Application of GLDAS Framework to the Next-Version Global Forecast System at NCEP

Accurate estimation of land surface states (e.g., soil moisture, soil temperature, snowpack) is critical for improving prediction skill of coupled global weather and climate forecast systems, especially in these regions with strong land-atmosphere interactions (e.g., semi-arid/”hot spots” and polar areas). Observed soil moisture and temperature are sparse and not appropriate for direct use as initial and/or boundary conditions of these systems. Land surface states produced from these coupled systems often have large errors and drifts owing to substantial biases in the surface forcing (e.g., precipitation, downward shortwave radiation). As a good alternative for this purpose, Global Land Data Assimilation System (GLDAS) is to utilize the best available in-situ and remotely-sensed observations of land surface properties and forcing data to produce enhanced fields of land surface states in uncoupled mode. Based such a concept, the NCEP GLDAS was implemented as the land analysis component in the NCEP Climate Forecast System Reanalysis (CFSR) and the NCEP operational CFS/CDAS version 2 in January 2011. Since then, it is an expected task to extend this framework to NCEP Global Forecast System (GFS).

Yet, GLDAS is not included in NCEP GFS system, even newly implemented version of GFS (GFS.v15). To utilize the CFS/GLDAS framework to next version GFS - GFS.v16, an effort initiated at NCEP includes to use new soil and vegetation fields and to replace Noah model with NoahMP. As NoahMP model contains groundwater and vegetation dynamics module, it needs much longer spin-up time than Noah model. This presentation summarizes GLDAS workflow, its development and tests with NoahMP model including spin-up, preliminary evaluation of test results for uncoupled GLDAS and weakly-coupled GFS.v16.