CLM in WRF

[edennis@umd_edu]

One of the most useful/interesting components of CLM is the telescoping (i.e., landuse; soil/snow columns; PFTs) structure in each grid cell. But when coupled to WRF, the preprocessing handles all of the land use/soil type data before WRF 'knows' which land surface model it will be using for the simulation. Does tht mean the heterogeneity of the soil type information has already been removed before WRF knows that it is using CLM (which is specified afterward in the model namelist)? Is my interpretation of this correct? For example, if you are running WRF with 15km horizontal grid spacing, the grid will be built with the land use/soil type information at 15 km before it 'knows' that CLM is the choice of LSM, even though CLM can handle more complex terrain data. So does that disallow the useful telescoping structure?
How does this affect the performance of CLM in WRF? Does this mean the version of CLM in WRF is not the same as the stand alone version, or the version in CESM?Any insights? —EJD

 

[oleson]

I personally don't know anything about this, but the following is a response from someone who knows a bit about it:CLM in WRF was done by an external university group. 1. My understanding is that the version in WRF is CLM 4.0, so to answer the question, the CLM in WRF is not updated with the CESM version, it was a one-off coupling. 2. It is true that the land cover for the grids are predetermined by the WRF pre-processor, independent of land model. That said, there is a land cover fraction for each of the land cover types from the high-res (typically 1km) input data to the WRF resolution (e.g. 15km in the question). However, in my understanding, this information is not used in WRF-CLM. Since WRF uses the USGS or IGBP classification and not a PFT approach, there are predetermined conversions to transfer the WRF classes to fractional coverage of 4-PFT (max) assumed in this CLM version. From a quick look at the code, it seems that most of the land covers are converted to 2 or 3 PFTs. It also looks to me that by default (i.e., it's hardcoded) that each PFT is running on its own soil column.