A hybrid approach to automated landsat pixel quality

Products derived from the MODIS sensor have typically been bundled with pixel quality assurance rasters/grids. This has enabled many researchers to pick and choose which pixels are to be included in their analysis. However, the same can´t be said for imagery acquired from Landsat 5 and 7 satellites. Researchers typically would employ their own algorithms in order to exclude erroneous pixels from their analysis. Or simply not include specific scenes if they were too cloudy. From a time-series perspective, there may be plenty of non-erroneous pixels that could be utilised as part of their analysis. By having a ready-made product, researches can spend more of their time focusing on the analysis at hand. The focus of this project was to develop a hybrid pixel quality product for Landsat 5 TM and 7 ETM+ sensors, targeting factors such as saturation, spectral contiguity, cloud and cloud shadow.