Initialization: Land

An important aspect of the GEOS-5 initialization concerns the treatment of the land. Observed precipitation data are used to construct a corrected version of the hourly MERRA (or GEOS-5 forward processing) precipitation fields, which are then used to force the land surface and generate enhanced soil moisture initial conditions for initializing the GEOS-5 seasonal forecasts. The corrections to the precipitation are obtained using GPCPv2.1 and CMAP pentad precipitation data following Reichle et al. (2011). First, the CMAP dataset is rescaled to match the (seasonally variable) long-term climatology of the GPCP data because the GPCP climatology is considered more correct than that of the CMAP data. (Because of its 3-month latency, however, GPCP data cannot be used directly in the precipitation corrections for operational seasonal forecasting.) Next, hourly MERRA total precipitation is time averaged and re-gridded to the scale of the correcting CMAP dataset (i.e., to pentad and 2.5° resolution). Next, separately for each pentad of each year and for each 2.5° grid cell, a scaling factor is computed by determining the ratio of the (climatologically adjusted) CMAP estimate to the MERRA data (i.e., on the grid and at the time scale of the correcting observations). Finally, these scaling factors are re-gridded back to the MERRA grid and applied to the MERRA data—a scaling factor derived for a given grid cell and year/pentad is applied to the MERRA precipitation rates (large-scale precipitation, convective precipitation, and snowfall separately) in each of the 120 hourly time steps within that pentad. By construction, the corrected MERRA precipitation is nearly identical to the CMAP estimates at the pentad and 2.5° resolution. The diurnal cycle, the frequency and relative intensity of rainfall events at the sub-pentad scale, and the sub-2.5° spatial variations are entirely based on MERRA estimates. To summarize, the precipitation corrections are constructed separately for each pentad of each year and thus go well beyond a mere climatological adjustment.

Reichle, Rolf H., Randal D. Koster, Gabriëlle J. M. De Lannoy, Barton A. Forman, Qing Liu, Sarith P. P. Mahanama, Ally Touré, 2011: Assessment and Enhancement of MERRA Land Surface Hydrology Estimates. J. Climate, 24, 6322-6338. doi: http://dx.doi.org/10.1175/JCLI-D-10-05033.1