questions about landuse changes in the CLM

[andraded@eas_slu_edu]

In an experiment, I changed the landuse (PFT coverage) in the CLM over a small area of the US. In less than 2 hours of model time, significant differences between the modified and control run output (CAM variables such as temp, sea level pressure) start to appear on opposite ends of the world, which would require waves to propagate faster than sound. However, nothing obviously unphysical seems to appear later in the modified run (just a different solution). I was wondering, if in theory, changing the landuse requires a complete spin up? Or if the initial fields such as soil moisture need to be adjusted before making changes?


Thanks for any help.

 

[slevis]

That's an interesting result if you started the two runs the same way and, therefore, expected no changes other than the ones caused by your regional pft change.

The answer to your questions depends on your experiment. E.g., you could be investigating the immediate (transient) land-atmosphere response to small scale deforestation. Or you could be investigating the long term (equilibrium) response. I don't think that the former would require a spin up, but it may require multiple realizations starting from slightly different initial conditions to tease out a signal from the model's natural/internal variability. On the other hand, the equilibrium response would need a long enough run to separate the signal from the noise, but probably not multiple runs. By the same token, for a transient response you may want to start with initial conditions from the run without your landuse change. For an equilibrium response you could do the same or modify the initial conditions to something closer to the new equilibrium. The problem there is that you don't know ahead of time what to expect for a new equilibrium. So in general we start with initial conditions from the control.

Sam Levis