I'm experiencing systematic cold bias of the night temperatures over high mountain terrain. The temperatures at night seem to always have cold bias by 2-4 degrees while the day temperatures are either fine or have very little bias (+1/-1 degree). From what I see - the higher the terrain, the worse the bias. Actually up to 1500m above sea level the night temperatures seem to be fine.
I'm fixing the temperature to the real location altitude using the geopotential height and estimating the temperature vertical lapse rate (e.g. very similar to what meteoadriatic have suggested in this topic - viewtopic.php?f=4&t=469&start=10).
An example location having the bias is Musala peak - http://www.windyweek.com/spots/bulgaria-samokov-musala (the highest peak of the Balkans peninsula).
Has anyone experienced that problem before or do you guys have any suggestion what can be changed to at least reduce the bias?
Thanks in advance for any information.
WRF EMS 3.4.1 using ARW core, adaptive timestep, initial and boundary conditions - GFS 0.25 degree. Outer (6km) and one-way nest domain (2km) (the bias seen for both domains). namelist.input excerpt:
Code: Select all
cu_physics = 0, 0 cudt = 0, 0 mp_physics = 5, 5 .... bl_pbl_physics = 1, 1 bldt = 0, 0 topo_wind = 0, 1 sf_sfclay_physics = 1, 1 .... sf_surface_physics = 2, 2 num_land_cat = 21 num_soil_cat = 16 num_soil_layers = 4 .... ra_lw_physics = 1, 1 ra_sw_physics = 2, 2 radt = 6, 6 ra_call_offset = 0 slope_rad = 0, 0 topo_shading = 0, 0 icloud = 0 co2tf = 1 sst_skin = 1 sst_update = 0 seaice_threshold = 271 fractional_seaice = 0