The ionosphere contains many small-scale electron density variations that are under represented in smooth physics-based or climatological models. This can negatively impact the results of Observation System Simulation Experiments, which use a truth model to simulate data. This paper addresses this problem by using ionosonde data to study ionospheric variability and build a new truth model with empirically-driven variations. The variations are studied for their amplitude, horizontal and vertical size, and temporal extent. Results are presented for different local times, seasons, and at two different points in the solar cycle. We find that these departures from a smooth background are often as large as 25\% and are most prevalent near 250 km in altitude. They have horizontal spatial extents that vary from a few hundred to a few thousand kilometers, and typically have the largest horizontal extent at high altitudes. Their vertical extents follow the same pattern of being larger at high altitudes, but they only vary from 10s of km up to 200 km in vertical size. Temporally, these variations can last for a few hours. The procedure for using these spatial and temporal distributions to add empirically-driven variance to a smooth truth model is outlined. This process is used to make a truth model with representative variations, which is compared to ionosonde data as well as GPS Total Electron Content (TEC) data that was not used to inform the model. The new model resembles the data much better than the smooth models traditionally used.