Look for complete geospatial metadata in this layer's associated xml document available from the download link

* Metric Name: Cumulative Shrub Cover Loss  

* Tier: 1  

* Data Vintage: 12/2020  

* Unit Of Measure: Cumulative fractional non-overlapping absolute shrub cover loss, where shrub cover is a continuous variable from 0 to 1. Cumulative loss can exceed 1 in cases with multiple disturbances.  

* Metric Definition and Relevance: The cumulative loss of shrub cover over a 30-year period (1992-2020). Shrub cover loss reflects fires, harvest/management and dieoff. Only disturbances that are sufficient to trigger the Continuous Change Detection and Classification algorithm are included; low-level, diffuse dieoff is likely missed.  

* Creation Method: Vegetation disturbances were identified over the Landsat TM/ETM+/OLI era using the Continuous Change Detection and Classification algorithm (CCDC). The corresponding annual change in shrub cover was determined with machine learning (Random Forest) using the National Land Cover Database for training and Landsat/CCDC observations as predictors; this produced a ~35-year stack of rasters that identified the locations and severity of shrub cover loss. This stack was then summed for 1992-2021 to calculate the cumulative tree cover loss over a 30-year period. See https://doi.org/10.1029/2021AV000654 for further information.

CECS data that reflect landscape changes resulting from disturbances require 6
to 12 months of Landsat observations **after a given year that included major
disturbances (such as a high severity wildfire)** to fully quantify that
disturbance. CECS data that reflect disturbance, such as this data layer, are
therefore available **through water year 2020 (i.e. through September 2020)**.

  

* Credits: CECS;