How to transiton land surface model improvements and value-added satellite data from university research to operation at NCEP/EMC?

In the past decade under JCSDA support, Zeng's team at the University of Arizona and the EMC Land Team (led by Ken Mitchell before, and by Ek now) have worked closely to accelerate the research-to-operation transition of land surface model (Noah) improvements and value-added satellite data. In this presentation, we will share our success and lessons learned through three related activities.

First, the NCEP global operational model (GFS) is found to have a significant cold bias in daytime surface skin temperature over global arid regions during the warm season. This issue also exists in NCAR's Community Earth System Model (CESM). Based on theoretical arguments and synthesis of previous observational and modeling efforts, three revisions are developed to address this issue (Zeng et al. 2012). The relevant improvement was implemented in GFS in 2011, improving the forecasting and increasing the assimilated infrared and microwave satellite data from land surface-sensitive channels (Zheng et al. 2012).

Second, the NCEP GFS has an early snowmelt problem. Based on NCEP land model (Noah) diagnostics and observational data analysis, we improved the Noah simulation of snow without changing the Noah model structure for easy implementation in GFS (Wang et al. 2010). GFS testing indicated the improvement in the prediction of snow processes, but its impact on other quantities (e.g., precipitation) is generally neutral (Wei et al. 2012). It is ready for operational implementation in GFS.

Recognizing that the maximum snow albedo data used in Noah are outdated, we developed the global 0.05-deg data over snow-covered land based on MODIS observations (Barlage et al. 2005). Even though this dataset is more realistic than that in Noah, its impact on Noah simulations and GFS forecasting was negative due to other deficiencies in Noah and GFS. Therefore it is ready but still has not been implemented in the operational GFS.