Underground Mining (UM) is a harsh environment that involves multiple threats to worker safety and needs adequate wireless communications to ensure worker well-being and optimize miner efficiency. While radio frequency signals can be used to manage such tasks, several limitations exist to its potential, as multiple phenomena affect their quality. Therefore, Visible Light Communications (VLC) has been proposed as a robust alternative to RF usage due to its many advantages over typical wireless technologies. Nevertheless, the conditions of the underground mines diverge importantly from the assumptions and simplifications of indoor models. To mitigate the problems generated by the UM environment in the VLC system, there must be adequate channel modeling that includes all the physical and external features that affect the communication system. To address this topic, in this paper, we present a generic Multiple-Input-Multiple-Output (MIMO) VLC channel analytic model for a UM environment. This channel model is evaluated with different positions of the optical receiver in a UM scenario in terms of typical metrics such as Channel Impulse Response (CIR) and Root Mean Square (RMS) delay spread. These results are the baseline that will allow us to propose solutions to improve the efficiency of the UM-VLC system in the future.