All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 193 33 9
PDF Downloads 63 14 2

The Image Navigation Cloud Mask for the Multiangle Imaging Spectroradiometer (MISR)

Larry Di GirolamoDepartment of Atmospheric and Oceanic Sciences, McGill University, Montreal, Canada

Search for other papers by Larry Di Girolamo in
Current site
Google Scholar
PubMed
Close
and
Roger DaviesDepartment of Atmospheric and Oceanic Sciences, McGill University, Montreal, Canada

Search for other papers by Roger Davies in
Current site
Google Scholar
PubMed
Close
Full access

Abstract

The authors have developed a cloud mask technique that may be applied to the efficient selection of “clear enough” scenes for image navigation. While the mask can be applied generally, the motivation for its development comes from its intended use on Multiangle Imaging Spectroradiometer (MISR) imagery. The difficulties in detecting clouds in the presence of land–water boundaries when using prenavigated imagery is overcome by using a simple two-step direct threshold technique. The two steps involve the thresholding of two observables derived for each pixel. The first is a 0.86-μm reflectance. The second is a new observable, D = | NDVI |bR1−2, where NDVI = (R2R1)(R2 + R1)−1, R2 is the 0.86-μm reflectance, R1 is the 0.67-μm reflectance, and b is chosen so as to maximize the separation between clear and cloudy pixels. The success of the cloud mask is shown by applying it to degraded AVIRIS data. The authors make comparisons with a more popular NDVI technique to show the advantage of our method.

Abstract

The authors have developed a cloud mask technique that may be applied to the efficient selection of “clear enough” scenes for image navigation. While the mask can be applied generally, the motivation for its development comes from its intended use on Multiangle Imaging Spectroradiometer (MISR) imagery. The difficulties in detecting clouds in the presence of land–water boundaries when using prenavigated imagery is overcome by using a simple two-step direct threshold technique. The two steps involve the thresholding of two observables derived for each pixel. The first is a 0.86-μm reflectance. The second is a new observable, D = | NDVI |bR1−2, where NDVI = (R2R1)(R2 + R1)−1, R2 is the 0.86-μm reflectance, R1 is the 0.67-μm reflectance, and b is chosen so as to maximize the separation between clear and cloudy pixels. The success of the cloud mask is shown by applying it to degraded AVIRIS data. The authors make comparisons with a more popular NDVI technique to show the advantage of our method.

Save