Time series analysis using R terra packages (Tomislav Hengl)

Earth Observation Summer School 2026 17 Aug 2026 – 21 Aug 2026, Istanbul, Turkey https://opengeohub.org/events/earth-observation-summer-school-2026 Tutorial (https://codeberg.org/opengeohub/MODIS_LST_trend_analysis_tutorial) will demonstrate how to use terra package to process time-series of MODIS Land Surface Temperature (LST) images. These are direct TIR images estimating surface temperature. The original MODIS Terra and Aqua products are 8-day mosaics. These are first aggregated to monthly distributions for 2000 to 2025. Terra package is then used in combination with the terra::app() to apply custom-based R functions to every cell across the layers. We will focus on using the Mann-Kendall test + Theil-Sen slope as the standard non-parametric approach to determine the trends in temperatures and map potential hot-spots of global warming. OpenGeoHub has previously run trend models and mapped beta coefficients for day-time and night-time LST (https://doi.org/10.5281/zenodo.1420114): these show that although the planet seems to be warming up, there are significant geographic differences with many areas even cooling down. We will also demonstrate how to use ARIMA methods to do forecasting of the future temperatures; and NIPALS package to derive Principal Components from time-series data. The tutorial emphasizes optimizing computing so it can be applied on large data sets by combining custom-based functions and tiling/parallelization.