This thesis presents a software tool which simulates the geological stratigraphy of a potash mine which is then used with gprMax (public domain Ground Penetrating Radar (GPR) simulation software) to examine and evaluate the effectiveness of auto-picking algorithms. The system is used to simulate the GPR response from clay seams in the roof of potash mining rooms. As it is extremely onerous to obtain in-situ data that captures all possible normal and anomalous geological conditions present in the mine roof, earth models are generated which accurately represents the geology of the mine. In particular, random clays in the mine roof can negatively affect the performance of auto-picking algorithms. These earth model simulations can be used to present these random clays accurately gprMax is an open source software that simulates Electro-Magnetic (EM) wave propagation in materials in order to support better understanding of the use of GPR in various application domains. Currently, GPR machines are being used in the potash mine as a secondary inspection of salt (roof) thickness. The goal of this thesis is to validate the ability to use gprMax with an effective earth model to generate realistic GPR signals to test/evaluate auto-picking algorithms. The use of simulated data in comparison to the experimental (actual) data and generate test bed models for an auto-picking algorithm has many benefits. The synthetic data is generated by gprMax using the Finite Difference Time Domain (FDTD) methodology. An effective methodology to develop and test robust auto- iii picking algorithms is created using simulated GPR signals because the ground truth is known from the earth models. Additionally, in this thesis results from both an industry standard auto-picking algorithm and a new auto-picking algorithm, called Clustered Ratio Derivative (CRD), are presented for the mine roof application. Finally, in this thesis we take advantage of cloud computing resources in order to execute this work and our use if this technology is summarized.