Mesoscale dynamics in the mesosphere and lower thermosphere (MLT) region are mainly due to atmospheric gravity waves (GWs) and stratified turbulence. GWs are assumed to be the dominant drivers of MLT mesoscale dynamics. However, characterising and observing these scales is challenging due to their scales and intermittent nature. This thesis investigates the mesoscale dynamics of the MLT region using ground-based observations. The main dataset used in this work is from multistatic specular meteor radars (SMRs). The study is complemented with observations from airglow imagers and satellites and with direct numerical simulations as well as high-resolution model data. This thesis uses a novel second-order statistics approach to investigate the correlation functions in temporal and spatial scales from one-dimensional projections of three-dimensional wind velocities. In a case study, we found that unexpected large horizontal scales dominate the dynamics. Intriguing high vertical velocity estimates using multistatic SMRs at different latitudes are also addressed through a validation study by constructing virtual multistatic radar systems based on winds from a high-resolution general circulation model.