Establishing a Baseline: Quantifying Observation Representativity Requirements for Proliferated Satellite Architectures

Future satellite Earth observation architectures may look quite different in the near future, upending a paradigm that has been nearly continuous and unchanged since the dawn of the space age. Traditionally, Earth science and operational satellites feature several instruments on the same large platform to ensure that measurements from the same field of regard could be considered collocated and simultaneously acquired. Looking at the same target at the same time provides great utility for calibrating instruments, reducing error sources, and investigating observed phenomena.

With the miniaturization of sensing payloads and the increased access to space launch services, observation planners can now consider disaggregated constellations to supplement or supplant large legacy assets. Besides the potential for cost savings, one of the major performance benefits of these hypothetical disaggregated architectures is the potential to increase revisit rate and enhance the temporal diversity of observations. However, without simultaneous and collocated measurements, there is a fundamental a prioriquestion of whether nearby samples in space and time can ever be considered functionally equivalent, and what determines the pattern of decay for representativity.

This study expands on prior work used in data assimilation and in situ observation network planning to evaluate fundamental decorrelation scales in space and time for target parameters from orbit. From there, we apply this metric to several hypothetical observations and satellite constellations to explore the behavior and relative representativity of the observation, all without ingest into a model or observing system simulation experiment. The immediate application of this methodology can be used to establish parameters for appropriate matchup conditions, and further, to establish baseline and quantitatively-rigorous performance requirements for observations that from disaggregated platforms.