Hi,
I would like to use gridded precipitation data (e.g., from radar data or regionalized data with kriging and the like). If I understood it right I have to use netCDF file with gridded data. In the manual I read that I have to construct an nc file with, e.g., precipitation variable in three dimensions (lat, lon, time). Then I could add another variable with the same dimensionality and so on.
1. How shall I include the time definition (starting time and time step, e.g., if I have hourly data)?
2. Either way the unit of hourly precipitation data is mm/d?
3. Can I mix gridded and station based data?
4. Is it better to clip the nc file for speed according to my regional coverage? Or is the reading speed not dependent on the amount of lat/lon tupels in the nc file but only on the amount lat/lon tupels in my region of interest?
5. Is an example nc file and an according rvt file available for download (I did not find anything, yet)?
Thanks,
Michael
netCDF support
-
jrcraig
- Posts: 57
- Joined: Tue Jul 05, 2016 1:08 pm
Re: netCDF support
Hi Michael,
Hope this addresses your questions:
(1) The time definition should be absolute time in your time zone (i.e., corrected if the data was originally in UTC); the time step is inferred from the continuous time stamps in the .nc file; for the next version, it is planned to let Raven handle the time zone shift.
(2) yes - it is always rainfall intensity in mm/d, regardless of time step
(3) somwhat - if you have gridded precip and temperature or windspeed at stations, you can mix these. If you want to mix gridded and gauged precip, you need to do that via a preprocessing step
(4) Cllipping will always be faster and less memory intensive, but it will not impact Raven's computational speed - it only accesses the data linked to HRUs
(5) Alas, not in the standard distribution - if you contact the author via e-mail - contact info on the Raven site - I can probably share something with you.
Cheers.
Hope this addresses your questions:
(1) The time definition should be absolute time in your time zone (i.e., corrected if the data was originally in UTC); the time step is inferred from the continuous time stamps in the .nc file; for the next version, it is planned to let Raven handle the time zone shift.
(2) yes - it is always rainfall intensity in mm/d, regardless of time step
(3) somwhat - if you have gridded precip and temperature or windspeed at stations, you can mix these. If you want to mix gridded and gauged precip, you need to do that via a preprocessing step
(4) Cllipping will always be faster and less memory intensive, but it will not impact Raven's computational speed - it only accesses the data linked to HRUs
(5) Alas, not in the standard distribution - if you contact the author via e-mail - contact info on the Raven site - I can probably share something with you.
Cheers.
-
Michael
- Posts: 3
- Joined: Wed Feb 07, 2018 9:53 am
Re: netCDF support
Hi,
thank you a lot! The mix of gridded and gauged data for different observation data is exactly what I want to do. Perfect! Regarding to the example file, I write you an Email.
Cheers.
thank you a lot! The mix of gridded and gauged data for different observation data is exactly what I want to do. Perfect! Regarding to the example file, I write you an Email.
Cheers.