A New Data Processing System for Generating Sea Ice Surface Roughness and Cloud Mask Data Products from the Multi-Angle Imaging SpectroRadiometer (MISR)

• Main Author: mosadegh, ehsan
• Other Authors: Nolin, Anne W., Samburova, Vera, Wilcox, Eric, Yan, Feng, Albright, Thomas P.
• Format: Doctoral or Postdoctoral Thesis
• Language: unknown
• Published: 2023
• Subjects: Arctic; cloud detection and classification; machine learning; MISR; sea ice; surface roughness; Airborne Topographic Mapper; Sea ice
• Online Access: http://hdl.handle.net/11714/9057

This study describes two novel data products derived from Multi-angle Imaging SpectroRadiometer (MISR) imagery: Arctic-wide maps of sea ice roughness and a binary cloud detection algorithm. The sea ice roughness maps were generated using a data processing system that matched MISR pixels with co-located and concurrent lidar-derived roughness measurements from Airborne Topographic Mapper (ATM), calibrated the multi- angle data to values of surface roughness using a K-Nearest Neighbor (KNN) algorithm, and then applied the algorithm to Arctic-wide MISR data for two 16-day periods in April and July 2016. The resulting maps show good agreement with independent ATM roughness data and enable characterization of the roughness of different ice types. The binary cloud detection algorithm was developed using a neural network approach and a training dataset constructed from Top-of-Atmosphere red band values from all MISR’s nine different viewing cameras for the same two months in various regions of the Arctic. The algorithm showed good performance in classifying pixels into cloudy and clear categories in MISR images, with better performance for clear pixels in April 2016 and better performance for cloudy pixels in July 2016. The algorithm also provides a significant advantage over existing MISR cloud mask products SDCM and ASCM in terms of accuracy and spatial resolution, with a resolution of 275 meters. The data products presented here can be used to gain insights into the seasonal and interannual changes in sea ice roughness and cloud cover over the Arctic and to develop and improve more accurate classification algorithms in the field of remote sensing.