ESIP
Browse

Deep Learning forecasts the spatial extent, but not the severity, of forest mortality in California

Download (795.23 kB)
poster
posted on 2025-01-14, 12:05 authored by Keenan Ganz, L. Monika MoskalL. Monika Moskal

In the American west, trees are killed by logging, fire, drought, and insects. Some tree death is normal, but climate change is making trees more likely to die from drought and insects in particular. In California, drought and insects killed 1-5% of all living trees between 2012 and 2016. Although scientists have studied drought and insect attack in individual trees, we still struggle to forecast where and when entire forests will die in the American west. One idea to improve these forecasts is to use deep learning. Deep learning is a way to teach a computer to forecast future forest death by studying where forests have died in the past. Past research on forecasting forest death did not utilize deep learning, so we think that this technique will help us make a more accurate map of predicted forest death over the entire western United States, benefitting both forest managers and other scientists.

Funding

Graduate Research Fellowship Program (GRFP)

Directorate for Education & Human Resources

Find out more...

History

Usage metrics

    ESIP January 2025

    Licence

    Exports

    RefWorks
    BibTeX
    Ref. manager
    Endnote
    DataCite
    NLM
    DC