SWF Analysis of LMRFC Hourly QPE (http) vs NCEP Stage 4 Mosaic

[adamscarberry]

Original email from SWF on 31-July-2024:

All,

We’ve compared the gridded precipitation data from difference sources to ensure they are similar across the products of cumulus and this is what we found for our district (SWF). We currently use Cumulus – NCEP Stage 4 Mosaic for forecasting and daily operation. Our district area is covered by both West Gulf RFC (WGRFC) and Lower Mississippi RFC (LMRFC).

We’ve compared data of 4 events this year for the Red River Basin (LMRFC) to evaluate the difference among the products. Here are the products we’ve compared:

1. Cumulus – NCEP Stage 4 Mosaic QPE Precip 1hr
2. Cumulus – LMRFC QPE Precip 1hr
3. LMRFC XMRG – Retrieved from LDM and https://tgftp.nws.noaa.gov/data/rfc/lmrfc/xmrg_qpe/, then processed with MetVue Grid reader

We used LMRFC XMRG as the baseline and compared both Cumulus product against it. The results are attached with this email. It seems like Cumulus – NCEP Stage 4 Mosaic QPE has very similar result with LMRFC XMRG when comparing basin averages from gridded precipitation. However, we noticed a significant difference in basin average when compare these results with the Cumulus – LMRFC QPE Precip 1hr.

We’ve did the similar comparison on the Trinity River Basin using the West Gulf RFC (WGRFC) XMRG(LDM and https://tgftp.nws.noaa.gov/data/rfc/wgrfc/xmrg_qpe/) and Cumulus – NCEP Stage 4 Mosaic QPE Precip 1hr , and the comparison results are similar. We are aware that WGRFC had issues sending over a format other than XMRG to cumulus from past discussions, but this does not impact our operation since we mainly utilize NCEP Stage 4 Mosaic product from Cumulus for forecasting and daily operations.

Since this does not impact our operation, we just wanted to inform you of our little discovery of the data discrepancy.

Please feel free to contact us if you would like any further research from us.

SWF-CumulusPrecipComparison.pdf

Thanks, Chris

[adamscarberry]

From James:

Hi Eric,

We were under the impression that all the RFC precip data was "stitched" together and it formed the national gridded data set. Is this correct?

Using the national gridded data has met our needs and has been confirmed after verifying this data against the gridded data we get from our local RFCs.

Thanks, James

[adamscarberry]

James,

Correct the NCEP Stage 4 Mosaic QPE Precip 1hr stitches together the data from the RFCs into a single grid. NOAA does the stitching and then makes it available. Some districts still like to use the file directly from the RFCs instead of using the national grid. Analysis like the one you did helps the case that the national grid is displaying the same data and those districts would be able to use the NCEP data instead.

We believe the difference that you are seeing between the LMRFC data in Cumulus and the data you grabbed directly is that cumulus only grabs the grid once. So if the LMRFC makes an update, and then overwrites the existing grid in the location we are grabbing it from, and saves it with the same filename cumulus doesn’t got back and grab the dataset again. Cumulus is grabbing the data from the same site you did https://tgftp.nws.noaa.gov/data/rfc/lmrfc/xmrg_qpe/. We are just grabbing the grb.gz file instead of the xmrg.

In term of NCEP data it is one of the only DAGs that we go back and grab the data for. Every time we go back 5 day and look for updated grids and grab the new files, process and overwrite what is being stored in cumulus. So in your analysis the LMRFC data that is in cumulus is likely the original file. Whereas the data in the file your grabbed directly from the http site and the NCEP data contain updates that the RFC might have applied after their original analysis.

Eric