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

WRF for Wind Speed Downscaling

This post was from a previous version of the WRF&MPAS-A Support Forum. New replies have been disabled and if you have follow up questions related to this post, then please start a new thread from the forum home page.

johnphua

New member
I am using hourly wind data from ERA5 at a resolution of 0.25°x0.25°, and did a successful run on WRF with default parameters.
I have some questions regarding how to improve the run and run it more efficiently.

Besides setting sst_update, is there anything else to watch out for when running for a time range of more than a year? ERA5 dataset begins at 1979 – I are trying to run it for all 40 years. Is there a maximum number of years we shouldn't run beyond? For either technical or meteorological reasons?

What is the maximum recommended downscaling resolution? The papers I have read usually downscale ERA5 to 11km x 11km. Can I downscale to 3km x 3km or worse 1km x 1km ? My understanding is that beyond some number, I am supposed to use WASP.

Can ungrib be sped up by running separate ungrib processes on different time segments (e.g. month by month) and combining back into a full time period? Would this create issues later?

When you choose a geographic bounding box to run WRF on, how much padding does WRF take outside of the box? Because I am assuming the edges of the bounding box are affected by conditions outside of the box.

Thank you.
 
Hi,

I have an unrelated question. You said you want to run WRF for 40 years using ERA5. Do you already download all the ERA5 data? I am also downscaling EAR5 in a long-term base, but I have problems downloading them from Climate Data Store using cdsapi. The download is too slow and sometimes the download script does not work. I just want to know if you also have this problem and how do you solve it?

Thank you
 
To answer your question for downscaling wind by WRF,
(1) sst_update is the necessary option for climate simulation. I don't think there is any other option that is critical for long term run.
(2) There is no limitation how many years you can run WRF. 40 years of simulation certainly should work.
(3) It is fine to downscale ERA5 to 3km WRF run. However, for 1-km WRF run, I would suggest running in nesting case (e.g., 3km-1km nesting).
(4) ungrib can run to process data on different time segments.
 
have you tried to download the ERA5 data from NCAR'S CISL website:
https://rda.ucar.edu/datasets/ds630.0/#!description

However, I understand that this website only has data that covers the period of 2003-2018
 
Thank you for the clarification.

It is fine to downscale ERA5 to 3km WRF run. However, for 1-km WRF run, I would suggest running in nesting case (e.g., 3km-1km nesting).
Regarding this point, may I ask how nesting helps to allow me to run on a finer resolution?
At 1km resolution, does WRF factor in local topography for wind data (mountainous regions, hills and valleys)? How does it do so (as ERA5 does not provide elevation data)?

The download is too slow and sometimes the download script does not work. I just want to know if you also have this problem and how do you solve it?
Yes, my data is also obtained from Climate Data Store using cdsapi and I did also face the same troubles downloading

To resolve the slow download speed, I wrote a script that sent out the requests in parallel, so that I could also download files in parallel and maximise the bandwidth (running on a cloud instance). However, this also resulted in more frequent occurrence of files that were corrupted.
To resolve the corruption of files / downloads that did not work, I re-ran the requests for the affected files sequentially.

Overall, the download process took a day or 2 for just 1 year of data, and I also made the following observations:
1) Requests were approved more frequently when sent on the cloud instance compared to my local machine.
2) For the same request size (variables, number of days), the file sizes are about the same and can be used as a preliminary check for file corruption or halted downloads
 
When running at higher resolution, WPS will provide more detailed description of surface features that will be used later in the run.
Please see the documents here: http://www2.mmm.ucar.edu/wrf/users/tutorial/201901/werner_nesting.pdf
 
Dear all,
I am a complete newbie in WRF but I need to simulate a long time series of wind data at a particular location in a complex orography starting from ERA5 wind data (1979-2021) at the nearest node and I thought about doing with WRF. Can someone point me out a tutorial? I will be running WRF on a MacOS. Is there a Docker image that I can use for this purpose?
Many many thanks in advance,
Letizia
 
Letizia,
Please take a look at the online tutorial, which will give you a complete picture of running WRF.

https://www2.mmm.ucar.edu/wrf/OnLineTutorial/

Tutorial presentation can be found at https://www2.mmm.ucar.edu/wrf/users/tutorial/tutorial.html

Specifically, WRF has a nice capability to output time series at any location inside the model domain. Please read the document "run/README.tslist"
 
Top