I discovered what explains the weird behavior in the finite crown + two-stream RTM. Turns out there's a difference in how the two RTMs handle crown area index in the longwave RTM. Basically, the two-stream model scales blackbody emissivity by the crown area index, while the finite canopy radius model does not (https://github.com/EDmodel/ED2/issues/278). I haven't totally resolved the math, but as far as I understand it, the reduction in emissivity associated with crown area << 1 leads to leaves that have on average higher leaf energy, and therefore higher temperature, respiration, etc. This reduced net productivity drives declines in leaf biomass that lead to a positive feedback of declining leaf area, light capture, productivity, etc.
We should discuss what this (and Marcos and Mike's comments on that GH issue) mean for the paper. One possibility is to drop the "finite canopy + multiple scatter" combination as invalid -- that's probably the option with the least amount of friction for given the paper's current status.
I discovered what explains the weird behavior in the finite crown + two-stream RTM. Turns out there's a difference in how the two RTMs handle crown area index in the longwave RTM. Basically, the two-stream model scales blackbody emissivity by the crown area index, while the finite canopy radius model does not (https://github.com/EDmodel/ED2/issues/278). I haven't totally resolved the math, but as far as I understand it, the reduction in emissivity associated with crown area << 1 leads to leaves that have on average higher leaf energy, and therefore higher temperature, respiration, etc. This reduced net productivity drives declines in leaf biomass that lead to a positive feedback of declining leaf area, light capture, productivity, etc.
We should discuss what this (and Marcos and Mike's comments on that GH issue) mean for the paper. One possibility is to drop the "finite canopy + multiple scatter" combination as invalid -- that's probably the option with the least amount of friction for given the paper's current status.