Anomaly detection for analysis of annual inventory data: a quality control approach

Abstract

Annual forest inventories present special challenges and opportunities for those analyzing the data arising from them. Here, we address one question currently being asked by analysts of the US Forest Service’s Forest Inventory and Analysis Program’s quickly accumulating annual inventory data. The question is simple but profound: When combining the next year’s data for a particular variable with data from previous years, how does one know whether the same model as used in the past for this purpose continues to be applicable? Of the myriad approaches that have been developed for changepoint detection and anomaly detection, this report focuses on a simple quality-control approach known as a control chart that will allow analysts of annual forest inventory data to determine when a departure from a past trend is likely to have occurred.