# HRRR variable documentation

**URL:** https://community.drivendata.org/t/hrrr-variable-documentation/7070
**Category:** Snowcast Showdown
**Created:** [January 14, 2022, 2:55pm UTC](https://community.drivendata.org/t/hrrr-variable-documentation/7070 "2022-01-14T14:55:40Z")
**Posts on this page:** 1
**Page:** 1

<div class="post-metadata">

### Author: ![simon.jegou](https://avatars.discourse-cdn.com/v4/letter/s/439d5e/32.png) [@simon.jegou](https://community.drivendata.org/u/simon.jegou)
#### Post date: [January 14, 2022, 2:55pm UTC](https://community.drivendata.org/t/hrrr-variable-documentation/7070/1 "2022-01-14T14:55:40Z")

</div>

Hello,

I struggle a bit to understand the exact definition of the variables accessible through the products of HRRR. I am using [Herbie](https://github.com/blaylockbk/Herbie) to retrieve the HRRR data, which is very convenient to retrieve all HRRR data since 2014, and fits the requirements of this challenge using the setting `priority='azure'`.

HRRR variables are listed [here](https://www.nco.ncep.noaa.gov/pmb/products/hrrr/hrrr.t00z.wrfsfcf00.grib2.shtml), and as it has been already noticed, it already contains a SWE variable (variable 068, WEASD). I compared this variable with the provided SNOTEL data and obtained the following correlations:

 ![image](https://canada1.discourse-cdn.com/flex028/uploads/drivendata1/original/2X/3/3168d4c9c55e1b9b317d935baf728929eacc4aec.jpeg)

We can notice a high correlation (around 0.80) with HRRRv2 released in 2016. The correlation is good in the first months (~december to february) but decreases in the melting season. The corresponding RMSE cumulates all these errors and thus increase over time before decreasing when the SWE levels go to zero. I guess this observation motivates the challenge overall 🙂

Back to my variable question, the documentation is very sparse for newcomers. For instance, how should I retrieve the total snowfall (ASNOW) or total precipitation (APCP) between 2 dates ?

So far if I want the accumulated snowfall from 12:00 to 13:00 on the 1st of january, I download the file at 12:00 and look at the forecast for 1h. Using Herbie: `Herbie('2022-01-01 12:00', product='sfc', fxx=1).xarray('ASNOW')`. Is it the good way to proceed ?

If you know any documentation on HRRR data it would be very welcome,  
Simon