[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[IDV #QEL-679004]: Use case for time matching, HRRR, etc...

> Yuan:
> 
> OK.  It took me a night's sleep, but I guess one can either grab
> the "Best Time Series" option or the "Latest Collection." I'm guessing
> the Latest Collection is the most recent model run? That seems to be the
> case when I load it up.  However, I'm still unable to do what I would
> like.  Here's the issue.
> 
> What I would like is observed reflectivity for part of the loop and
> then HRRR reflectivity for the other part (i.e., the future).  I would
> also like to overlay HRRR analyses on top of the observed reflectivity.
> I can do this if I set my time driver through the Time Animation
> Properties Widget and I can get it to work if I time match the HRRR Best
> Time Series to the widget. However, when I do this, it displays the HRRR
> reflectivity along with the observed reflectivity, which I don't want.

Jim,
      How about this kind of design: setup the driver time through the 
animation widget as past 12 hour to
future 18 hour like before. Then, use this driver time to time match the real 
observed radar reflectivity
from the adde server, you will only load the observed part. Again, use the 
driver time to load the 
reflectivity field from HRRR latest dataset. The result will be closed to what 
you want, you may see
one time step observed overlay from both datasets.

Yuan
> 
> I've tried to get around this in the attached bundle by loading the
> HRRR Latest Collection, which I use to plot just the reflectivity in the
> forecast.  Then, I loaded the Best Time Series in the hopes of plotting
> the vector winds on top of both the observed reflectivities and the HRRR
> forecast reflectivities. However, when I do this, it only plots wind up
> to the period that the HRRR Latest Collection is available for.
> 
> I suppose what I am doing here is a bit unusual, but it is
> something very useful for teaching nowcasting.  Basically, I'm trying to
> blend an image/analysis loop (in this case, radar reflectivity with HRRR
> analyses) seemlessly into a forecast loop (in this case HRRR
> reflectivity and analyses).  I'd actually like to do this with a number
> of my bundles as it would make class much easier to teach than what I'm
> doing now, which is loading up analysis and forecast bundles separately.
> 
> Jim
> 
> 
> On 2/9/15 4:52 PM, Jim Steenburgh wrote:
> > Attached is a screenshot showing it is missing.  I am running 5.0u3
> > from about a week ago.  Is it something you added recently?
> >
> > Jim
> >
> >
> >
> > On 2/9/15 4:48 PM, Unidata IDV Support wrote:
> >>> Oh yeah, what ever happened to the most recent model run option on the
> >>> THREDDS server.  There were some advantages of this over Best Time
> >>> Series for some display options.  For this particular bundle, for
> >>> example, one might want to not plot the HRRR reflectivity until the
> >>> forecast period in order to not conflict with the observed
> >>> reflectivity.
> >>>
> >>> Jim
> >> Most recent run (or latest rung) should be available in the list. If
> >> not, you should update to the latest nightly release.
> >>
> >>
> >> Yuan
> >>>
> >>> On 2/9/15 4:06 PM, Unidata IDV Support wrote:
> >>>>> IDVers:
> >>>>>
> >>>>> I have something I would like to do with IDV and I don't know if it
> >>>>> can be done with the current IDV.  Perhaps it can, but it is not
> >>>>> immediately obvious.
> >>>> Jim,
> >>>>         Yes, it can be done in the current IDV.
> >>>>> What I would like is a loop extending from 12 hours into the past
> >>>>> to 12 (or 18) hours into the future that seemlessly blends *observed*
> >>>>> radar reflectivity, HRRR *analyzed* surface wind vectors (and
> >>>>> possibly
> >>>>> other fields), and HRRR forecasts of radar reflectivity and
> >>>>> surface wind
> >>>>> vectors.  The idea is to have a real-time short-range forecast
> >>>>> tool for
> >>>>> precipitation the blends the observed/analyzed past with the future
> >>>>> forecast.  The bundle would need to grab past HRRR analyses and the
> >>>>> latest HRRR forecast.  It would probably be best if it were done with
> >>>>> hourly increments for performance reasons.
> >>>>>
> >>>>> Of course, we need the HRRR for this and I don't think it is
> >>>>> presently available on motherlode (hopefully it is coming).  In it's
> >>>>> place, we could use the RAP as a test case.
> >>>> HRRR is available in this catalog:
> >>>>
> >>>> http://thredds.ucar.edu/thredds/catalog.xml
> >>>>
> >>>> There may be some minor changes in the catalog, but the best time
> >>>> series should be consistent.
> >>>>
> >>>>> So, I sat down to do this today with the RAP and it wasn't
> >>>>> immediately obvious how to do it and it highlights, potentially,
> >>>>> some of
> >>>>> the limitations of the IDV, the need for more user friendliness with
> >>>>> regards to time matching, or the need for some improvements with
> >>>>> THREDDS.
> >>>>>
> >>>>> For example, I thought I might be able to specify the from 12 hours
> >>>>> into the past to 12 hours into the future using the time animation
> >>>>> properties widget, but I don't think this is possible. Either the
> >>>>> start
> >>>>> or end time needs to be Current time for a real-time bundle.  An
> >>>>> alternative would be to use the maximum time for all data and grab
> >>>>> the
> >>>>> latest HRRR/RAP, but if one does that, you need to use the somewhat
> >>>>> obtuse "reverse times" options with the best time series option.
> >>>>> This
> >>>>> gets messy very quickly.  I've done it for many of my bundles, but
> >>>>> it's
> >>>>> very hard to setup right.  We need to do better.  BTW, whatever
> >>>>> happened
> >>>>> to the latest model run option on the motherload THREDDS? All
> >>>>> there is
> >>>>> now seems to be best time series.
> >>>>>
> >>>> The only possible way is to use the time animation widget to set
> >>>> the driver time, you can specify past 12 hours
> >>>> to future 18 hours by selecting both start and end time as current,
> >>>> and then set the offset values
> >>>> to be -12 for past and 18 for future.
> >>>>
> >>>>> I've been thinking of making this capability in most of my
> >>>>> bundles.  It would be cool to have analyses and forecasts blended
> >>>>> together.  Presumably there is an easy way to do this, but I'm not
> >>>>> sure
> >>>>> how if one starts mixing in model forecasts, satellite data, radar
> >>>>> data,
> >>>>> etc.
> >>>>>
> >>>> Time driver feature is the best fit for this kind of requirement.
> >>>>
> >>>>> Let me know if I'm clueless or if we have a topic worth of
> >>>>> addressing.  Oh yeah, to beat a dead horse, this brings up the
> >>>>> omniprescent need to be able to identify forecast vs. analysis hours
> >>>>> with the model grids.
> >>>>>
> >>>>> Jim
> >>>>>
> >>>> Let me know if you need more help.
> >>>>
> >>>>
> >>>> Yuan
> >>>>>
> >>>> Ticket Details
> >>>> ===================
> >>>> Ticket ID: QEL-679004
> >>>> Department: Support IDV
> >>>> Priority: Normal
> >>>> Status: Open
> >>>>
> >>>
> >>
> >> Ticket Details
> >> ===================
> >> Ticket ID: QEL-679004
> >> Department: Support IDV
> >> Priority: Normal
> >> Status: Open
> >>
> >
> 
> 
> 


Ticket Details
===================
Ticket ID: QEL-679004
Department: Support IDV
Priority: Normal
Status: Open