1DVar problem: allocatable array or pointer is not allocated

[Michael Rennie]

Came across problem during testing. Following command gives output (intel fortran compiler):

ropprepos/ropp_1dvar/build/ropp_1dvar_refrac_ec_ifort_linux -y ropprepos/ropp_test/data/1.0/IT-1DVAR-02_y.nc -b ropprepos/ropp_test/data/1.0/IT-1DVAR-02_b.nc --bg-corr ropprepos/ropp_test/data/1.0/IT-1DVAR-02_c.nc -o ropprepos/ropp_test_tmp/IT-1DVAR-02.1.nc

Output from shell:

---

ROPP Refractivity 1DVar for ECMWF Model Level Data - v0.9

---

INFO (from ropp_1dvar_refrac_ec):

Reading observation data for profile 1 from the file

ropprepos/ropp_test/data/1.0/IT-1DVAR-02_y.nc.

INFO (from ropp_1dvar_refrac_ec):

Reading background data for profile 1 from the file

ropprepos/ropp_test/data/1.0/IT-1DVAR-02_b.nc.

INFO (from ropp_qc_genqc):

Background quality control lets all refractivity values pass.

INFO (from ropp_1dvar_solve):

Using background error covariance matrix for preconditioning.

INFO (from ropp_1dvar_cost):

Using absolute decrease of cost function and relative change of state

vector

as additional convergence criterium.

n_iter = 1 J = .76999E-09 max(relative change in state) =

-

INFO (from ropp_1dvar_cost):

Convergence assumed to be achieved as the cost function did not

change by more

than 0.1000 for the last 2 iterations.

forrtl: severe (153): allocatable array or pointer is not allocated Image PC Routine Line Source ropp_1dvar_refrac 08356EA2 Unknown Unknown Unknown ropp_1dvar_refrac 08355575 Unknown Unknown Unknown ropp_1dvar_refrac 08335BD3 Unknown Unknown Unknown ropp_1dvar_refrac 08309196 Unknown Unknown Unknown ropp_1dvar_refrac 0832200B Unknown Unknown Unknown ropp_1dvar_refrac 080533D4 Unknown Unknown Unknown ropp_1dvar_refrac 0804ED8E Unknown Unknown Unknown ropp_1dvar_refrac 0804B48F Unknown Unknown Unknown ropp_1dvar_refrac 0804A6B5 Unknown Unknown Unknown libc.so.6 0048CDE3 Unknown Unknown Unknown ropp_1dvar_refrac 0804A5F1 Unknown Unknown Unknown

and for portland group compiler:

---

ROPP Refractivity 1DVar for ECMWF Model Level Data - v0.9

---

INFO (from ropp_1dvar_refrac_ec):

Reading observation data for profile 1 from the file

ropprepos/ropp_test/data/1.0/IT-1DVAR-02_y.nc.

INFO (from ropp_1dvar_refrac_ec):

Reading background data for profile 1 from the file

ropprepos/ropp_test/data/1.0/IT-1DVAR-02_b.nc.

INFO (from ropp_qc_genqc):

Background quality control lets all refractivity values pass.

INFO (from ropp_1dvar_solve):

Using background error covariance matrix for preconditioning.

INFO (from ropp_1dvar_cost):

Using absolute decrease of cost function and relative change of state

vector

as additional convergence criterium.

n_iter = 1 J = .76999E-09 max(relative change in state) =

-

INFO (from ropp_1dvar_cost):

Convergence assumed to be achieved as the cost function did not

change by more

than 0.1000 for the last 2 iterations.

0: DEALLOCATE: memory at (nil) not allocated

Mike

[marq]

Changesets [1042] and [1043] should fix this issue on a technical side, i.e. the core dump shouldn't occur any more.

What worries me more is that the cost function evaluator is called only once; using the default settings, it should be called at least twice. This is (hopefully) fixed in [1044].