Evaluating MERRA-2 Reanalysis Surface Radiation and Turbulent Fluxes with Satellite and Ground-Based Observations

The land surface energy in MERRA-2 is evaluated, in terms of errors in incoming and outgoing radiation, and the surface turbulent heat fluxes. The evaluation is performed by comparison to multiple satellite remotely sensed and ground-based observations. Since the ground-based observations are not necessarily representative of model grid scale estimates, and remotely sensed estimates inevitably involve modeling assumptions, none of these data sets can be assumed to represent the truth. Hence we focus on consistency between the comparisons to different data sets. Comparison is also made to screen level relative humidity and temperature observations, again focusing on consistency with comparisons to other data sets. The performance of MERRA-2 is compared to that of the antecedent reanalyses, MERRA, and also to the offline land-only MERRA-Land. Both MERRA-2 and MERRA-Land use observed precipitation in place of model-generated precipitation at the land surface. Compared to MERRA, the use of observed precipitation results in MERRA-Land and MERRA-2 both having significantly improved land surface hydrology, leading to improved partitioning between the surface turbulent fluxes. Compared to MERRA-Land, MERRA-2 further benefits from the feedback of the improved surface turbulent fluxes back to the atmosphere. For applications currently using MERRA or MERRA-Land, for example as forcing for offline land surface models, we recommend switching to MERRA-2, to take advantage if its improved accuracy, and greater consistency between the land surface hydrology and low-level atmospheric fields.