Variability and Trend Analysis of a Grid-Scale Solar Photovoltaic Array above the Arctic Circle

Abstract

As solar photovoltaic (PV) power generation continues to grow in popularity, the variability in solar irradiance caused by weather effects such as clouds poses an increasing challenge to maintaining grid stability. Characterizing the variability and trends present in historical PV production data is vital to the development of effective models for predicting rapid changes. This is particularly important at higher latitudes where seasonal changes in PV generation are more extreme. In this paper, we analyse data from a small, grid-scale PV array in Kotzebue, Alaska (66.8969° N, 162.5931° W), located above Arctic Circle. We also successfully validate the variability index (VI), a previously proposed metric which quantifies the volatility of solar PV data over a given time span using a synthetic cloudless (clear-sky) dataset as a reference. We also include an examination of the stationarity of the PV production data at various timescales as well as the efficacy of using clear-sky models as a reference for de-trending solar irradiance data via, showing that better results can be obtained from data closer to solar noon. To the best of our knowledge, this is the first use of VI to assess PV production data from above the Arctic Circle.