Fire raster holds the latest fire at or before dataYear

[eliotmcintire]

Stand ages inside old burns were wrong where a pixel burned twice: the later fire hid the earlier one, so for an earlier vegetation year the pixel kept its satellite age. The module now builds its fire map for the chosen vegetation year only.

Affects: stand age (and biomass re-estimated from it) inside reburned fire perimeters, for vegetation years earlier than the latest fire.

Depends-on: PredictiveEcology/LandR#275

Technical details

Defect: Biomass_borealDataPrep.R (the prepInputsFireYear() call in .inputObjects, ~line 1810) rasterized all fires with fun = max, so a pixel burned in 1980 and 2015 held 2015 and replaceAgeInFires(startTime = dataYear) skipped it for dataYear 1990. The same raster is used for the young-biomass override (~line 1415, non-NA test only), which was missing the same pixels and is now right without other changes; earliestFireYear is untouched.

Fix: pass latestYear = P(sim)$dataYear; reqdPkgs floor is LandR (>= 1.2.0.9057), the version LandR development (now 1.2.0.9056) will have once LandR#275 is merged and bumped. A firePerimeters supplied by another module is used as given.

Test: a static check that the fire call passes latestYear = P(sim)$dataYear (fails on development, passes on the branch); the raster behaviour is tested in LandR#275. Module suite (convertToPackage() + test_local(), scratch library phase4/lib with LandR from #275 installed as 1.2.0.9057): development 127 passed, 3 errors; branch 130 passed, 3 errors. The 3 errors are in test-estimateCoverModel.R (Copy not found) on both: the local harness does not attach reproducible, CI does, so they are not a module defect.

🤖 Generated with Claude Code