Scheduled Downtime
On Friday 21 April 2023 @ 5pm MT, this website will be down for maintenance and expected to return online the morning of 24 April 2023 at the latest

Nudging

dyf

New member
Hello, I have been using WRF for a continuous period of four months. I have used ERA5 data as the initial field and boundary conditions. I have plotted a time series of surface rainfall in the watershed obtained by integration. Upon comparison, I have noticed that the simulated precipitation is more than twice the observed precipitation. I would like to reduce the simulated precipitation by nudging it to the observation level. I have attached a list of my "namelist.iput" and simulation results. As my domain is 4km resolution with a 150 x 150 grid, I think it would be more reasonable to use grid nudging.
Could you provide me with suggestions for setting up nudging parameters to lower my simulated precipitation?
Thanks a lot.
 

Attachments

  • Rainrate.png
    Rainrate.png
    20.5 KB · Views: 7
  • namelist.input
    4.1 KB · Views: 12
The overestimation of precipitation could be cased by many factors, including possible problems in ERA5 data. I have this impression that the ERA5 precipitation is not as good as expected. For your case, I would suggest that
(1) you may replace WSM6 by Thompson microphysics, and the latter is supposed to yield better results especially for high resolution run
(2) please use RRTMG for both shortwave and long wave radiation
(3) set radt = 4
(4) set hybrid_opt = 2 (need to rerun real.exe)
(5) set diff_opt = 2, which is more physically reasonable
(6) enlarge the model domain by increasing the grid numbers. The current domain is 600 x 600km, which is way small.
(7) if you want to check whether nudging can give you better result, then analysis nudging is a good option (grid_fdda = 1). For other options related to analysis nudging, please stay with the default option.
 
Thank you for your response, and the fact that I'm going to be integrating WRF for a couple of months is because I'm running Fully-Coupled WRF-hydro, which is to have the precipitation from WRF integration provides a hydrological model to simulate changes in runoff, I'm sorry I didn't make myself clear.
Since the hydrological model is sensitive to the total precipitation in the convective flow, I would like to ask if the total rainfall in the basin can be closer to the observation by adjusting the parameterization scheme, except nudging
At present, my parameterization scheme seems to capture the number of days of heavy rain. Still, the magnitude is twice as large as the observation, whether I can adjust the parameterization scheme to reduce the simulated precipitation, and thus close to the observation.
 
The performance of WRF is often case-dependent, and it is hard to tell which options/schemes can yield better results over a specific region/case.
You may have to run various tests and verify the results before you can figure out what options are the best.
 
Last edited:
Top