Removed some obviously wrong trait data that was leading to unrealistically high Pine SLA values (which could explain their excessively high LAI)
Several were mislabeled and therefore were therefore silently ignored by ED2.
My flux aggregations are probably wrong because they are disproportionately weighted towards instantaneous fluxes around midday. ED's monthly outputs average evenly over all time steps, and are therefore more correct. But, ED also has a weird monthly file bug (https://github.com/EDmodel/ED2/issues/283), which I can work around.
After lots of deep exploration, I have a clearer picture of where to go with the paper. A few notes:
Figures showing the impacts of the three sub-models, compared to default ED, at the same set of parameters:
Finite canopy radius increases the amount of light available to understory trees, with positive impacts for growth. Side-by-side LAI and light levels by PFT, for a few select years during the simulation.
Multiple scatter model produces slightly more diffuse radiation in the understory, leading to a slight boost in overall productivity.
Enabling trait plasticity seems to slightly reduce productivity. My guess is this is because the penalty to high-light canopy trees, which are the most productive, outweighs the bonus for understory trees. Ideal thing would be to hack ED2 into spitting out the light-adjusted Vcmax/SLA values for each PFT, but I have a few ideas about easier ways to show this.
For time-series summary figures, add lines for ED2 run with default parameters and at the mean/median of the trait meta-analysis. Also, add ED2 default values to parameter distribution figure. Likely discussion topic: ED2 sometimes needs unrealistic default parameters (e.g. zero growth respiration for deciduous trees) to fit observations.
Look at impacts of each individual parameter on ED2 simulations, all else being equal. Show a few of these figures for the most important parameters in the results (alongside the sensitivity analysis); dump the rest into the supplement. I now know how to do these runs quickly.
Modeled NPP averaged from 1930-1950, by model configuration. Here, ED2 is run with default parameters. C = closed canopy, F = finite canopy, T = two-stream RTM, M = multiple-scatter RTM, S = static traits, P = plastic traits:
NPP time series, by model configuration:
Other updates
Hector
Need to prepare presentation for meeting with EPA for 10/7.
FoRTE paper 1
After lots of deep exploration, I have a clearer picture of where to go with the paper. A few notes:
Modeled NPP averaged from 1930-1950, by model configuration. Here, ED2 is run with default parameters. C = closed canopy, F = finite canopy, T = two-stream RTM, M = multiple-scatter RTM, S = static traits, P = plastic traits:
NPP time series, by model configuration:
Other updates