In this section we go over what has been extensively tested and scientifically validated with CLM4, and maybe more importantly what has NOT been tested and may NOT be scientifically validated. You can use all features of CLM, but need to realize that some things haven't been tested extensively or validated scientifically. When you use these features you may run into trouble doing so, and will need to do your own work to make sure the science is reasonable.
The standard version of the model is CLMCN at 1-degree horizontal resolution (0.9x1.25). This version has been scientifically validated with long simulations for: fully coupled simulations ("B" cases), coupled to atmosphere model CAM ("F" cases), and stand-alone CLM cases ("I" cases). We've also done both long simulations for 1850 conditions, and transient 20th century simulations from 1850 to 2005 (with transient land-use, Nitrogen and Aerosol deposition). There have also been transient future scenario simulations done for fully coupled cases for different "representative concentration pathway" (RCP) scenarios (RCP2.6, RCP4.5, RCP6.0, and RCP8.5). To a lesser extent there have also been simulations done at T31 and 2-degree horizontal resolution (1.9x2.5), and with CLMSP for these resolutions. As such we have provided appropriate 1-degree, 2-degree, and T31 initial condition datasets for these configurations. The irrigation and prognostic crop models were both validated at 2-degree resolution. The irrigation model for CLMSP for present day conditions for an "I" compset, and the prognostic crop model for present day conditions for a case coupled to the active land model, but using a data ocean model (an "F" compset). Other resolutions, configurations, and namelist options are less well tested or scientifically validated. The further you get away from the standard configurations and resolutions, the more likely you are to run into trouble, and/or need to scientifically validate your work.
In the sections below we go through configuration and/or namelist options or modes that the user should be especially wary of using. You are of course free to use these options, and you may find that they work functionally. Although in some cases you will find issues even with functionality of using them. If so you will need to test, debug and find solutions for these issues on your own. But in every case you will need to go through more extensive work to validate these options from a scientific standpoint.