Grid Spacing vs. Resolution

If you spend enough time with meteorologist, you'll inevitably hear criticism against using the non-interchangeable terms "grid spacing" and "resolution" to describe the spacing between grid points in a numerical weather model. Just last week I reviewed a paper where one of the other reviewers blasted the authors for incorrectly using the terms interchangeably. I cringed at the comment. What drives me nuts about this is that the reviewer didn't even explain why the terms are not interchangeable. The supposed incorrect use of the term model "resolution" in place of "grid spacing" will go on forever because those who are deeply offended by those who interchange the term rarely explain how the two terms are different.  

What is the difference?

In a March 2000 BAMS commentary, Lewis D. Grasso explains: 

The two terms  grid  spacing and resolution refer to two different length scales that characterize a grid configuration. Because of this, they are not interchangeable in linear and nonlinear numerical  models. It is not possible to resolve a wave on the scale of one grid spacing in any spatial direction.  Since 2Δx and 3Δx waves are removed to prevent nonlinear instability, waves on the scale of at least 4Δx may be resolved. Other terms that may be used in place of grid spacing are grid interval, grid length, and grid increment. 

There you have it; resolution is not the correct term to describe a model's grid spacing because resolution refers to the length scale that waves are resolved in the model.

More resources on the topic

• https://weather.mailasail.com/Franks-Weather/Grid-Length-Resolution
• https://www.meted.ucar.edu/nwp/model_structure/navmenu.php?tab=1&page=3-0-0

Documentation for my Packages

  I finally got around to consolidating some of my more useful Python code and generated Sphinx documentation.

Please check out the following packages:

• GOES-2-go: 
• Download GOES-16 or GOES-17 data from Amazon Web Services (ABI and GLM data).
• RGB recipes for ABI images (Natural Color, Air Mass, Water Vapor, etc.)
• SynopticPy:
• Retrieve mesonet data from the Synoptic API
• HRRR-B:
• Download HRRR data from the HRRR archives at the University of Utah's Pando archive, Google Cloud Platform, and Amazon Web Services.

Pygrib now on Windows via anaconda conda-forge

I was very excited to hear this news that pygrib is now available on Windows...

So there are now (FINALLY) Windows builds of PyGrib available from @condaforge. You can thank @wxchick10 for nerd-sniping me today.

Actually, a lot of thanks should go to @ECMWF for getting eccodes to build on windows.

I love my job.

— Ryan May (@dopplershift) November 15, 2019

I gave it a try, and it seems to open HRRR grib2 files just fine. This means my HRRR_Pando scripts will work on a Windows machine now. Happy day!

How to download specific variables from HRRR GRIB2 files from the Pando archive

Don't have much time, but wanted to paste in a reply to an email concerning downloading specific variables from the HRRR archive on Pando.

The person wanted to download a few variables from files on the archive, rather than the full file. They also wanted to get data from a specific latitude/longitude location in the model grid.

-------------

Unfortunately, it isn't possible to pluck out specific lat/lon locations from the grid before you download the data. The best thing to do would be to download the full grid for the variables you want.

I have described how to do this on the "Scripting Tips" page under the "cURL and wget" and "Python" tabs.

http://home.chpc.utah.edu/~u0553130/Brian_Blaylock/hrrr_script_tips.html

The approach here is that by knowing the byte range for the grid of a single variable (as found in the .idx file), you can download just that chunk from the file (~1 MB) rather than the full file (>125 MB).

An example .idx file for the model analysis (F00) valid at 0000 UTC on 10 July 2019 is here: https://pando-rgw01.chpc.utah.edu/hrrr/sfc/20190710/hrrr.t00z.wrfsfcf00.grib2.idx

This file corresponds to the full GRIB2 file here: https://pando-rgw01.chpc.utah.edu/hrrr/sfc/20190710/hrrr.t00z.wrfsfcf00.grib2

In this example, if you want surface temperature, you would search the .idx file for the line with the variable TMP:surface

The grid for this variable begins on byte 31889605 and the next variable begins on byte 33367664.

Thus, you can download just that grid, which returns a valid GRIB2 file with one layer, with a curl command:

curl -o 20190710_00zf00_TMPsurface.grib2 --range 31889605-33367664 https://pando-rgw01.chpc.utah.edu/hrrr/sfc/20190710/hrrr.t00z.wrfsfcf00.grib2

Because we don't archive the HRRR on the native grids we don't have the data you need at specific model levels. But we do archive the surface file forecasts (F00-F18) and the pressure file analyses (F00) that interpolate the data on specific levels. These still might be useful for what you want to do.

You can browse the variable names and descriptions here: (though, these pages might not be a complete list)

Surface Files: http://home.chpc.utah.edu/~u0553130/Brian_Blaylock/HRRR_archive/hrrr_sfc_table_f00-f01.html

Pressure Files: http://home.chpc.utah.edu/~u0553130/Brian_Blaylock/HRRR_archive/hrrr_prs_table_f00-f01.html

I listed the variables you might be interested in below:

-geopotential (at surface and at first level of model)

HGT:surface

(no HGT at first model level)

-wind speeds (u and v components at first level)

UGRD:10 m

VGRD:10 m

UGRD:80 m

VGRD:80 m

WIND:surface This is the maximum wind speed at 10 m for the previous hour

GUST:surface This is essentially the maximum wind speed in the lowest XX layers (not sure exactly how gust is calculated)

(again, wind is not available at first model level)

-snow depth

SNOD:surface

SNOWC:surface (snow cover)

-temperature (at surface and at first level)

TMP:surface

TMP:2 m

(Not available at first level)

-solar irradiance

DSWRF:surface (downward shortwave radiation flux)

-relative humidity

RH:2 m

-cloud fraction

TCDC:entire atmosphere (total cloud cover)

-hourly precipitation amount

This is a tricky one because precip is an accumulated quantity, thus there isn't a value for the analysis (F00). Instead, you would have to look at accumulated precipitation over some period of time. Some people like to look at the precip accumulated during the F00-F01 period as the "best guess"

For example:

APCP:surface:0-2 hour acc fcst Is the accumulated precip between F00-F02

APCP:surface:1-2 hour acc fcst Is the accumulated precip between F01-F02

If you target the specific variables you want, you will end up downloading much less data than downloading the full files. This will likely save a lot of download time and bandwidth (on your end and our end). It would probably be best to execute a curl command for each variable separately (and for each hour of the day).

I often have to pluck out values at specific lat/lon locations, too. My method is available on github, but there are probably many other ways to do the same thing.

https://github.com/blaylockbk/pyBKB_v3/blob/master/demo/Nearest_lat-lon_Grid.ipynb

I hope this helps!

Brian Blaylock

Ph.D. Candidate

Atmospheric Science

University of Utah

RAP and HRRR model domain boundaries

I made a neat map of the RAP (version 4) and HRRR model domain boundaries with Cartopy. I think it looks nice...