Subdividing Rain Drop Arrivals into Steady Intervals

Student: Brianna Brunson

Major: Mathematics

Mentors: Dr. Michael L. Larsen

Department: Physics and Astronomy

Subdividing Rain Drop Arrivals into Steady Intervals

In this project we adapted an algorithm built on a statistical framework designed to divide time series data into steady intervals. The algorithm applies a Bayesian framework through a "blocking" method and defines a likelihood parameter to unambiguously localize change-points in the underlying arrival rate. This algorithm was then applied to synthetic data to confirm that it effectively localized the (known) change points in the simulated data. Real rain drop data measured by a 2-dimensional video disdrometer was then explored. The Bayesian Blocks algorithm identified change-points that provided a non-parametric partitioning of the data into non-equal, steady temporal intervals. This partitioning was then used to explore the changing physical properties of rain like rainfall rate and mean drop diameter.