In July 2001 a major upgrade to the Noah land-surface model (LSM) was implemented in the operational NCEP mesoscale Eta model. The Noah LSM improvements dealt largely with cold season processes (frozen soil and snowpack physics), but also included improvements related to bare soil evaporation (reduction of excessive early-spring surface evaporation), soil heat flux under vegetation (reduction of excessive heat flux through vegetation canopies), and minor changes to the canopy conductance formulations.

More recently, the Noah LSM has been 'unified' among several operational and research groups (i.e. NCEP, AFWA, NCAR, and UCLA), with further upgrades to the unified Noah LSM. These upgrades include additional changes to the calculation of surface fluxes for patchy snow conditions. The soil heat flux now explicitly includes contributions from both the snow and non-snow covered portions of a model gridbox. As such, soil heat flux generally increases due to the contribution of the soil heat flux from the non-snow covered surface, allowing a stronger soil-atmosphere transfer, e.g. such that a cold bias in the nighttime 2-meter air temperature over snow was reduced, especially for warmer and moister soil conditions where the conduction of heat from below is increased. (This change was implemented in the operational mesoscale Eta model in February 2002.) Additionally, latent heat flux over patchy snow conditions now also explicitly accounts for the flux from both snow and non-snow covered portions of a model grid box. As such, latent heat flux generally decreases due to a less-than-potential evaporation (sublimation) from the non-snow covered surface which allows snow cover to persist longer (all other conditions being equal).

Finally, the green vegetation fraction is used in the Noah LSM to determine the portion of a model gridbox where vegetation actively contributes to surface evaporation. Currently in the operational Eta model the green vegetation fraction comes from a high resolution (0.144 deg) global data base based on a 5-year climatology. The implementation of this data base was superior to previous lower-resolution approaches, but can still suffer from not representing the current state, e.g. the green vegetation fraction may differ from climatology due to less (or more) optimal growing conditions at a particular time for the current year. We will present results of testing in the coupled setting of the Eta Analysis and Forecast System using a near-realtime weekly green vegetation data base and assess model performance.