In this paper, a classical CPT-based liquefaction model is integrated with geostatistical tools and random field models to evaluate and map liquefaction potentials over an extended area. Verification of random field-based liquefaction prediction is challenging and rarely addressed in current literature due to limitation of soil data and liquefaction observations. To this end, a three-dimensional synthetic digital soil field is generated through novel numerical models, providing extreme detailed soil properties and corresponding liquefaction hazard information quantified in terms of the liquefaction potential index (LPI). Different virtual field testing plans are designed to investigate the effect of data inference on the model performance. Three ranking criteria are used to evaluate model performance. Knowledge gained through the model verification process can be used to guide the inference of model parameters and understanding the performance of random field-based liquefaction hazard mapping.