# JEDI-MPAS Error with Thompson Tables

**URL:** <https://forums.jcsda.org/t/jedi-mpas-error-with-thompson-tables/738>\
**Category:** Uncategorized\
**Created:** [April 13, 2026, 9:48pm UTC](https://forums.jcsda.org/t/jedi-mpas-error-with-thompson-tables/738 "2026-04-13T21:48:52Z")\
**Posts on this page:** 5\
**Page:** 1

<div class="post-metadata">

**Author:** ![amercer](https://avatars.discourse-cdn.com/v4/letter/a/7cd45c/32.png) [@amercer](https://forums.jcsda.org/u/amercer)\
**Post date:** [April 13, 2026, 9:48pm UTC](https://forums.jcsda.org/t/jedi-mpas-error-with-thompson-tables/738/1 "2026-04-13T21:48:52Z")

</div>

I am new to MPAS-JEDI and am trying to assimilate Nexrad Level II data to a single time in an existing MPAS regional simulation. I have successfully generated the appropriately formatted ioda file (I think), as well as the appropriate yaml file for JEDI. However, when I go to run the mpasjedi\_variational.x executable against this yaml I get an error saying  
"At line 867 of file /mercer/williams/mpas-bundle/MPAS/src/core\_atmosphere/physics/physics\_wrf/module\_mp\_thompson.F (unit = 101, file = ‘MP\_THOMPSON\_QRacrQG\_DATA.DBL’)  
Fortran runtime error: End of file  
"

I have verified that this table (as well the others generated by mpas\_atmosphere\_buildtables) are present in my working directory and my previous MPAS run that did not do any data assimilation worked fine with these tables. It is unclear why this is failing with JEDI when it worked with a vanilla MPAS run without any DA. I’ve included my .yaml and namelist/stream files below. Everything has been run in a single directory. I am using a build of mpas-bundle, if that helps. What could be causing this?

namelist.atmosphere:

&nhyd\_model  
config\_time\_integration\_order = 2  
config\_dt = 3  
config\_start\_time = ‘2025-01-05\_18:00:00’  
config\_run\_duration = ‘0\_00:15:00’  
config\_len\_disp=9000.0  
config\_split\_dynamics\_transport = true  
config\_number\_of\_sub\_steps = 2  
config\_dynamics\_split\_steps = 3  
config\_horiz\_mixing = ‘2d\_smagorinsky’  
config\_visc4\_2dsmag = 0.05  
config\_scalar\_advection = true  
config\_monotonic = true  
config\_coef\_3rd\_order = 0.25  
config\_epssm = 0.1  
config\_smdiv = 0.1  
/  
&limited\_area  
config\_apply\_lbcs=true  
/  
&damping  
config\_zd = 22000.0  
config\_xnutr = 0.2  
/  
&io  
config\_pio\_num\_iotasks = 0  
config\_pio\_stride = 1  
config\_restart\_timestamp\_name = ‘restart’  
/  
&decomposition  
config\_block\_decomp\_file\_prefix = ‘MS\_Regional.graph.info.part.’  
/  
&restart  
config\_do\_restart = false  
/  
&printout  
config\_print\_global\_minmax\_vel = true  
config\_print\_detailed\_minmax\_vel = false  
/  
&IAU  
config\_IAU\_option = ‘off’  
config\_IAU\_window\_length\_s = 21600.  
/  
&physics  
config\_sst\_update = false  
config\_sstdiurn\_update = false  
config\_deepsoiltemp\_update = false  
config\_radtlw\_interval = ‘00:10:00’  
config\_radtsw\_interval = ‘00:10:00’  
config\_bucket\_update = ‘none’  
config\_gwdo\_scheme = ‘off’  
config\_physics\_suite=‘convection\_permitting’  
/  
&soundings  
config\_sounding\_interval = ‘none’  
/

streams.atmosphere:  
  
\<immutable\_stream name=“input”  
type=“input”  
filename\_template=“mpas\_ics.nc”  
input\_interval=“initial\_only” /\>

\<immutable\_stream name=“restart”  
type=“input;output”  
filename\_template=“restart.$Y-$M-$D\_$h.$m.$s.nc”  
input\_interval = “initial\_only”  
output\_interval=“0:15:00” /\>

\<immutable\_stream name=“iau”  
type=“input”  
filename\_template=“x1.40962.AmB.$Y-$M-$D\_$h.$m.$s.nc”  
filename\_interval=“none”  
packages=“iau”  
input\_interval=“initial\_only” /\>

\<immutable\_stream name=“lbc\_in”  
type=“input”  
filename\_template=“lbc.$Y-$M-$D\_$h.$m.$s.nc”  
filename\_interval=“input\_interval”  
packages=“limited\_area”  
input\_interval=“3:00:00” /\>

radar.yaml

cost function:  
cost type: 3D-Var  
analysis variables: [u, v, w, theta, qv, qc, qr]  
time window:  
begin: 2025-01-05T18:05:00Z  
length: PT20M  
geometry:  
model: MPAS  
nml\_file: /mercer/williams/MPAS-Model/namelist.atmosphere  
streams\_file: /mercer/williams/MPAS-Model/streams.atmosphere  
background:  
filename: restart.2025-01-05\_18:15:00.nc  
date: 2025-01-05T18:15:00Z  
background error:  
covariance model: SABER  
saber central block:  
saber blocks:

- covariance model: BUMP  
bump:  
io:  
data directory: ./bump  
files prefix: mpas\_b  
drivers:  
multivariate strategy: univariate  
compute covariance: true  
modeling:  
horizontal length-scale: 500000.0  
vertical length-scale: 2000.0  
observations:
- obs space:  
name: KGWX radial velocity  
obsdatain:  
engine:  
type: H5File  
obsfile: /mercer/williams/jedi/kgwx\_velocity\_ioda.nc  
obsdataout:  
engine:  
type: H5File  
obsfile: /mercer/williams/jedi/kgwx\_velocity\_ioda\_out.nc  
simulated variables: [radialVelocity]

obs operator:  
name: RadarRadialVelocity  
variables: [wind]  
options:  
vertical coordinate: height  
use\_reflectivity: false

obs error:  
covariance model: diagonal  
diagonal:  
value: 25.0

obs filters:

- filter: Domain Check  
where:
- variable: height  
minvalue: 0.0  
maxvalue: 12000.0
- filter: Bounds Check  
where:
- variable: radialVelocity  
minvalue: -80.0  
maxvalue: 80.0

---

<div class="post-metadata">

**Author:** ![byoung-joo](https://yyz1.discourse-cdn.com/flex031/user_avatar/forums.jcsda.org/byoung-joo/32/126_2.png) [@byoung-joo](https://forums.jcsda.org/u/byoung-joo)\
**Post date:** [April 14, 2026, 12:05am UTC](https://forums.jcsda.org/t/jedi-mpas-error-with-thompson-tables/738/2 "2026-04-14T00:05:55Z")

</div>

Thanks for your interest in using mpas-jedi, @amercer .

I wonder if your mpas-bundle build is successful. Have you check if all (or most) mpas-jedi ctests pass with your build ?

Also, the following environmental setting is needed in mpas-jedi (but not for the standalone MPAS Model build).

**gcc compiler**  
export GFORTRAN\_CONVERT\_UNIT='big\_endian:101-200"  
or  
setenv GFORTRAN\_CONVERT\_UNIT ‘big\_endian:101-200’

**intel compiler**  
export F\_UFMTENDIAN=‘big\_endian:101-200’  
or  
setenv F\_UFMTENDIAN ‘big\_endian:101-200’

BJ

---

<div class="post-metadata">

**Author:** ![amercer](https://avatars.discourse-cdn.com/v4/letter/a/7cd45c/32.png) [@amercer](https://forums.jcsda.org/u/amercer)\
**Post date:** [April 15, 2026, 4:09pm UTC](https://forums.jcsda.org/t/jedi-mpas-error-with-thompson-tables/738/3 "2026-04-15T16:09:50Z")

</div>

That gives me something to check. I am not running JEDI on Derecho but on a custom server at my institution, but I have the full mpas-jedi-env spack-stack build running successfully. I will try to rebuild with this option enabled and run through the ctests to see how things turn out and report back. Thanks!

---

<div class="post-metadata">

**Author:** ![byoung-joo](https://yyz1.discourse-cdn.com/flex031/user_avatar/forums.jcsda.org/byoung-joo/32/126_2.png) [@byoung-joo](https://forums.jcsda.org/u/byoung-joo)\
**Post date:** [April 15, 2026, 4:59pm UTC](https://forums.jcsda.org/t/jedi-mpas-error-with-thompson-tables/738/4 "2026-04-15T16:59:34Z")

</div>

For additional information:

Setting GFORTRAN\_CONVERT\_UNIT or F\_UFMTENDIAN is only required at “run time”, not for building the code. So, you may not need to rebuild the code.

Also they are _automatically_ applied when we run mpas-jedi ctest.

---

<div class="post-metadata">

**Author:** ![amercer](https://avatars.discourse-cdn.com/v4/letter/a/7cd45c/32.png) [@amercer](https://forums.jcsda.org/u/amercer)\
**Post date:** [April 17, 2026, 7:29pm UTC](https://forums.jcsda.org/t/jedi-mpas-error-with-thompson-tables/738/5 "2026-04-17T19:29:53Z")

</div>

Thanks for the reply. I ended up having to rebuild the mpas-bundle as I forgot to enable lfs within git. After that, it passed all but about 10 tests and when I added the above environment variable it got past this error. I’m working my way through my example now, but so far what you provided helped. Thanks!