Hyper-Spatial Remote Sensing
Abstract
Large volumes of sub-meter commercial satellite data coupled with high performance computing & machine learning provide new insights for Earth Science. This includes digital elevation retrievals at the meter x-y scale, the ability to map trees & bushes over large areas with high accuracy, and the possibility to map other features with high accuracy using machine learning. Several examples of these new possibilities will be discussed & illustrated with recent successful studies.
Date and Time
September 12, 2024, 1:00 - 2:00 PM EST
Link
Speaker Biography
Dr. Tucker specializes in studying the earth with satellite data. He was among the first researchers to employ coarse-resolution satellite data to exploit the time domain for studying global photosynthesis on land, determining land cover, monitoring droughts, providing famine early warning, and predicting ecologically-coupled disease outbreaks. He has also used large quantities of Landsat data to study forest condition, deforestation, and forest fragmentation in temperate, subtropical, and tropical forests; and to study glacier extent. From 2005 to 2010, he was on NASA detail to the U.S. Global Change Program where he was the co-chairperson of two Interagency Working Groups, for Observations and for Land Use and Land Cover Change. He was active in NASA’s Space Archaeology Program and has conducted ground-based radar and magnetic surveys at Troy, the Granicus River Valley, and Gordion in Turkey, with University of Pennsylvania projects working at these locations from 2001 to 2012. Since 2014, he has been active in mapping land and forest degradation and attempting to quantify arid and semi-arid woody biomass using Landsat, MODIS, and large volumes of commercial satellite data. More recently, Tucker and colleagues have mastered mapping billions of trees and converting these into carbon at tree-level using machine learning coupled to high-performance computing. https://science.gsfc.nasa.gov/sci/bio/compton.j.tucker