Introduction to Forecasting Phenology with NEON and phenocam R packages

<p><strong>Intro to Material </strong></p> <p>This resource is written in Rmarkdown, with knitted .html files included for viewing. The first part is also available as a python notebook. The steps are split into 3 parts:</p> <ol> <li>Data Visualization (NEON_Pt1_DataVis.Rmd): the first part downloads phenology data from the NEON Ecological Forecasting Challenge and visualizes it as a time series using ggplot. We also download summary weather statistics using the NEON Utilities package, and join this data with the phenology data. This dataset is saved to be used again in the following parts. There is also a section that includes an example of using generative AI to improve the ggplot output for presentations. </li> <li>Forecasting Daily (NEON_Pt2_ForecastingDaily.Rmd): the second part uses daily values to build a linear model and fit it to our data and future predictions </li> <li>Forecasting Annual (NEON_Pt3_ForecastingAnn.Rmd): the third part uses phenocamR data to look at annual metrics of phenology, the spring transition dates. We relate these to annual weather metrics, exploring whether winter temperatures cause an earlier spring green up. </li> </ol> <p>Supplementary material:</p> <ul> <li>NEON_workflow_Rmd.ipynb - jupyter notebook version of the first part of part 1. This can be opened in Google Colab or Jupyter. </li> <li>cheat-sheet-neonUtilities_0.pdf - helpful overview page of neonUtilities in R, created by NEON (source: <a href="https://github.com/NEONScience/NEON-utilities/blob/main/cheat-sheet-neonUtilities.pdf">https://github.com/NEONScience/NEON-utilities/blob/main/cheat-sheet-neonUtilities.pdf</a>)</li> <li>gcc_neon_join.csv is an example of a data set that includes daily GCC_90 from NEON forecasting download and summary temperature data from NEON Utilities download. </li> </ul> <p> </p> <p><strong>More Resources</strong></p> <p>The steps followed in the forecasting workflow are from the <a href="https://projects.ecoforecast.org/neon4cast-ci/">NEON Ecological Forecast Challenge</a>, a platform to bring together ecology and data science for forecasting data from the <a href="https://www.neonscience.org/">National Ecological Observatory Network</a>, created by the <a href="https://ecoforecast.org/">Ecological Forecasting Initiative</a>.</p> <p>The R packages used include:</p> <ul> <li>tidyverse: <a href="https://tidyverse.org/">https://tidyverse.org/</a></li> <li>neonUtilities: <a href="https://www.neonscience.org/resources/learning-hub/tutorials/neondatastackr">https://www.neonscience.org/resources/learning-hub/tutorials/neondatastackr</a></li> <li>phenocamR: <a href="https://bluegreen-labs.github.io/phenocamr/">https://bluegreen-labs.github.io/phenocamr/</a></li> </ul> <p> </p> <p><strong>Who is this for?</strong></p> <p>This resource is intended for anyone who is new to using R code for working with ecological data, specifically from NEON sites. It was developed while working with undergraduate STEM students on a semester-long research project. Examples of student projects can be found here:</p> <ul> <li>Modeling Peak Photosynthesis in the Great Smoky Mountains [<a href="https://digitalcommons.humboldt.edu/ideafest2025/121/">poster</a>]</li> <li>Change in Snowmelt Timing and its Effects on the Green-up Timing of Mixed Conifer Forests [<a href="https://digitalcommons.humboldt.edu/ideafest2025/69/">poster</a>]</li> </ul> <p>Instructors are welcome to adapt the material for teaching purposes,  and students are welcome to use the material for self-learning.</p> <p> </p> <p><strong>Acknowledgements</strong></p> <p>This work was funded by the Exemplary Pathways to STEM Graduate Education grant from the Alfred P. Sloan Foundation.</p>