Procedure for gridding NEXRAD level II files
- Obtain level 2 files from the LDM. These files have no extention on the end. It will be something like:
- Use a script "runtrans" which sets up a bunch of environmental variables and runs the xltrsii translator that is available as part of the Soloii NCAR package. This script converts the level2 format to uf or dorade sweep files (specified under the OUTPUT_FLAGS setting). One very important thing to remember is that the reflectivity and velocity fields have different gate sizes (at the lowest 2 elevation angles. I believe they change the PRF in order to help detect second trip). I originally tried using a program called any_to_uf, but it does not account for the different DZ and VR gate sizes. However, the xltrsii translator will resample the VR field to be the same as the DZ field. This script will leave you with a file beginning with "ufd." for the uf file. I rename this back to KFTG20070610_234523.uf
--If you are interested in the Velocity data (specifically to run dual-Doppler in CEDRIC), we found that the level II files get labeled as 'SUR' (surveillance), but CEDRIC requires 'PPI'. Thus, run the changetoPPI.pl script.
- Setup a REORDER input file. There are a lot of options in REORDER, so this is a bare bones example. If you want to grid the data to AGL, put in the radar altitude as the grid altitude. If you want MSL, set the grid altitude to 0.
A few things I have learned by frustrating trial-and-error are:
- being sure to set the STOP date and time WAY past the end of your dataset
- making sure that you set the SCRATCH environmental variable. I do this using:
setenv SCRATCH .
Thus, I only have to set the SCRATCH once in a terminal window and I can change directories without REORDER complaining. Or you could set it to /home/$USER/tmp or something else as well. The output files will be written to the SCRATCH directory.
- You are only allowed to have a finite number of grid points. The documentation states a 500x500x500 grid, but there are other restrictions, and if I end up gridding too many variables to a very fine grid (such as all the polarimetric variables), it will run out of memory and complain.
- Be sure that log variables, such as DZ and DR, are specified with the 'DBZFIELD'. All other variables can go into the 'FIELD' specification.
- I have occasionally run into a problem where REORDER will only grid part of the data (like the first few levels). In this case, it fixed the problem to turn on 'SPANVOLUMES'. I think this is needed if you generate DORADE Sweep files, then combine them into a volume input file such as with swp2uf.
- REORDER will write out netcdf files or cedric binary files. We prefer to do the cedric binary files and then convert that to netcdf ourselves. It seemed like the netcdf files generated by REORDER weren't exactly what we were used to. But, to turn on netcdf output, you include: NETCDF: "/path/outfilename" under the INPUT line.
- To run REORDER, type the following:
qreou < reo.inp > reo.out
The cedric binary output will be in the form of ".ced". (If you are going to perform a dual-Doppler analysis, you can feed this file directly into CEDRIC).
- We then use a script called ced2nc (or ced2nc4swap) which writes the cedric binary file into a gridded netcdf. Note this requires the perl netcdf module to be installed.
Thanks to Brenda Dolan for documenting this.