This dataset presents Harmonized Landsat Sentinel-2 (HLS) tiles labeled with forest structural information at ~90 sites (minimum 50 m by 500 m) in the boreal forests of Alaska and Northwest Canada. UAV-LiDAR data was collected at each site between 2022 and 2024. Canopy height and crown coverage were derived from the point clouds. The HLS pixels were collected as time series for two seasons: peak summer (June 20th to August 10th 2022, 2023, 2024) with 648 tiles, and late summer (August 20th to September 30th, 2022, 2023, 2024) with 258 tiles. We applied a 30 % cloud cover filter and masked remaining clouds, water bodies, and snow with the HLS Fmask bit filters. We computed and added one NDVI band per season. The median pixel value was taken from the median composites at each site. We resampled the crown coverage and median canopy height with the HLS grid resolution (30 m) and stacked the rasters together.

The dataset contains: • HLS and forest structure as Cloud Optimized GeoTIFF (COG)It contains all the HLS peak summer and late summer bands, and UAV-lidar derived information. One file per site, and the associated read_me file with more details.• PolygonsPolygons shapefiles of each site and extent• Pixel tableOne table containing all band information from all pixels in the dataset, Principal components 1 to 3, and Tasseled Cap values, and the associated read_me file with more details.This dataset supports the findings of Enguehard et al., (in preparation), and can be used as training and validation data for canopy height or crown coverage prediction models in the boreal forest.