Forums

Back

Cold temperature bias at Yosemite National Park

RSS (Opens New Window)(Opens New Window)
JB
Jeffrey Barlow, modified 6 Years ago.

Cold temperature bias at Yosemite National Park

Youngling Posts: 4 Join Date: 3/24/16 Recent Posts

We have been consistently seeing 8-16 degree cold Max temperatures compared to OBS Database. 

Jeffrey Craven, modified 6 Years ago.

RE: Cold temperature bias at Yosemite National Park

Youngling Posts: 90 Join Date: 9/24/12 Recent Posts
Hello Jeff.  I obviously have a keen interest in this because of the impact of the URMA on NBM.   I took a look and this is not an easy problem to fix.   

The valley is rather narrow, and there appear to be observations both in the valley floor and also on the ridges either side of the valley.  So looking at 20F differences in MaxT as a result.   Elevation differences in this short distance for the observations range from 2,000 to 8,000 feet MSL.  

I drew a black line on the approximate location of the valley/Merced River.  MatchObsAll in this plot appears to be offset about one to two grid points to the south of the actual location (not sure if this is a real problem or just an issue with NBM Viewer).  

I looked at the current plot (around 1100 UTC) this morning and there is a 50F spread in observations across a short distance.  Omitting the 13F observation from a HADS (hopefully this is blacklisted), there still appears to be roughly a 25F difference in observations close to the Merced River Valley in this area.   The HADs (AHIC1) appears to be accepted by the URMA analysis.  



Here is a recent example showing URMA, URMA First Guess, MatchObsAll, and the difference between MOA and URMA.  



The URMA FG in this case for AHIC1 grid point was raised 4F to account for that observation. 

Here is another day, where an observed 2,000 foot observation downstream on Merced River was 101F, and the URMA analyzed about 98F.   The AHIC1 HADS site was 87F, after being raised 5F from the first Guess.  



JPC

Jeff Craven
Chief, Statistical Modeling Branch
National Weather Service, W/STI-12
Meteorological Development Laboratory (MDL)
Room 10410, SSMC2
Silver Spring, MD 20910
(301) 427-9475 office
(816) 506-9783 cell/text
@jpcstorm

On Thu, Jul 5, 2018 at 4:26 AM, VLab Notifications <VLab.Notifications@noaa.gov> wrote:

We have been consistently seeing 8-16 degree cold Max temperatures compared to OBS Database. 


--
Jeffrey Barlow RTMA/URMA Discussion Group Virtual Lab Forum https://vlab.noaa.gov/web/715073/discussions-forums-/-/message_boards/view_message/4373192VLab.Notifications@noaa.gov

Attachments:

Attachment

Attachment

Attachment

U
Anonymous, modified 6 Years ago.

RE: Cold temperature bias at Yosemite National Park

Jeff,

Looking at the output it does appear that there is some kind of projection issue, at least in the Yosemite Valley with MatchAllObs. It appears like the grids are shifted one box down. That being said, even with the projection corrected there still appears to be a 5-10 degree discrepancy between the two across the Yosemite Valley grids. Of course, this isn't the only area we are seeing discrepancies across the Sierra based on point observations. Down around Shaver Lake/Huntington Lake the morning lows via the RTMA appear to be too warm, 10-20 degrees at times. Possibly due to the RTMA not taking the lake into account? These are smaller lakes, but high elevation snow melt lakes. Though not as bad of a discrepancy this looks to be happening across the high Sierra in other spots. We all know observations are scarce in this region, though those we do have are several degrees cooler than the RTMA. Not a big deal as we focus on these areas a lot closer when there is an impact, say early season snow across Tioga Pass. 

 

Thanks for your attention on this.

 

Jerald

JC
Jacob Carley, modified 6 Years ago.

RE: Cold temperature bias at Yosemite National Park

Youngling Posts: 69 Join Date: 12/17/14 Recent Posts
Hello,

Sorry for taking so long to respond.  We (the RTMA group) have been busy with code finalization for v2.7 as well as with the recent holiday.

In this upcoming upgrade (v2.7) we've got a couple changes in place to help address this issue.  Unfortunately the parallel RTMA/URMA system has not been running reliably due to ongoing WCOSS dev issues over the past several weeks.  Once the machine is back running regularly you'll see graphics here:


I've also shared here the science briefing which explains a bit more (see slides 14-17):
So you should see an improvement in the next version, which is planned to be implemented in early October.

Thanks,
Jacob


On Fri, Jul 6, 2018 at 11:00 AM, VLab Notifications <VLab.Notifications@noaa.gov> wrote:

Jeff,

Looking at the output it does appear that there is some kind of projection issue, at least in the Yosemite Valley with MatchAllObs. It appears like the grids are shifted one box down. That being said, even with the projection corrected there still appears to be a 5-10 degree discrepancy between the two across the Yosemite Valley grids. Of course, this isn't the only area we are seeing discrepancies across the Sierra based on point observations. Down around Shaver Lake/Huntington Lake the morning lows via the RTMA appear to be too warm, 10-20 degrees at times. Possibly due to the RTMA not taking the lake into account? These are smaller lakes, but high elevation snow melt lakes. Though not as bad of a discrepancy this looks to be happening across the high Sierra in other spots. We all know observations are scarce in this region, though those we do have are several degrees cooler than the RTMA. Not a big deal as we focus on these areas a lot closer when there is an impact, say early season snow across Tioga Pass. 

 

Thanks for your attention on this.

 

Jerald


--
Anonymous RTMA/URMA Discussion Group Virtual Lab Forum https://vlab.noaa.gov/web/715073/home/-/message_boards/view_message/4380775 VLab.Notifications@noaa.gov

JB
Jeffrey Barlow, modified 6 Years ago.

RE: Cold temperature bias at Yosemite National Park

Youngling Posts: 4 Join Date: 3/24/16 Recent Posts

Im writing to inform you that we seeing a warm temperature bias of 8-10 degrees at Tuolumne Meadows indicated in the images. .The RTMA is not as accurate as the Match all Obs data.

Attachments:

Attachment

Attachment

JB
Jeffrey Barlow, modified 6 Years ago.

RE: Cold temperature bias at Yosemite National Park

Youngling Posts: 4 Join Date: 3/24/16 Recent Posts

Also, does the RTMA use the SFORTPHNX data for initialization?

JC
Jacob Carley, modified 6 Years ago.

RE: Cold temperature bias at Yosemite National Park

Youngling Posts: 69 Join Date: 12/17/14 Recent Posts
Hi Jeff B.,

Could you tell me a little bit more about the images you've provided?  Are they both RTMA?  I'm having a little trouble locating labels/filenames indicating which is RTMA and MatchObs.   Are the numbers shown obs or grid values?  And can you tell us the date/cycle time so we can look a bit more deeply?

I know we're talking about an 8-10 deg difference here, and we'll look into it.  But it is also worth noting that the RTMA/URMA is not intended to fit observations exactly.  The RTMA/URMA analysis reflects a statistical combination of a background forecast and observations weighted by their respective errors.  This accounting for observational error, as well as considerations for the analysis grid-spacing and the desire to avoid introducing sub-grid scale non-meteorological features (e.g. via aliasing), generally means that the analysis will not fit the observations exactly - especially if the background forecast differs quite a bit from the observation in question.  Of course there are situations where the ob should have been fit a bit better and those are good cases to look into.

Thanks!
Jacob


On Mon, Sep 24, 2018 at 7:19 PM VLab Notifications <VLab.Notifications@noaa.gov> wrote:

Also, does the RTMA use the SFORTPHNX data for initialization?


--
Jeffrey Barlow RTMA/URMA Discussion Group Virtual Lab Forum https://vlab.noaa.gov/web/715073/home/-/message_boards/view_message/4924230 VLab.Notifications@noaa.gov
JB
Jeffrey Barlow, modified 6 Years ago.

RE: Cold temperature bias at Yosemite National Park

Youngling Posts: 4 Join Date: 3/24/16 Recent Posts

Jacob,

 

Sorry it took so long to reply. The top image is the Match Obs All and the bottom image is the RTMA. I understand your reply and that makes sense. I will continue to provide examples we are seeing in our CWA.

 

Jeff

SL
Steven Levine, modified 6 Years ago.

RE: Cold temperature bias at Yosemite National Park

Youngling Posts: 174 Join Date: 11/13/14 Recent Posts
Jeffrey,

I assume you are talking about this when referring to SFORTPHNX data.  We do use all of the obs mentioned in this file, we use all METAR data and most of the 'mesonet' data we get from MADIS.  I should note that over the past couple days, the ASOS at Edwards AFB has had intermittent outages for some reason.

You can find more information about obs used in the system (and request that certain obs not be used) here on our VLab community site.

As Jacob mentions, RTMA will not match obs in the way MatchObsAll does, but that does not mean that the obs are not being used.  A new algorithm to go into operations later this fall (and available for viewing here as URMA27Exp) should give much more weight to METAR obs in complex terrain (such as these).  If you can tell us more about where and when your images were taken, we can run comparisons for you.

Steve

On Mon, Sep 24, 2018 at 7:19 PM VLab Notifications <VLab.Notifications@noaa.gov> wrote:

Also, does the RTMA use the SFORTPHNX data for initialization?


--
Jeffrey Barlow RTMA/URMA Discussion Group Virtual Lab Forum https://vlab.noaa.gov/web/715073/home/-/message_boards/view_message/4924230 VLab.Notifications@noaa.gov