NOUS41 KWBC 041930
Technical Implementation Notice 12-42
National Weather Service Headquarters Washington DC
330 PM EDT Tue Sep 4 2012
-Family of Services
-NOAA Weather Wire Service
-Emergency Managers Weather Information Network
-Other NWS Partners, Users and Employees
From: Tim McClung
Science Plans Branch Chief
Office of Science and Technology
Subject: Change in Land Surface Model in Global Forecast
System and Associated Cool and Moist Bias in Near
Surface Temperature and Moisture Fields.
Beginning with the 1200 Coordinated Universal Time (UTC) model run on Wednesday, September 5, 2012, the National Centers for Environmental Prediction (NCEP) Central Operations will implement a fix to the Land Surface Model used to drive the Global Forecast System (GFS). This correction is expected to improve the cool and moist bias in the near surface air temperature and moisture fields during the warm season.
Starting in mid-June 2012, NCEP confirmed a problem with the GFS near-surface temperature and moisture simulations. The model was not verifying in the late afternoon over the central United States when drought conditions existed. Specifically, users noted a significant 2m cold and wet bias in both the MOS and GFS gridded products. The NCEP Environmental Modeling Center (EMC) traced the problem to a look-up table used in the land surface scheme that modulates evapotranspiration based on vegetation type and root zone depth. Current settings allowed for excessive transpiration and plant-extraction of soil moisture from deeper soil layers, which caused the lower atmospheric boundary layer to become too moist and cool.
This summer, EMC conducted a parallel test of the GFS with corrected land surface parameters. EMC found the change to the land surface model significantly reduced the cold/wet bias over and improved the 0-3 day precipitation forecasts. An analysis of the parallel test results can be viewed here:
In addition, the Meteorological Development Lab (MDL) evaluated the GFS MOS guidance generated from the GFS parallel output. MDL concluded the corrections in the GFS parallel output had the desired effect of reducing the cold/wet bias resulting in improved bias and error scores in the GFS MOS temperature and dewpoint guidance. Slight improvement was also noted in the GFS MOS probability of precipitation guidance generated from the GFS parallel output. A summary of MDL’s evaluation and comparisons of GFS MOS text bulletins generated from the GFS parallel output can be viewed at:
NCEP and MDL also conducted a 1.5 month GFS parallel evaluation from Jan-Feb 2012 to test the land surface model correction during the U.S. cold season. As expected, analysis of results indicated the change in the land surface model had little impact on the simulated 2m temperature and dewpoint given the low solar forcing during the winter months. MDL’s analysis of the MOS guidance from the GFS parallel during the cool season sample also indicated minimal impact on the 2m temperature and dewpoint.
Based on the positive results of the GFS parallel tests and the severity of the cold and wet bias in the near surface air temperature, NCEP will implement the correction to the land surface model beginning with the 1200 UTC cycle on Sep. 5, 2012. This change requires no action by users to continue receiving data.
NCEP thanks the users who alerted us to this problem. For questions regarding the correction to the GFS Land Surface Model and resulting improvement to the near surface cold and moist bias please contact:
NCEP EMC/College Park, Maryland
For questions regarding the impacts on the GFS MOS text and gridded products please contact:
MDL/Silver Spring, Maryland
301-713-0023, Ext. 130
For questions regarding the impacts to the GFS LAMP text and gridded products please contact:
MDL/Silver Spring, Maryland
301-713-0056, Ext. 194
Links to the MOS products and descriptions are online at:
Descriptions of previous and upcoming GFS implementations can be found at:
NWS national TINS are online at: