It is well known that nuclear magnetic resonance (NMR) transverse relaxation measurements of porous media at high magnetic field strengths provide only an effective relaxation time T(2,eff), as opposed to the true T(2), due to molecular diffusion through magnetic field gradients induced by the magnetic susceptibility contrast between the adsorbent and the adsorbate. Here, we deconvolve the diffusion and surface relaxation contributions to measurements of T(2,eff) and thus obtain the true T(2) relaxation time distribution. This technique is applicable within the short time diffusion regime where the diffusion exponent varies as t(E) (3), while the surface relaxation exponent varies as t(E), where t(E) is the echo time in a standard Carr-Purcell Meiboom-Gill measurement. We demonstrate this technique on measurements of water in contact with glass spheres across a range of magnetic field strengths from B(0)=50 mT to 7.4 T. A direct measurement of T(2,eff) suggests that the transverse relaxation rate increases with field strength, in contrast to theoretical predictions. We show that when the effects of the susceptibility induced gradients, which are known to increase with magnetic field strength, are deconvolved from the T(2,eff) measurement, the true T(2) relaxation rate does indeed decrease with increasing field strength. We also apply the T(2) correction in multidimensional NMR experiments using the example of a T(1)-T(2) relaxation correlation. Here, the correction is essential in order to obtain the true T(1)/T(2) ratio as a function of magnetic field strength, which provides a measure of mobility for surface-adsorbed species; without this correction, we see surface residence times overestimated by up to two orders of magnitude. This novel approach enables the accurate determination of T(2) distributions, and hence T(1)/T(2) ratios, on high-field spectrometers that would have previously been deemed inappropriate for the study of liquids in porous media because of the intrinsic susceptibility effects.