Identifying and Characterizing Cyclicality and Seasonality in Environmental Data with An Adaptive Filter Implementation of A Phase Locked Loop

Abstract

Phase Locked Loops are provided as a manner of assessing, monitoring and forecasting with seasonal or cyclical data. Phase Locked Loops using adaptive filtering were initially utilized for gravitational wave data analysis. In this paper we examine this implementation as an alternative approach for identifying the presence of one or more cycles in environmental data. Unlike most techniques for assessing seasonality phase lock loops can identify the presence of cycles without prior suspicion of their nature and existence. This is important for identifying either the initiation of new cycles or changes in existing cycles. As environmental data can be impacted by the presence of numerous cyclical phenomena and these cycles can change over time, this is a suitable and valuable technique. The technique can be used on existing data sets or assessing data streams. Phase Locked Loops are advantageous over many currently as the presence of multiple different cycles can be identified without prior suspicion and specification of the nature of the cyclicality of the data. The speed and parsimonious nature of the technique makes Phase Lock Loops ideal for Big Data and data flows that are so large that they must be assessed on a real time basis. The technique can manage and advise of changes in frequency, amplitude and phase shifts. The technique is illustrated utilizing weather data.