What CLASS-related control flag options should I select?

Here, 3 control flags are discussed: IDISP, IZREF and ISFLD.

IDISP:
If IDISP = 0, vegetation displacement heights are ignored, because the atmospheric model considers these to be part of the "terrain".
If IDISP = 1, vegetation displacement heights are calculated.
By default, IDISP = 1.

IZREF:
If IZREF = 1, the height to the bottom of the atmospheric model is taken from CLASS intialization file and MESH extrapolates to the surface.
If IZREF = 2, the bottom of the atmospheric model is taken to lie at the local roughness height.
By default, IZREF = 1.

IZREF = 1 should be okay for both forcing data from NWP models or field measurements so long as the height of the measurement is specified correctly.

ISLFD:
If ISLFD = 0, DRCOEF is called for surface stability corrections and the original GCM set of screen-level diagnostic calculations is done.
If ISLFD = 1, DRCOEF is called for surface stability corrections and SLDIAG is called for screen-level diagnostic calculations.
If ISLFD = 2, FLXSURFZ is called for surface stability corrections and DIASURF is called for screen-level diagnostic calculations.
By default, ISLFD = 2.

The recommended settings for IDISP and IZREF are complicated if field data is used as meteorological forcing. Ideally (normally) set IDISP = 1 and IZREF = 1 if using field-observed forcing data. Difficulties may arise (i.e. CLASS will likely crash) if the input wind speed measurement height is below the calculated canopy displacement. If this is the case, then set IDISP = 0 and IZREF = 2; however, in some cases it may be physically difficult to justify doing so. Setting IDISP = 0 and IZREF = 2 when using field observations is reasonable if the open and forested areas are relatively large such that the overlying atmosphere is well mixed and there are negligible differences above the forested and open areas (edge effects are negligible).
Do not set ISFLD = 0 because it uses the crudest set of screen-level diagnostic calculations. This option has only been left in for comparison with earlier GCM runs. Note that if setting ISLFD =1 , surface stability corrections are made in the subroutine DRCOEF. DRCOEF is unable to deal with forcing data where where wind speed and temperature/humidity forcings are at different heights. If the wind speed and temperature/humidity forcings are set at the same height in MESH_parameters_CLASS.ini then it is OK to set ISFLD = 1. If this is not the case, however, then set ISFLD = 2. In light of the above discussion, we recommend setting ISFLD = 2 and not to bother deviating from this.