Abstract:
|
Spacetime data with regularly sampled temporal measurements can be modeled reliably using Gaussian Processes with a spectral-in-time approach, where time is exchanged for frequency in the specification of the covariance model (Stein, 2005). In this talk we examine environmental data where the cyclostationary structure is accounted for in the model by specifying +/- 1 cycle per day offsets of the Loeve spectrum.
|