A very common question I get is how to tune the "albedo" in the new delta-Eddington radiation scheme. One actually tunes the inherent optical properties of the snow, bare ice, or pond surfaces via the r_snw, r_ice, and r_pnd parameters respectively. These change the effective albedo in the same direction by adjusting up or down by the number of standard deviations from observations. For example, here is a discussion of adjusting the snow grain radius.
The best way to think of all of these is a linear plot (attached). The x-axis is temperature and the y-axis is effective snow grain radius. There are a couple curves here. First r_snw is used to determine the non-melting snow grain radius as follows:
rsnw_nonmelting = 500. - r_snw*250., where 50. < rsnw_nonmelting < rsnw_mlt_in (next)
Where 500. is the base value and 250. is the standard deviation from the observations. So r_snw < 0, increases the dry snow grain radius and lowers the effective albedo of dry snow, while r_snw > 0 does the opposite. The r_snw parameter was designed to adjust the albedo of dry snow with the sign in the same direction. The r_ice and r_pnd are similar animals for the bare ice and ponded surfaces. For the other two snow parameters look at the attached plot. The linear curves are both:
rsnw = rsnw_nonmelting + (rsnw_mlt_in + rsnw_nonmelting)*fT
where fT = (dTs/dT_mlt_in)=1., 0 <= fT <= 1
dTs is Tmelt-Tsfc and is computed during the run.
The namelist input parameters: rsnw_mlt_in and dT_mlt_in have two defaults currently:
Solid linear curve: rsnw_mlt_in = 1000., dT_mlt_in = 1. (Used for most configurations with CAM5)
Dashed linear curve: rsnw_mlt_in = 2000.0, dT_mlt_in = 2.
Not shown: rsnw_mlt_in = 1500., dT_mlt_in = 1.5 (Used for CAM4 configurations)
The dash-dot and dotted curves are possible nonlinear fits for melting and are not implemented currently but are provided for comparison.
The other curves are