Long-term data sets are required to investigate sources of natural variability in the upper atmosphere. Understanding the influence of sources of natural variability such as the solar cycle is needed to characterize the thermosphere + exosphere, to understand coupling processes between atmospheric regions, and to isolate signatures of natural variability from those due to human-caused change. Multi-year comparisons of thermospheric + exospheric Balmer α emissions require cross-calibrated and well-understood instrumentation, a stable calibration source, reproducible observing conditions, separation of the terrestrial from the Galactic emission line, and consistent data analysis accounting for differences in viewing geometry. We discuss how we address these criteria in the acquisition and analysis of a mid-latitude geocoronal Balmer α column emission data set now spanning two solar cycles and taken mainly from Wisconsin and Kitt Peak, Arizona. We also discuss results and outstanding challenges for increasing the accuracy and use of these observations.