Detecting change in a dynamic world

• Main Author: Simpson, Gavin
• Format: Other/Unknown Material
• Language: unknown
• Published: Zenodo 2022
• Subjects: time series; generalised additive models; trends; penalized splines; space; time; spatiotemporal; Sea ice
• Online Access: https://doi.org/10.5281/zenodo.7272350

As we've collected data over longer and longer periods and at ever finer scales, the questions we ask of those data have also changed. Previously, we might have been happy just asking whether we can detect linear change. Today, we are asking more nuanced questions, such as are rates of change themselves changing? We are attempting to explain those changes by relating drivers to responses. And we're interested in change beyond the average; are the systems we study becoming more variable? Are the frequencies of extreme events changing? In many cases however, we're trying to do this with the same tools that we used to answer the simpler, original questions. In this talk I'll show how a particular type of model, a generalized additive model or GAM, can be used to answer these new questions about spatiotemporal change without requiring a high degree of statistical expertise. I'll illustrate my talk with examples from my own research and a reanalysis of the Arctic sea-ice extent time series.