High Quality Cloud, Cloud-Shadow, Water, and Snow/Ice Masks for Landsat 8
Joe Hughes, Robert, Kennedy, Rakesh Solanki
Locations were selected by taking one scene from each of the 14 World Wildlife Fund terrestrial Major Habitat Types plus 'Inland Water' and 'Rock and Ice' for each of the 7 Biogeographical Realms (64 scenes, not all habitat types occur in each Realm). Plus, an additional scene for each of the MHTs chosen randomly (originally intended for testing of algorithms).
The manuscript describing this dataset, as well as the algorithm that uses it, can be found at:
Hughes, M. Joseph and Kennedy, Robert E. (2019). "High Quality Cloud Masking
of Landsat 8 Imagery Using Convolutional Neural Networks". Remote Sensing
(2019) Vol 11(21), 2591. MDPI. doi: 10.3390/rs11212591. https://www.mdpi.com/2072-4292/11/21/2591
*_labels.tif: a coded cloud, shadow, water, ice mask (see below for codes).
*_data.tif: a 10-band geotiff of original USGS data for the subscene for bands 1-10.
*_mtl.txt: the original metadata file for the entire scene (the extents have not been changed for the subscenes)
*_photo.png: the image sent to the interpreter
*_qmask.tif: the original USGS quality mask for the subscene
*_c1bqa.tif: the quality mask associated with the Collection 1 data for the subscene. More information about the differences between these two quality masks can be found at: QA Tools Userguide. Note that LC81130672013142LGN01 was not included in Collection 1 and does not have this file.
