Hi,
I am running a real-case simulation of a tornadic event in Manitoba, Canada. We performed our simulation with different PBL and surface layer schemes to see which scheme(s) would give us the best results based on convection timing/spatial distribution/structure, surface condition, and sounding verifications. Thus far, we have found that the simulation with QNSE PBL (with the daytime EDMF option- 'mfshconv', turned on) and surface layer schemes gives us the best results for our area of interest (southern Manitoba). However, we noticed that about 300 km to the west and southwest of our study area (which is roughly over North Dakota), some locations experience a rapid increase in the 2-m air temperature between 23 UTC - 00 UTC of the next day (6-7 pm LT). I have attached a PDF document demonstrating this problem. As a reference, our study area is at the center of the domain shown.
As you can see, the simulated 2-m air temperature field looks normal even at 22:40 UTC. However, beginning at 22:50 UTC, some small, isolated spots start to experience significant warming. The area experiencing rapid warming expands over time, with the maximum simulated air temperature increasing from 306 to 315 K (33 to 42 deg. Celsius) in 30 minutes! I checked the observed surface air temperature over North Dakota during that day, but I didn't see any reports higher than 36 deg. Celsius. Of course, the really high temperatures may have been missed in the observation due to the sparsity of the observational network, but I just feel that the timing and the spatial pattern of the temperature increase look strange, given that 6-7 pm LT is already well past the peak solar heating hours. Some other 2D fields such as HFX, LH, QFX, and PBLH also exhibit this feature.
The simulation with the daytime EDMF turned off produces a similar problem. Other simulations with different PBL schemes (e.g., YSU, MYJ, and MYNN) do not show this behavior, but each has its own disadvantage when comparing to the observed conditions over Manitoba. I wonder if this is normal, or there is a bug in either the QNSE PBL or its surface layer scheme, or like I have read at some places that the QNSE schemes only work in stably-stratified regimes. If the latter, I think there should be a brief mentioning of this in the WRF User Manual.
I have also attached the namelist.input for this QNSE simulation. Any insights would be greatly appreciated.
Happy New Year,
Chun-Chih
I am running a real-case simulation of a tornadic event in Manitoba, Canada. We performed our simulation with different PBL and surface layer schemes to see which scheme(s) would give us the best results based on convection timing/spatial distribution/structure, surface condition, and sounding verifications. Thus far, we have found that the simulation with QNSE PBL (with the daytime EDMF option- 'mfshconv', turned on) and surface layer schemes gives us the best results for our area of interest (southern Manitoba). However, we noticed that about 300 km to the west and southwest of our study area (which is roughly over North Dakota), some locations experience a rapid increase in the 2-m air temperature between 23 UTC - 00 UTC of the next day (6-7 pm LT). I have attached a PDF document demonstrating this problem. As a reference, our study area is at the center of the domain shown.
As you can see, the simulated 2-m air temperature field looks normal even at 22:40 UTC. However, beginning at 22:50 UTC, some small, isolated spots start to experience significant warming. The area experiencing rapid warming expands over time, with the maximum simulated air temperature increasing from 306 to 315 K (33 to 42 deg. Celsius) in 30 minutes! I checked the observed surface air temperature over North Dakota during that day, but I didn't see any reports higher than 36 deg. Celsius. Of course, the really high temperatures may have been missed in the observation due to the sparsity of the observational network, but I just feel that the timing and the spatial pattern of the temperature increase look strange, given that 6-7 pm LT is already well past the peak solar heating hours. Some other 2D fields such as HFX, LH, QFX, and PBLH also exhibit this feature.
The simulation with the daytime EDMF turned off produces a similar problem. Other simulations with different PBL schemes (e.g., YSU, MYJ, and MYNN) do not show this behavior, but each has its own disadvantage when comparing to the observed conditions over Manitoba. I wonder if this is normal, or there is a bug in either the QNSE PBL or its surface layer scheme, or like I have read at some places that the QNSE schemes only work in stably-stratified regimes. If the latter, I think there should be a brief mentioning of this in the WRF User Manual.
I have also attached the namelist.input for this QNSE simulation. Any insights would be greatly appreciated.
Happy New Year,
Chun-Chih