CLOUD MASKING USING RED AND NEAR-INFRARED BANDS

Hits: 99
Research areas:
Year:
2016
Type of Publication:
In Proceedings
Keywords:
DIWATA-1, Spaceborne Multispectral Imager, Landsat8-OLI
Authors:
Felix, Mark Jayson; Perez, Dr. Gay Jane
Abstract:
In this work, a simple cloud mask algorithm that only utilizes the red and near-infrared (NIR) bands of a multispectral imager was employed and assessed using daytime Landsat8 dataset. The cloud mask algorithm uses an ancillary land/water mask generated using a threshold applied to the calculated NDVI values. Accuracy assessment using the quality assessment band provided in each Landsat8 dataset shows a good agreement on the computed cloud cover area percentage. Moreover, the red-NIR cloud mask algorithm for each processed Landsat8 dataset achieved an overall accuracy no less than 90%. Further analysis showed undetected cloud pixels by the quality assessment band of Landsat8 dataset which are detected by the red-NIR cloud mask algorithm. This shows the potential efficacy of the red-NIR cloud mask algorithm for satellites without onboard thermal imager such as Philippines’ first earth-observation microsatellite named DIWATA-1.
Full text: Ab 0581.pdf [Bibtex]
You are here: Home ACRS ACRS Overview Proceedings CLOUD MASKING USING RED AND NEAR-INFRARED BANDS