I think the "amplitude" in those things is the weight of the drum divided by eccentric moment. In other words: the eccentricity of the drum as it rolls about it's axis. Therefore your sinusoidal forcing equation will be:
Fo=m*e*ω2
Where:
m=mass of your drum
e=eccentricity of the drum (0-2.5mm)
ω=circular frequency of rotation [for 28 Hz that would 175.9 rad/sec]
As far as the equivalent static load, it would be = the drum weight + (Fo*DLF)
Since you [I assume] don't have a spring constant for the soil (and you also cannot guarantee the time period it will be applied), it will be best to just use 2 as your DLF. Conceivably, it could be more than that (thinking of it as a transmissiblity problem)......but I have a hard time picturing the frequencies getting so close as to causing that.
EDIT: I kind of have to re-think my statement above (i.e. about the DLF/transmissiblity not exceeding a factor of 2). Running some rough numbers, I can see the possibility of the frequencies getting close depending on the site characteristics. However, I think damping would bail you out here to some degree. Again running some rough numbers I came out with a (minimal) damping ratio of about 15%. With that, the transmissiblity factor would be about 3.4