Questions About WPS/WRF Application

[Josh]

GOAL: Use High-Resolution Rapid Refresh (HRRR) model data as input to WRF to recreate initial atmospheric conditions, specifically for temperature and wind conditions at a specific location of interest.

• I am using hourly HRRR "0 hour forecast" analysis data as my meteorological data input to WPS.
• I am using the highest resolution static geography data provided by UCAR: Static Data Downloads
• I am using a nested domain: 3 km input met data --> 1 km resolution (child domain).
• I am not interested in making a forecast. I just want to have the best possible representation of the actual conditions that happened for some time in the past at each specific hour. The levels of interest would be between 50 and 2000 meters above ground level.

Questions:

1. I understand metgrid.exe's primary function is to perform horizontal interpolation of meteorological fields from the input datasets onto the model's computational grid. I understand real.exe performs vertical interpolation of the meteorological fields onto the requested levels, and generates the initial and boundary conditions, while wrf.exe generates the forecast. Since I am not interested in a forecast, and I am happy with the default levels provided in my input met data, do I really need to run real.exe and wrf.exe? What advantage would this give me compared to just using the output of metgrid for my analysis? During interpolation of variables such as wind speed, does metgrid adjust the wind speed depending on terrain features and land use characteristics or is this only really accounted for in WRF?
2. If you think it is reasonable to just run metgrid.exe, would I only need to enter a few key variables of interest to me for interpolation calculations or would this be inappropriate?
3. Do you think the static geography data I am using is a significant limiting factor if I am interested in 1 km resolution or is it "good enough".
4. Since I am only interested in a snapshot in time, and not a forecast, how small do you think my parent domain could be without causing significant issues in model stability/results?

Thanks for your time.

 

[kwerner]

The point of using the WRF model is to use it's functions (e.g. fine-tuned physics schemes) to get a higher-resolution and accurate solution. As you mentioned, through the metgrid program, the only data you have is available is the solution from the larger-scale (more coarse) input data (e.g., ERA5 or GFS). If your only intent is to see that data on a specific domain, then you can certainly just use the WPS program; however, keep in mind that you will just be using the exact data from the input model.