J1.4
Reducing the wintertime warm bias in NCAR GCMs through the use of a new snow cover fraction scheme

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner
Monday, 30 January 2006: 2:15 PM
Reducing the wintertime warm bias in NCAR GCMs through the use of a new snow cover fraction scheme
A313 (Georgia World Congress Center)
Zong-Liang Yang, Univ. of Texas, Austin, TX; and G. Y. Niu

The wintertime warm bias over central Eurasia has been a problem in several recent versions of climate models developed by the National Center for Atmospheric Research (NCAR), including CCM3/LSM, CAM2/CLM2, and CAM3/CLM3. By comparing the modeled wintertime surface albedos with those produced from the Moderate Resolution Imaging Spectroradiometer (MODIS), we found that the modeled albedos are too low, which is consistent with the fact that the modeled snow cover fraction (SCF) is too low. A series of studies has been focused on sensitivity tests using different forms of SCF. The SCF scheme of Yang et al. (1997) largely reduced the wintertime warm bias, but it produced a cold bias during the melting season in late spring. We developed a different SCF scheme for the melting season based on watershed observations. The SCF scheme appeared to reduce the wintertime warm bias over snow covered areas. However, the scheme caused spurious oscillations in the simulated surface temperature at sub-daily to daily time scales during the melting season. We proposed a new SCF scheme, in which the SCF in melting season is represented by snow density. This scheme ensured a smooth transition from accumulation to melting periods. The new SCF scheme largely reduces the warm bias during December-January-February (DJF) produced by the original NCAR GCM (i.e., CAM2/CLM2). The March-April-May (MAM) warm bias is reduced as well. The new SCF scheme largely improves the snow depth simulations in mid-latitude regions such as Eurasia (40-55N, 50-85E) and North America (40-55N, 95-110W), where the topography is relatively flat and low vegetations (grassland and croplands) dominate. The new SCF scheme also improves simulations of precipitation in DJF and MAM. The root mean square error (RMSE) of precipitation with the new scheme decreases in northern China and west North America, which is consistent with temperature improvements in these regions. Additional numerical experiments are planned to fully assess the impacts of this new parameterization on the simulations of snow depth, air temperature, and runoff. Topography will also be incorporated in the new scheme.