Speaker Curtis Mobley
Affiliation Sequoia Scientific
Host External Research; Earth, Energy and Environment
Date recorded 1 December 2010
During the last 40 years, satellite “ocean color” remote sensing has revolutionized our understanding of the global ocean at kilometer spatial scales. However, the era of multispectral sensors (5-10 wavelengths) and correlational algorithms is coming to an end. Recent work shows the value of hyperspectral sensors (~100 wavelengths) and radiative-transfer-based retrieval algorithms that relate light measurements to environmental information. For example, hyperspectral imagery can generate accurate maps of water quality parameters, water depth, and bottom type in optically shallow waters at the meter scale needed for management of coastal ecosystems and military operations. Not surprisingly, the computational requirements for hyperspectral image processing at high-spatial-resolution are much greater than for low-resolution multispectral images. After reviewing traditional multispectral methodologies for remote sensing of quantities such as chlorophyll, I will survey recently developed hyperspectral techniques for retrieval of depth and bottom type. Present computational requirements will be outlined, and anticipated future computational requirements for global-scale hyperspectral remote sensing will be estimated.
©2010 Microsoft Corporation. All rights reserved.