ESPC-D-V02 2026-08-17 12Z incomplete u3z/v3z forecast files

[Stephen Woodbridge]

Subject: ESPC-D-V02 2026-08-17 12Z incomplete u3z/v3z forecast files

Hello,

We are using the HYCOM ESPC-D-V02 forecast data from tds.hycom.org and have run into what looks like a partially published forecast cycle for the 2026-08-17 12Z run.

The run appears in the FMRC catalogs for all four variables we use:

• u3z

• v3z

• t3z

• s3z

However, when we check the underlying fileServer files directly, the temperature and salinity files are present, while most of the early U/V forecast files return HTTP 404.

For the exact set of files our processing requires through forecast hour 48, we see:

Temperature (t3z): 9/9 available
Salinity    (s3z): 9/9 available
U current   (u3z): 4/17 available
V current   (v3z): 4/17 available

Total: 26/52 available

For U/V, only these leads are currently available:

f039
f042
f045
f048

while:

f000 through f036

return HTTP 404.

For example, this 2026-08-17 12Z file currently returns 404:

https://tds.hycom.org/thredds/fileServer/datasets/ESPC-D-V02/data/forecasts/US058GCOM-OPSnce.espc-d-031-hycom_fcst_glby008_2026081712_t0000_u3z.nc

By comparison, the preceding 2026-08-16 12Z cycle is complete for the same manifest: all 52 required files are available.

The unusual part is that later U/V leads are present while the earlier leads are missing, so this does not look like a normal case of forecast files gradually appearing from f000 forward.

Could you let us know whether the 2026-08-17 12Z ESPC-D-V02 cycle is known to be incomplete or failed upstream, or whether the missing U/V files are expected to appear later?

We are mainly trying to determine whether applications should treat a cycle like this as permanently incomplete and continue using the previous complete forecast cycle.

Thank you,

Stephen Woodbridge

[Michael McDonald]

Stephen,

There will generally be a 12-24+ hour period when new forecast data is being acquired and integrated into the thredds catalogs (this is daily and unavoidable). During this time, there are rolling restarts of all the hycom data services (that require one) in order to update the data catalogs.

If you pick the penultimate forecast run, then you should not see missing dates as often, but those too can sometimes need a 3rd...4th (manual) transfer attempt when there is server maintenance being done at the Navy and primary data transfers get delayed/interrupted. 

e.g., as of writing this email on Aug 20th at 8:30 AM EDT we are still waiting on the t0111~t0192 forecast bundles for some of the 3z variables (as these are created last in the model run workflows at the Navy). So if you ask for ice, sur, ssh, salinity, temperature at a time in the upper t0111+ tau range that is when the ice and sur might be fulfilled, as those hourly tar bundles have already arrived, have been unpacked, and are updated in the system. Not the case for the larger 3z variables, which are still updating. Hope this helps. 

 $ /nexsan/people/pumps/scripts/hycom-espc-forecast-check-ALL.sh
[Thu 20 Aug 2026 07:34:10 AM EDT] t3z
=================================================================================
/hycom/ftp/datasets/ESPC-D-V02/data/forecasts
[  START  ] Mon 10 Aug 2026 12:00:00 PM UTC
[ MISSING ] 20260819 (t3z): t0111 t0114 t0117 t0120 t0123 t0126 t0129 t0132 t0135 t0138 t0141 t0144 t0147 t0150 t0153 t0156 t0159 t0162 t0165 t0168 t0171 t0174 t0177 t0180 t0183 t0186 t0189 t0192
[   END   ] Wed 19 Aug 2026 12:00:00 PM UTC
|  'Missing t3z'=28;1;10;;

[Thu 20 Aug 2026 07:34:15 AM EDT] s3z
=================================================================================
/hycom/ftp/datasets/ESPC-D-V02/data/forecasts
[  START  ] Mon 10 Aug 2026 12:00:00 PM UTC
[ MISSING ] 20260819 (s3z): t0111 t0114 t0117 t0120 t0123 t0126 t0129 t0132 t0135 t0138 t0141 t0144 t0147 t0150 t0153 t0156 t0159 t0162 t0165 t0168 t0171 t0174 t0177 t0180 t0183 t0186 t0189 t0192
[   END   ] Wed 19 Aug 2026 12:00:00 PM UTC
|  'Missing s3z'=28;1;10;;

[Thu 20 Aug 2026 07:34:20 AM EDT] u3z
=================================================================================
/hycom/ftp/datasets/ESPC-D-V02/data/forecasts
[  START  ] Mon 10 Aug 2026 12:00:00 PM UTC
[ MISSING ] 20260819 (u3z): t0111 t0114 t0117 t0120 t0123 t0126 t0129 t0132 t0135 t0138 t0141 t0144 t0147 t0150 t0153 t0156 t0159 t0162 t0165 t0168 t0171 t0174 t0177 t0180 t0183 t0186 t0189 t0192
[   END   ] Wed 19 Aug 2026 12:00:00 PM UTC
|  'Missing u3z'=28;1;10;;

[Thu 20 Aug 2026 07:34:25 AM EDT] v3z
=================================================================================
/hycom/ftp/datasets/ESPC-D-V02/data/forecasts
[  START  ] Mon 10 Aug 2026 12:00:00 PM UTC
[ MISSING ] 20260819 (v3z): t0111 t0114 t0117 t0120 t0123 t0126 t0129 t0132 t0135 t0138 t0141 t0144 t0147 t0150 t0153 t0156 t0159 t0162 t0165 t0168 t0171 t0174 t0177 t0180 t0183 t0186 t0189 t0192
[   END   ] Wed 19 Aug 2026 12:00:00 PM UTC
|  'Missing v3z'=28;1;10;;

--
--
You received this message because you are a member of HYCOM.org
To ask a question, send an email to fo...@hycom.org

--

Michael McDonald
HYCOM.org Administrator

[Knut-Frode Dagestad]

Hi,

I don't know whether it is related, but we are seeing an issue with the same dataset.

ncdump -h https://tds.hycom.org/thredds/dodsC/FMRC_ESPC-D-V02_uv3z/FMRC_ESPC-D-V02_uv3z_best.ncd | grep time

shows that there are now three time dimensions:
time = 121 ;
time1 = 121 ;
time_1 = 1 ;

and water_u and water_v have respectively time and time1, which contain the same values/times:

float water_u(time, depth, lat, lon) ;
water_u:coordinates = "time_run time depth lat lon " ;
float water_v(time1, depth, lat, lon) ;
water_v:coordinates = "time1_run time1 depth lat lon " ;