Greenland snowmelt estimation using multi-spectral passive and active microwave observations

Principal component analysis (PCA) has previously been used to characterize snowmelt on the Greenland ice sheet using all seven SSM/I channels. Three extensions to this algorithm are presented in this work. First, a location dependent thresholding technique is developed that facilitates improved sensitivity in snowmelt detection as compared to prior studies. Second, an adapted PCA algorithm is formulated that is able to link key physical processes with vectors in the SSM/I data eigenspace. Finally, the inclusion of active scatterometer data in the PCA implementation is shown to offer additional information that assists in snowmelt estimation.