IFLAG1 problem. The model didn't converge!

[psharma-gif]

I am having trouble converging the model. Has anyone faced a similar issue?

                 !!! FATAL ERROR !!!                          

---

---> File:        rk4_integ_utils.f90
---> Subroutine:  print_rk4patch
---> Reason:      IFLAG1 problem. The model didn't converge!

---

ED execution halts (see previous error message)...

---

Likely to be a rejected step problem.

MEND_SANITY_CHECK
POM1, CNP -225.7839 0.2258064 NaN
POM2, CNP -80.63709 8.0645151E-02 NaN
DOM, CNP -4.838226 4.8387093E-03 NaN
MOM, CNP -677.3516 0.6774193 NaN
AMB, CNP 2.3000000E-02 4.6000001E-03 5.5000000E-04
DMB, CNP 0.2700000 4.6000000E-02 5.5000000E-03
INVARS, NH4,NO3,PSOL 9.9999997E-06 9.9999997E-06 -0.8063840
INVARS, NH4ADS,NO2,NO 9.9999997E-06 9.9999997E-06 -1.3216177E-10

I suspect the issue is related to NaN. Should I address it in ED2IN or the source code?

[mdietze]

Check you met data to make sure there are no odd values or sudden changes that would mess up the energy balance. If that looks find I'd try making the timestep smaller. But I've never run the version of ED2 with MEND (which I don't think is even in the github repo?) so I don't know what could be causing NaN's there.

[mpaiao]

The NaN is probably the source of this issue indeed, and it does look like something specific to nutrients. As @mdietze mentioned, MEND was never submitted as a pull request to the main ED2 repository (though it would be great to have it), so I don't know what it could be causing the issues. I am flagging @dmedvigy who may have some insight.

For future reference, the other common cause for model failure related to met drivers is setting the variable NL%IMETAVG incorrectly in the namelist. NL%IMETAVG will eventually be dropped from ED2IN and become part of the meteorological driver header, but in the meantime, make sure it reflects what the time stamp in the meteorological drivers indicate (beginning, middle or end of the average window). But this does not seem to be your case, because the error is associated with MEND variables.