRFv 3. 2 – 108 -km domain, 34 layers, model top at 50 h. Pa – CAM LW and SW Radiation – WSM 6 Microphysics – Yonsei University PBL – NOAH Land-Surface Model – Grell Cumulus Parameterization – Various analysis and spectral nudging configurations 5

Domains and Topography Model. E (cropped to WRF domain and projection) 6 WRF at 108 -km

Multi-year downscaled simulations (even without nudging) can capture interannual variability from the GCM. Monthly Area-Averaged 2 -m Temperature Distribution (Plains) for “ 2002 -2007” Model. E (“obs”) “ 2003” “ 2004” “ 2005” “ 2006” “ 2007” Increased range “ 2002” No Nudging mean maximum 95 th %ile 90 th %ile 75 th %ile 50 th %ile 25 th %ile 10 th %ile 5 th %ile minimum 7 Analysis Nudging Spectral Nudging

Multi-year WRF runs using Model. E have persistent and systematic regional and seasonal biases, even with nudging. Monthly Area-Averaged 2 -m Temperature Difference by Region Midwest Northwest 2002 2003 2004 2005 2006 2007 No Nudging nearly zero annual bias 4 Nudging Options Plains Southwest annual periodicity of biases Northeast annual periodicity of biases Jan. distinct cold bias Southeast annual periodicity of biases Apr. Sept. 8 Result of inconsistencies in GCM/RCM physics or errors in modeling physical processes? Corroborate with multi-year reanalysis runs?

Multi-year WRF runs using Model. E can broadly capture monthto-month trends in precipitation within different regions. Monthly Area-Averaged Accumulated Precipitation Difference by Region Northwest 2002 2003 2004 2005 Southwest 9 2006 Midwest Northeast Plains Southeast 2007

Subtle changes in nudging strategy can have a large impact on results. Plains U. S. region Precipitation Analysis Nudging (Sens. 1) It’s not just about whether or not nudging is used but how it is used. 2 -m Temperature JFMAMJJASOND Precipitation Model. E Analysis Nudging (Sens. 2) 10 WRF 2 -m Temperature JFMAMJJASOND

One size (configuration) will not fit all (regions and applications). 6 -Year Monthly Error from 17 Nudging Sensitivities 2 -m Temperature Difference (Plains U. S. region) 11 Strategy may ultimately be dictated by application and region(s) of interest.

Next Steps • Analysis of upper-air and column-averaged fields • Space-time analysis • Additional nudging sensitivities, as warranted by analysis • Nested (36 -km) runs and nesting strategies • Longer time slices (and climate change) 12