WRF Noah MP Nighttime Cold Bias over Snow

Dec 2021 to Jan 2022 cases

1. 12-km Sea-Tac Soundings at hour 2
2. see table 2 below to plot geographic plots and some simple statistics

   differences from wrfgfs are indicated with *asterisk

   Case Name	Description	Full namelist
   3. enshrrr	HRRR physics: Thompson Aerosol Aware microphysics*, RRTMG LW and SW, radt=15*, MYNN sfc layer*, Noah LSM, MYNN 2.5-level TKE PBL*, no mosaic landuse in Noah*, no shallow cu*, no FDDA*, diff_opt=1*, diff_6th_factor=0.25*, diff_6th_opt=2*, tke_adv_opt=1*	namelist.input.enshrrr
   4. enswrf4mynn	Thompson microphysics, RRTMG LW and SW, Noah LSM, MYNN sfc layer*, no shallow cu*, tke_adv_opt=1*	namelist.input.enswrf4mynn
   5a. gfsn	same as wrfgfs except Noah LSM for all runs Dec 2021 through Jan 2022	namelist.input.gfsn
   5b. gfsnmp	same as wrfgfs except Noah-MP LSM* for all runs Dec 2021 through Jan 2022	namelist.input.gfsnmp
   5c. gfsruc	same as wrfgfs except RUC LSM* for all runs Dec 2021 through Jan 2022	namelist.input.gfsruc
   5d. gfsruc3	same as hn3 except RUC LSM* for all runs Dec 2021 through Jan 2022 snow physics: Thompson Aerosol Aware microphysics*, RRTMG LW, Goddard SW*, radt=30, Old MM5 surface layer*, RUC LSM*, MYNN 2.5-level TKE PBL*, mosaic landuse and soil, no shallow cu, FDDA and real-time dynamics settings: diff_opt=2, diff_6th_otp=0, tke_adv_opt=0	namelist.input.gfsruc3
   6. hn2	same as enshrrr except: Goddard shortwave radiation*, mosaic landuse in Noah, and real-time dynamics settings: diff_opt=2, diff_6th_otp=0, tke_adv_opt=0	namelist.input.hn2
   6b. hn3	snow physics: Thompson Aerosol Aware microphysics*, RRTMG LW, Goddard SW*, radt=30, Old MM5 surface layer*, Noah LSM, MYNN 2.5-level TKE PBL*, mosaic landuse in Noah, no shallow cu, FDDA and real-time dynamics settings: diff_opt=2, diff_6th_otp=0, tke_adv_opt=0	namelist.input.hn3
   6c. hn3c273	exact same as hn3 except remove snow from wrfinput where top soil temp >= 273.15K exact same as hnmp3s3c273 except Noah vs Noah-MP LSM snow physics: Thompson Aerosol Aware microphysics*, RRTMG LW, Goddard SW*, radt=30, Old MM5 surface layer*, Noah LSM, MYNN 2.5-level TKE PBL*, mosaic landuse in Noah, no shallow cu, FDDA and real-time dynamics settings: diff_opt=2, diff_6th_otp=0, tke_adv_opt=0	namelist.input.hn3c273
   7. hnmp2	exactly like hn2 except Noah-MP LSM*	namelist.input.hnmp2
   8. hnmp3	exact physics as hnmp2 except Noah-MP code mods wrf.exe.nmpmods1: increase minimum snow-depths from from 0 and 5 cm to 15 cm	namelist.input.hnmp3 module_sf_noahmplsm.F.nmpmods1 compare to module_sf_noahmplsm.F.orig
   9. hnmp3s1	like hnmp3 used code wrf.exe.npmmods1, except Thompson microphysic,RRTM LW*, Dudhia SW*, radt=15*, Old MM5 surface layer, YSU PBL, FDDA	namelist.input.hnmp3s1
   10. hnmp3s2	like hnmp3 used code wrf.exe.npmmods1 except radt=30, Old MM5 surface layer*, FDDA best snow physics: Thompson Aerosol Aware microphysics*, RRTMG LW, Goddard SW*, radt=30, Old MM5 surface layer*, Noah-MP LSM, MYNN 2.5-level TKE PBL*, mosaic landuse in Noah, no shallow cu, FDDA and real-time dynamics settings: diff_opt=2, diff_6th_otp=0, tke_adv_opt=0 identical physics to hn3 except Noah-MP instead of Noah LSM	namelist.input.hnmp3s2
   11. hnmp3s3	exact same physics as hnmp3s2 except used code wrf.exe.npmmods2 which uses Verseghy 1991 thermal conductivity of snow calculation though constrained to be <= 0.25 best snow physics: Thompson Aerosol Aware microphysics*, RRTMG LW, Goddard SW*, radt=30, Old MM5 surface layer*, Noah-MP LSM, MYNN 2.5-level TKE PBL*, mosaic landuse in Noah, no shallow cu, FDDA and real-time dynamics settings: diff_opt=2, diff_6th_otp=0, tke_adv_opt=0	namelist.input.hnmp3s3
   11a. hnmp3s3c273	exact same physics as hnmp3s3 except removed snow from wrfinput files where top soil temp >= 273.15K best snow physics: Thompson Aerosol Aware microphysics*, RRTMG LW, Goddard SW*, radt=30, Old MM5 surface layer*, Noah-MP LSM, MYNN 2.5-level TKE PBL*, mosaic landuse in Noah, no shallow cu, FDDA and real-time dynamics settings: diff_opt=2, diff_6th_otp=0, tke_adv_opt=0	namelist.input.hnmp3s3
   11b. hnmp3s3myn	exact same physics and code as hnmp3s3 except used MYNN surface layer instead of old MM5	namelist.input.hnmp3s3myn
   12. hnmp3s4	exact same physics as hnmp3s2 except used code wrf.exe.npmmods3 which 1) increases minimum snow-depths from from 0 and 5 cm to 25 cm 2) uses Verseghy 1991 thermal conductivity of snow calculation though constrained to be <= 0.15	namelist.input.hnmp3s4
   13. hnmp	exact same physics as hnmp2 and hnmp3 except with RRTMG SW instead of Goddard	namelist.input.hnmp
   2. mm5eta	NAM initialization, MM5 3.6.1 code with simple Dudhia thermal diffusion surface layer, Kain-Fritsch cumulus, Dudhia simple ice microphysics, MRF PBL, Dudhia radiation, multil-layer soil model, no shallow convection	mmlif
   14. simple	Thompson microphysics, RRTM LW*, Dudhia SW*, radt=30, Old MM5 surface layer, thermal diffusion land-surface, YSU PBL, shallow cu	namelist.input.simple
   1. wrfgfs	real-time configuration, Noah-MP through Dec 18, 2021, Noah LSM from 2021121900 onward physics: Thompson microphysics, RRTMG LW and SW, radt=30, revised MM5 Monin-Obukhov sfc layer, Noah or Noah-MP LSM, YSU PBL, shallow cu,	namelist.input.wrfgfs

   notes from plots

   1. Goddard SW does improve 2-m temperature (RMS for hnmp2 (Goddard SW) vs hnmp (RRTMG sw)
   2. Noah-MP can be improved (mostly in afternoon) by using MYNN sfc layer and MYNN 2.5-level TKE PBL (RMS for gfsnmp vs hnmp)
   3. So far, I believe from other tests that Noah-MP is still desired, but would recommend switching, at least for the winter, to Goddard SW, MYNN sfc, MYNN 2.5-level TKE PBL
3. All Stats for Dec 2021 - Jan 2022 for the cases highlighted above, all fields, both hours: Mean Absolute Error, Mean Error
4. Geographic loops for Dec 2021 - Jan 2022 for the cases highlighted above: a. t2m MAE all sites 00Z b. t2m ME all sites 00Z c. t2m MAE SA sites 00Z d. t2m ME SA sites 00Z
5. Geographic loops for Dec 2021 - Jan 2022 for the cases highlighted above: a. t2m MAE all sites 12Z b. t2m ME all sites 12Z c. t2m MAE SA sites 12Z d. t2m ME SA sites 12Z
6. 12-km Sea-Tac Soundings at hour 12
7. 12-km Sea-Tac Soundings at all times
8. Ground heat flux conus3 case (left side) vs conus3nmp case (right side) at 30 minutes into the simulations range -200 to +200
9. Ground heat flux for Noah vs Noah-MP (optstc2 case) with range -20 to +20

 conus3nmp         = same as conus3 but with Noah-MP instead
 conus3            = Noah + Thompson + RRTMG + MYNN* + Tiedtke
                     mynn_tkeavect, mynn_cloudpdf=2, mynn_mixlength=2,
                     mynn_edmf=2, mynn_edmf_tke=1, mynn_mixscalars=1
 conus2            = Noah + Thompson + RRTMG + MYNN + Tiedtke
 conus             = Noah + Thompson + RRTMG + MYJ + Tiedtke
                     this is the CONUS physics suite
 gfssfc            = Noah-MP opt_stc=2 + Thompson + RRTMG + GFS sfclay + GFS PBL
 noah              = Noah + Thompson + RRTMG
 optstc2           = Noah-MP opt_stc=2 + Thompson + RRTMG
 optstc2wsm        = Noah-MP opt_stc=2 + WSM-6 + RRTMG
 optstc2wsmgoddard = Noah-MP opt_stc=2 + WSM-6 + Goddard radiation

Older Cases

• Vis 1km Satellite, IR Satellite + surface obs, IR Satellite + 500mb obs, EOS Woldview Image

• Ensemble Plumes
• KSEA table data at hour 12, KSEA table data at hour 15, KSEA table data at hour 24
• KOLM table data at hour 12, KOLM table data at hour 15, KOLM table data at hour 24
• 00.12 t2m ME Puget Sound plots