One of the significant issues in the inversion of geophysical data is the evaluation of the reliability of inverted models. This means that using mathematical tools, the degree of certainty or uncertainty of the models obtained from solving the inverse problem is determined quantitatively, and this helps to better interpret geophysical models. Evaluating the quality of water content and comfort time models resulting from the inversion of surface nuclear magnetic resonance data is also an unavoidable necessity. In this research, we extract the model resolution matrix using Single Value Analysis (SVD) of the forward MRS function. In this method, we evaluate components such as surface loop size (receiver), maximum moment pulse value, conductivity distribution of subsurface layers as well as ambient noise level conditions as input parameters affecting the resolution and penetration depth of MRS data. The effect of each of these components on the resolution of magnetic resonance sounding data is measured through synthetic models.