Bugfix: wind boundary

[scrasmussen]

TYPE: bug fix

KEYWORDS: wind field, boundary conditions

SOURCE: Soren Rasmussen, NCAR

DESCRIPTION OF CHANGES: Boundary conditions of the wind fields causes significant answer changes. The calculation of the the halo region bounds was slightly off for the wind fields.

ISSUE: The GitHub Issue that this PR addresses. For issue number 123, it would be Fixes #123

Checklist

• [ ] Closes issue #xxxx (An issue must exist or be created to be closed. The issue describes and documents the problem and general solution, the PR describes the technical details of the solution.)
• [ ] Tests added (unit tests and/or regression/integration tests)
• [X] Backwards compatible
• [ ] Requires new files? If so, how to generate them.
• [ ] Fully documented

[scrasmussen]

Ok, I think what happened was my “upstream base case” directory was accidentally overwritten with “wind_fix base case” so I thought it was working. Here are current results

• The first is upstream 1 vs 36 images

• The second is wind_fix 1 vs 36 images, with halo of 1 for both

• The third is wind_fix 1 vs 36 images with halo of 3 for the 36 images

• The fourths is upstream 1 image vs wind_fix 1 image

Ran a test and the difference in upstream 1 image vs wind_fix 1 image is due to the corr_factor changes. The PR looks good to go, these runs were done with a single wind iteration, rather than the default 100. Will run tests again with 50 and 100 wind iterations to see the results

[gutmann]

change map = (old-new) / old

Fractional change in precipitation over a ~6month period changes are <5%, mostly <1%, with the "biggest" changes being a decrease on top of mountains, and an increase in valleys (that is a good thing). Other changes seem randomly distributed.