Space Applications for Ensemble Detection and Analysis (EDA)

NASA makes extensive investments to circumvent the engineering challenges posed by naturally occurring random processes for which conventional statistics do not provide adequate numerical description. Ensemble Detection and Analysis is a transformative, crosscutting technology that addresses measurement challenges for these naturally occurring random events.  Ensemble Detection produces data that are admissible to statistical analysis and filtering algorithms that are otherwise not possible to implement. EDA is a form of Noise Assisted Data Analysis (NADA) that has broad application to dynamic systems characterized by changing statistics. The underlying mathematics predicts the existence of stochastic waves on which time points toward the direction of greater uncertainty.

Ensemble Detection is both a measurement technique and analysis tool. Like a prism that separates light into spectral bands, an ensemble detector mixes a signal with noise references that span a range of noise powers that produce multiple realizations that comprise an ensemble data set. The resulting ensemble data set represents a comprehensive stochastic description of the signal that is admissible to statistical analysis not otherwise possible with a single realization. Ensemble data sets provide time-varying probability distribution functions from which time varying statistics can be derived. The linear relationship between the reference mean and standard deviation provides a way to discriminate changes in the process mean from changes in the spread of the process’ probability distribution. Why is this important? Distinguishing between a mean state change and natural variability is a problem that spans multiple scientific disciplines. One important application is discriminating between climate change and natural variability of extreme weather events.