Forums (Forecast Builder)

Back

Diurnal T Troubles

DB
David Barjenbruch, modified 7 Years ago.

Diurnal T Troubles

Youngling Posts: 16 Join Date: 9/2/14 Recent Posts

 

We had troubles with the blotchy up/down significant hourly T differences.  They varied greatly across grid points and through time.  Ran in FB using default diurnal, also occurred outside of FB using SuperBlend and National Blend.  Images attached (sorry, VLAB would not post in chronological order). 

Dave

 

 

JM
Joseph Moore, modified 7 Years ago.

RE: Diurnal T Troubles

Youngling Posts: 72 Join Date: 11/1/13 Recent Posts

The default ForecastBuilder code lists Superblend as the primary model and NationalBlend as secondary. However, since the last NationalBlend update (3.0), the model only comes in through so many hours depending on the run. For example, the 00z NBM goes through 264 hours but the 01z run only comes in through 18 hours - see attached graphic from the NBM Vlab. This is by design, and NIC 7.x has a fix wherein the "NationalBlend" model will vary in length but an "NBM" model will always go out through 10 days and update with the latest NationalBlend guidance at each time-step.  (The NIC 7.x update is apparently coming to CR by way of a tech note some time this month along with a FB update.)

I say all this because I wonder if the reason for this problem is that NationalBlend does not always come in through the same time range, and thus the dirunal curve that gets built is messed up? I have no idea if that's possible, but as a work-around I would recommend manually running Diurnal with Superblend selected for both primary and secondary models to see if that solves the issue.

 

 

JW
Jerry Wiedenfeld, modified 7 Years ago.

RE: Diurnal T Troubles

Youngling Posts: 4 Join Date: 11/21/12 Recent Posts
I believe there is a lot going on here with the diurnal tool.  There are known issues with that tool. With that said our latest version of FB will use the NBM singleton db to help with the cadence issues with that NationalBlend.

I'll let Andy Just add more to this but it could be a bit as he is transitioning to CRH this week.

Jerry

On Monday, October 16, 2017, VLab Notifications <VLab.Notifications@noaa.gov> wrote:

The default ForecastBuilder code lists Superblend as the primary model and NationalBlend as secondary. However, since the last NationalBlend update (3.0), the model only comes in through so many hours depending on the run. For example, the 00z NBM goes through 264 hours but the 01z run only comes in through 18 hours - see attached graphic from the NBM Vlab. This is by design, and NIC 7.x has a fix wherein the "NationalBlend" model will vary in length but an "NBM" model will always go out through 10 days and update with the latest NationalBlend guidance at each time-step.  (The NIC 7.x update is apparently coming to CR by way of a tech note some time this month along with a FB update.)

I say all this because I wonder if the reason for this problem is that NationalBlend does not always come in through the same time range, and thus the dirunal curve that gets built is messed up? I have no idea if that's possible, but as a work-around I would recommend manually running Diurnal with Superblend selected for both primary and secondary models to see if that solves the issue.

 

 


--
Virtual Lab Message Boards https://vlab.noaa.gov/web/cr-soo/forecast-builder-forum/-/message_boards/view_message/3262929 VLab.Notifications@noaa.gov http://vlab.noaa.gov


--
Jerry Wiedenfeld 
Information Technology Officer 
National Weather Service, Milwaukee/Sullivan, WI


RL
Robb Lawson, modified 7 Years ago.

RE: Diurnal T Troubles

Youngling Posts: 8 Join Date: 11/6/15 Recent Posts
Wanted to bump this post as we are still having issues with hourly temps across CWA borders. After the initial load, we seem to mesh well with some offices while others we are consistently off 3-6 degrees.  Would think while the initial load may not be correct, it would at least be consistent across CWA borders.  FWIW, max and mins load fine.