track C dens of non-regenerating forest patches and apply in conversion to shrubland

[msimmond]

Update CALAND so that the C density of non-regenerating patches is tracked within the annual loop so that when the conversion to shrubland happens it doesn't use the area-weighted average of forest. Instead, it will use the C density post-fire of that specific area that does not regenerate.

[aldivi]

do some tests to figure out how much of impact this has this will may not be fixed before initial release

[aldivi]

note that carbon is still conserved in this system. there is a very small distortion of forest and shrubland carbon depending on whether the fire emissions leave lower carbon in the forest than shrubland (results in lower forest carbon and higher shrubland carbon than if tracked explicitly) or leave higher carbon in the forest than shrubland (results in higher forest carbon and lower shrubland carbon than if tracked explicitly). These two cases both happen in across the state and cancel each other out to some extent.

The correct option would be to add an internal nonregen forest type and deal with the carbon appropriately during the fire module. Then the correct carbon could be transferred in the land change module from the nonregen to the shrubland. This special case would need to be dealt with explicitly. Probably would need a shrubland fire decay pool also.

A simpler method would do the conversion to shrubland directly in the fire module. This may affect other shrubland conversion in the land change module.

[aldivi]

actually, there are some cases where some forest carbon is leaked (when the non-regen fire emissions are less than the difference between the forest c density and the shrubland c density in the conversion module). currently, these leaks are small (and occur in only about half of the land units) based on the initial state and both immediate and delayed fire emissions. delayed fire emissions are still attributed to the forest and not the shrubland.

and the fix does not need to have a new land type, I think. As long as the post-fire density of the non-regen land is stored (for all pools), and likewise the delayed fire emission stored in an additional shrubland array, a non-regen condition can be added to the conversion module that transfers the correct carbon (the atmos emissions are already accounted for).

however, the implementation may effectively add a non-regen land type internally to keep track of the carbon data