Evaluation of Diagnostic CO2 Flux and Transport Modeling in NU-WRF and GEOS-5

Friday, 18 December 2015: 10:50
3006 (Moscone West)
Stephan R Kawa1, George James Collatz1, Zhining Tao2, James S Wang3, Lesley E Ott4, Yuping Liu5, Arlyn E Andrews6 and Colm Sweeney7, (1)NASA Goddard Space Flight Center, Greenbelt, MD, United States, (2)Universities Space Research Association Columbia, Columbia, MD, United States, (3)NASA Goddard Space Flight Ctr, Greenbelt, MD, United States, (4)NASA Goddard Space Flight Center, Global Modeling and Assimilation Office, Greenbelt, MD, United States, (5)SSAI, Greenbelt, MD, United States, (6)NOAA Earth System Research Lab, Boulder, CO, United States, (7)NOAA Boulder, ESRL, Boulder, CO, United States
We report on recent diagnostic (constrained by observations) model simulations of atmospheric CO2 flux and transport using a newly developed facility in the NASA Unified-Weather Research and Forecast (NU-WRF) model. The results are compared to CO2 data (ground-based, airborne, and GOSAT) and to corresponding simulations from a global model that uses meteorology from the NASA GEOS-5 Modern Era Retrospective analysis for Research and Applications (MERRA). The objective of these intercomparisons is to assess the relative strengths and weaknesses of the respective models in pursuit of an overall carbon process improvement at both regional and global scales. Our guiding hypothesis is that the finer resolution and improved land surface representation in NU-WRF will lead to better comparisons with CO2 data than those using global MERRA, which will, in turn, inform process model development in global prognostic models. Initial intercomparison results, however, have generally been mixed: NU-WRF is better at some sites and times but not uniformly. We are examining the model transport processes in detail to diagnose differences in the CO2 behavior. These comparisons are done in the context of a long history of simulations from the Parameterized Chemistry and Transport Model, based on GEOS-5 meteorology and Carnegie Ames-Stanford Approach-Global Fire Emissions Database (CASA-GFED) fluxes, that capture much of the CO2 variation from synoptic to seasonal to global scales. We have run the NU-WRF model using unconstrained, internally generated meteorology within the North American domain, and with meteorological ‘nudging’ from Global Forecast System and North American Regional Reanalysis (NARR) in an effort to optimize the CO2 simulations. Output results constrained by NARR show the best comparisons to data. Discrepancies, of course, may arise either from flux or transport errors and compensating errors are possible. Resolving their interplay is also important to using the data in inverse models. Recent analysis is focused on planetary boundary depth, which can be significantly different between MERRA and NU-WRF, along with subgrid transport differences. Characterization of transport differences between the models will allow us to better constrain the CO2 fluxes, which is the major objective of this work.