Landtrendr Tuning

Summary

The Landtrendr (LT) disturbance detection algorithm appears in many different CAFRI projects (AGB, Shrubland), and is a powerful tool in the pocket of anyone trying to look at historical forest disturbance at a landscape scale. However, the algorithm is not regionally specific and was not initially trained or tested for the disturbance regimes or management practices of eastern forests. This project seeks to meet that need by creating a version of the algorithm specifically tuned for the northern forest region.

Nearly all change detection algorithms, LT included, work in a similar fashion. They take inputs of spectral reflectance values from time series of aerial images and look for deviations from the average trajectory to mark as disturbances. The LT algorithm specifically used Landsat imagery and fits linear segments to the time series. While some algorithms offer more detailed predictions, the LT algorithm can identify canopy disturbances at an annual time scale, as well as estimates of disturbance magnitude, canopy decline, and growth and recovery trajectories.

Through our partnerships with forest managers and private land owners we have access to extensive forest harvest records spanning the entirety of the Adirondack Park. This mixed use landscape is the ideal location for validating these disturbance outputs because thanks to its ‘Forever Wild’ designation there are areas that we know for sure have not been harvested in the last 30 years. The goal is to use these harvest records, along with data from a field survey conducted in the summer of 2022 as reference data in hyper parameter tuning of the Landtrendr algorithm.

People

Some Papers

  • Kennedy, R. E., Yang, Z., & Cohen, W. B. (2010). Detecting trends in forest disturbance and recovery using yearly Landsat time series: 1. LandTrendr — Temporal segmentation algorithms. Remote Sensing of Environment, 114(12), 2897–2910.
  • Zhu, Z. (2017). Change detection using landsat time series: A review of frequencies, preprocessing, algorithms, and applications. ISPRS Journal of Photogrammetry and Remote Sensing, 130, 370–384.
  • Desrochers, M. L., Tripp, W., Logan, S., Bevilacqua, E., Johnson, L., & Beier, C. M. (2022). Ground-Truthing Forest Change Detection Algorithms in Working Forests of the US Northeast. Journal of Forestry.