Temperature at 2m is 2-4 degrees lower than observation with WRF & NoahMP

[Ming Chen]

It seems that "snow" in ERA5 is overestimated. I will talk to our expert and get back to you ...

 

[Ming Chen]

The initiial value of snow is way too large. Please reduce it to a smaller value in wrfinput, then rerun this case, ---- I expect that the cold bias should decrease. Just for a test, you may set snow =0 and see how it affects the model behavior.

 

[xiao.zh]

The initiial value of snow is way too large. Please reduce it to a smaller value in wrfinput, then rerun this case, ---- I expect that the cold bias should decrease. Just for a test, you may set snow =0 and see how it affects the model behavior.

Hi Ming,
Thanks for your reply and sorry for my late response. I am busy with other works this month. I will run a test case later as you suggest and get back here when I get some result.

 

[xiao.zh]

The initiial value of snow is way too large. Please reduce it to a smaller value in wrfinput, then rerun this case, ---- I expect that the cold bias should decrease. Just for a test, you may set snow =0 and see how it affects the model behavior.

Hi Ming,
I tried a scenario with SNOW=0 (as well as SNOWC, SNOWH). The cold bias is relatively smaller (RMSE decreased from 2.40 to 2.29) but still exits.

 

[cenlinhe]

Is this over a snowpack region that cold bias occurred? If so, the snow would build up even if with 0 snow initial condition. could you please check SNOWC (snow cover)?

 

[Ming Chen]

Cold bias can also be attributed to unrealistically large cloudiness, overestimation of surface albedo, veg fraction, et al. How is the diurnal cycle of the bias?

 

[xiao.zh]

Is this over a snowpack region that cold bias occurred? If so, the snow would build up even if with 0 snow initial condition. could you please check SNOWC (snow cover)?

Hi, thanks for your reply.
This is the SNOWC and SNOWH in 2010-08-15. My simulation starts at 2010-01-01.
SNOWH should be 0 during the summer in my study area.
This scenario runs with WRF 4.7.0.

 

[xiao.zh]

Cold bias can also be attributed to unrealistically large cloudiness, overestimation of surface albedo, veg fraction, et al. How is the diurnal cycle of the bias?

Hi Ming,
I only preserved 24h's data for now. I may run a simulation with hourly output later.
This is my comparision in daily data. The peak and vally seems to have delays.

Do you think whether simulation with LAI and SST updated would help?

 

[Ming Chen]

I expect that the model should perform better with sst_update and time-varying LAI. Please try and let me know whether the results can be improved.

Thanks.