The warning you received says that some formulas may be no usable since a building code assumes that the stories weight are (more or less equally) distributed over the building height.
I'm not familiar with NBC 2010. A reasonable approach would be to find out the period of the first mode of vibration (T1) and to look into the seismic spectrum which is the corresponding spectral acceleration. The idea is that a rigid body (T=0) on soil will feel the PGA, a vertical vessel with period T1 can be assumed to be subject to an increased acceleration as the spectrum shows. In fact this is not fully-true, theoretically all the modal responses of vessel will be involved in a seismic event.
There are Codes asking to assume that a non-structural component (as the vessel is in relation with a building) is subject to the maximum acceleration of the spectrum (it may be 2.5 to 3.0 times PGA) regardless T1 (in fact, this assumption means that T1, T2, T3, .... all will fall in the maximum values zone of the spectrum). Also, depending on the Code, it may be required to make a more accurate analysis considering more periods of vibrations (T1, T2, T3, ....etc) by effectively introducing the seismic spectrum.
Another issue may be the "ductility" involved or not in a seismic event. In fact is about the capacity to develop "plastic hinges" that will dissipate the seismic energy. If yes, a coefficient shall be considered in a denominator, if no- the spectrum is told to be "elastic" because the legs are supposed to remain into elastic domain of stress-strain, case in which their section must be increased.
Referring your query: you may consider the equivalent static x*PGA in X direction, Y dir, 0.7X and 0.7Y (45 degree, lets say), alone or together with a vertical component, as separate scenarios or combined by SRSS, etc.