Improvements towards using subgrid CLM output to study land-atmosphere interactions

The land surface in the Community Land Model (CLM), the land component of the Community Earth System Model (CESM), is represented as a nested hierarchy of subgrid levels. By default, the vegetated fraction of a gridcell contains up to 16 plant functional types (PFTs), all of which share a single soil column. The PFTs specify the differences in biophysical and biogeochemical processes across different vegetation types, and provide an opportunity to examine how the different biophysical processes of each PFT respond to and interact with the atmosphere. While the distribution of PFTs within the vegetated fraction of a grid cell is represented in a tile or mosaic configuration, PFTs still compete for water and nutrients through the shared soil column. In this study, we compare PFT-level output from a modified land surface configuration (PFT-COL) where each PFT is given its own soil column to the PFT-level output from the default configuration (CTRL) where all PFTs share a single soil column. We use CLM4.0 and CLM4.5 (CESM1.2) forced by the CRUNCEP atmospheric dataset to examine how the individual vs. shared soil column configurations affect CLM output both at the PFT-level and the gridcell-level. Our results show that there are large differences in the PFT-level output between the PFT-COL and CTRL simulations, but likely due to compensating changes at the PFT-level, there are only minor differences in the gridcell-level output between the PFT-COL and CTRL simulations. Overall, we find that the PFT-level output is more reasonable and realistic in the PFT-COL simulations. We find that in the CTRL simulations, the shared soil column allows heat and moisture to transfer horizontally from one PFT to another, producing erroneous values most notably for air temperature, sensible and latent heat fluxes, and the ground heat flux. As the individual column configuration (PFT-COL) may more closely represent "real world" conditions, these results have implications for future subgrid land surface model experiments, including the ability to examine how different PFTs respond to atmospheric forcings, and how different land cover types influence climate at the local scale. In addition, this new configuration presents an opportunity to validate CLM at the PFT-level.