logical inconsistency (possible bug) with missing zeta in stability.correction

[Thomas Wutzler]

Unstable conditions use stability parameter zeta in stability correction. Currently, if zeta is missing both, the test for unstable and the test for stable conditions are TRUE.

https://bitbucket.org/juergenknauer/bigleaf/src/2c6f0e4e502af9565e946ba314a6819042988d78/R/stability_correction.r#lines-175

The risks of results depending on evaluation order and of missing values in psi. I suggest

• either - explicitly setting psi to NA if zeta is missing (introduce third stability case: missing), or
• treat missing zeta as stable condition.

for the second option, I propose replacing all occurences of “[unstable]” by “[!stable]” which ensures that no record is classified twice.

‌

‌

[Juergen Knauer]

Bug fix in stability.correction(): test for stable/unstable conditions are false if zeta is missing. Function returns NA in this case. Fixes Issue #8.

→ <<cset 3c49e4b7059a>>

[Juergen Knauer]

Thanks, well spotted. I have implemented the first option but instead of introducing a third case I have initialised the output to NA which will not be overwritten if zeta is missing.