Good, bad, and the ugly: Seasonal Filtering and Autoregressive Moving Average (ARMA) models for detecting and replacing spikes in velocity profile data