Investigating the Spatial Distribution of Scotch Broom (Cytisus scoparius) in Garry Oak (Quercus garryana) Dominant Regions Using Sentinel-2 Multispectral Imagery

The dataset was derived from Sentinel-2 multispectral imagery and includes both raw spectral bands and derived vegetation indices. Predictor variables include surface reflectance bands (B01–B12), Tasseled Cap Transformations (brightness and greenness), and temporal change metrics in vegetation indices (NDVI - Normalized Vegetation Index, ciRED - Red-Edge Chlorophyll Index, and TREI - Red-Edge Chlorophyll Index) calculated across seasonal intervals (January–April, April–September, and January–September).

These variables were used as inputs to a Random Forest classification model to identify areas of Scotch broom presence and absence. The dataset also includes model outputs as continuous probability surfaces and binary classification maps, generated via thresholding approaches (e.g., Youden’s index and high-confidence thresholds). Accuracy assessment was conducted using validation data, with performance evaluated through confusion matrices, overall accuracy, and class-specific producers’ and users’ accuracies.