Time Series Analysis for Environmental Data

An R-Based Introduction

Read it here: timeseries.andybunn.org

This started as weekly handouts for ESCI 504, a graduate course in time series analysis at Western Washington University, and grew into a short book. It's free to read online and openly licensed.

What it covers

The book works through time series analysis from the ground up, in four parts:

• Foundations - what a time series is, how R represents temporal data, and trend/seasonal decomposition
• The Anatomy of Dependence - autocorrelation, the ACF and PACF, stationarity, and ARMA models for stationary series
• Working with Dependence - forecasting, cross-correlation between two series, regression with autocorrelated errors, trend detection, and reconstruction/out-of-sample validation
• The Frequency Domain - filtering and smoothing, and spectral methods for finding cycles

The emphasis is on doing the analysis in R and understanding what you're doing, not on mathematical derivation. Scattered Asides dig into the math and the R internals for readers who want them, but you can skip every one and still follow the core material.

Who it's for

Graduate students and researchers in the environmental sciences who can get around in R and want to take time seriously as a dimension in their data. It assumes an introductory linear-modeling course and comfort with basic statistics. Everything time-series-specific is built up from there.

The example data lives in data/. Every build bundles it into a data.zip you can download straight from the site, so you can run the code against the exact data the chapters use.

How it's built

The book is written in Quarto and rendered as a book site. Every push to main triggers a GitHub Actions workflow that renders the book and deploys it to timeseries.andybunn.org, so the live site is always current.

To build it locally you'll need R, Quarto, and the packages listed in DESCRIPTION. From the project root:

quarto render

Found an error or have a suggestion? Open an issue, or use the "Edit this page" link on any chapter. I'd love to hear from you.

License

The book is dual-licensed (see LICENSE.md):

• Prose and figures under CC BY-NC-SA 4.0. Share and adapt for noncommercial use, keep derivatives open under the same terms, and give credit.
• Code under the MIT License. Use it however you like.