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

Re: Fwd: NCEP model GRIB records may have errors



Hi Geoffrey:

Thanks much for your explanation. We have done a review on all of the models that we get over the IDD, and appears that this issue only appears in the RUC2. We do see the "bucket dump" (we are calling it "mixed intervals") in a lot of models, for example in the GFS half-deg:

***Total_precipitation/Ground or water surface
all: (60) 0-3 0-6 6-9 6-12 12-15 12-18 18-21 18-24 24-27 24-30 30-33 30-36 36-39 36-42 42-45 42-48 48-51 48-54 54-57 54-60 60-63 60-66 66-69 66-72 72-75 72-78 78-81 78-84 84-87 84-90 90-93 90-96 96-99 96-102 102-105 102-108 108-111 108-114 114-117 114-120 120-123 120-126 126-129 126-132 132-135 132-138 138-141 138-144 144-147 144-150 150-153 150-156 156-159 156-162 162-165 162-168 168-171 168-174 174-177 174-180 Mixed

The issue appears in the RUC2 pressure output:

Open Q:\cdmUnitTest\tds\new\RUC2_CONUS_20km_pressure_20100509_1300.grib2
***Convective_precipitation/Ground or water surface
all: (46) 0-0 0-1 0-2 1-2 0-3 2-3 0-4 3-4 3-5 4-5 0-5 0-6 5-6 3-6 6-7 0-7 0-8 7-8 6-8 0-9 8-9 6-9 0-10 9-10 0-11 10-11 9-11 11-12 9-12 0-12 12-13 0-13 12-14 13-14 0-14 14-15 12-15 0-15 15-16 0-16 16-17 15-17 0-17 0-18 17-18 15-18 Mixed

***Large_scale_precipitation/Ground or water surface
all: (46) 0-0 0-1 1-2 0-2 2-3 0-3 3-4 0-4 3-5 4-5 0-5 0-6 5-6 3-6 6-7 0-7 7-8 0-8 6-8 8-9 6-9 0-9 9-10 0-10 10-11 9-11 0-11 9-12 0-12 11-12 12-13 0-13 12-14 0-14 13-14 12-15 0-15 14-15 15-16 0-16 15-17 0-17 16-17 17-18 0-18 15-18 Mixed

Now that we understand whats going on, we may just save the 1-hour accumulations, since the others can be derived if needed. Hopefully, this will be less confusing to our users.

Thanks again to you and Patrick for your help!

John Caron

On 6/8/2010 4:30 PM, address@hidden wrote:
    hi John-

         There are indeed 3 time accumulation periods for RUC precipitation.
    In a nutshell, we have 1-hr, 3-hr, and run total, but they're not quite
    that simple.

         The simplest thing to note is that we generate at each forecast
   hour the precip that has fallen within the past 1 hour.

         Also,the model "dumps" its precip bucket every 3 hours.
    As a result, we start by producing the fraction of the 3-hr bucket that
    has been accumulated.    In other words, if you are at a forecast hour
    divisible by 3 (03,06,09...), you get a 3-hr accumulation.   If you're
    at a forecast hour 02,05,08..  you get a 2-hr accumulation since that's
    what is in the bucket.   If you're at 01,04,07...  you would get a 1-hr
    accumulation since that's what is in the bucket, but you already get 1-hr
    accums, so we don't repeat it.   It's this feature that probably causes the
    most confusion, since it looks so different from hour to hour.

         Finally, you always get a "run total" which is how much precip has 
fallen
    since the start of the model.   So, at forecast hour 8, you get an 8-hr
    accumulation and so on.

         This is probably confusing, but we have users who want different 
things.
    Perhaps the 1-hr totals are the most logical records to save, since a user
    can always add them up on his/her own as needed.   Some users are
    concerned with hourly details about where precipitation will initiate
    and move, while others are only concerned about how much falls over
    a longer time period.

                  -Geoff


---- Forwarded message ----
Subject: Re: [DATAFLOW #KCY-33448-726] NCEP model GRIB records may
have errors
From: John Caron<address@hidden>
Date: Tue, 08 Jun 2010 12:33:16 -0600
To: "Patrick OReilly, NCEP Support"<address@hidden>

Hi Patrick:

Thanks much for your reply. I have figured out that these are all
the
same model parameter, differing by the time interval that they
cover.
The problem is that Im trying to turn the collection of 2D GRIB
records
into 3D (or 4D if there are vertical levels) "variables" (in netCDF
lingo), where the 3rd dimension is time, usually forecast hour. Im
looking for help on how to do that grouping.

So perhaps another way to ask my question is: how are these
different
time intervals used? Why are those particular ones output?

thanks again for any help,
John