cleanup

Author: wpbonelli

autotest/test_prt_clamp.py

@@ -1,14 +1,5 @@
 """
-Subcell tracking methods "snap" particles to whichever (sub)cell they're
-assigned to. This is both a separation of concerns (the tracking methods
-can be kept smaller/simpler, avoiding redundant logic to cross-check the
-cell ID against the particle's coordinates) and a user-facing feature: a
-particle released close to a cell boundary may be determined invalid due
-to floating point imprecision, by default, so the user may choose to set
-COORDINATE_CHECK_METHOD NONE to tolerate particles ending up outside the
-cells they are assigned to. This is effectively a "snap to cell" feature.
-
-These test cases exercise this "snap" or "clamp" behavior in lateral and
+These test cases exercise cell "snap" or "clamp" behavior in lateral and
 vertical dimensions. There is currently no distance limit; particles may
 begin arbitrarily far outside the cell.
 

doc/ReleaseNotes/develop.toml

@@ -253,5 +253,10 @@ description = "The groundwater cell number written to the binary budget file for
 
 [[items]]
 section = "fixes"
-subsection = "advanced"
+subsection = "model"
 description = "The PRT model's Particle Release Point (PRP) package performs release point validation using a point-in-polygon test that compares raw coordinates for exact floating point equality. This could incorrectly flag points close to a cell boundary as outside the cell, especially for rotated cells or cells with very large coordinate values. Add a small tolerance to the containment check, scaled to the cell's extent, to reduce the odds that a point will fail the check due to numerical inexactness."
+
+[[items]]
+section = "fixes"
+subsection = "model"
+description = "The PRT model did not consistently clamp particle coordinates to the extent of the cell before applying the tracking method. This could cause an incorrect exit face or travel time solution, for example due to roundoff in the coordinate transformations computed as the particle moves between cells. This has been corrected; coordinates will now be clamped to the cell boundary in all dimensions. Results may change slightly for models with particle positions lying close to cell boundaries."