[check_swatinit] Cells flagged as UNKNOWN

[berland]

This is a dataframe snip where some cells have been classified as UNKNOWN, due to SWAT being approximately equal to SWATINIT. This should probably be categorized as PC_SCALED, but the difference between SWAT and SWATINIT is a little too large to uncritically assume they are equal (and what should be tolerance?)

          I    J   K              X             Y            Z       VOLUME  GLOBAL_INDEX        PERMX      PORO  NTG  SATNUM  EQLNUM  FIPNUM       SWL  SWATINIT     SWLPC        PCW       PPCW      SWAT   PRESSURE  PCOW_MAX  Z_DATUM  PRESSURE_DATUM  OIP_INIT     OWC        PORV  PC_SCALING  SWATINIT_SWAT  SWATINIT_SWAT_WVOL  QC_FLAG        PC
291187  122  162  43  486301.979135  6.829820e+06  3656.905762  2618.914988     2960666.0  1386.720947  0.205573  1.0       1       2       2  0.108680  0.301594  0.108680  10.405046  10.405046  0.301583  630.89860   8.82698   3660.0          631.14         0  3664.0  538.378777    1.178778       0.000010            0.005503  UNKNOWN  0.135648

The tolerance is currently set to 1e-6. To cover all UNKNOWN cells in the datamodel where this is found (and where it looks like PC_SCALED is the correct flag), the tolerance must be bumped to 1e-3.

[berland]

subscript/src/subscript/check_swatinit/check_swatinit.py

Line 342 in 589f045

np.isclose(qc_frame["SWAT"], qc_frame["SWATINIT"], atol=1e-6)