8.2 Interactive Computing and Processing of NASA Land Surface Observations using Google Earth Engine

Wednesday, 13 January 2016: 10:45 AM
Room 348/349 ( New Orleans Ernest N. Morial Convention Center)
Andrew L. Molthan, NASA/MSFC, Huntsville, AL; and J. E. Burks and J. R. Bell

Remote sensing offers numerous means of identifying and documenting damage to vegetation that results from severe convective storms, including tornadoes, damaging winds, and hail, and also other hazards such as flooding and the expanse of wildfires. Impacts to seasonal vegetation, such as planted crops, can be limited to a single season while longer-lived damage can be noted in areas such as a stand of trees or large, forested area. Since many remote sensing imagers provide their highest spatial resolution bands in the red and near-infrared to support monitoring of vegetation, these impacts can be readily identified as short-term decreases in common vegetation indices such as NDVI, along with increases in land surface temperature that are observed at a reduced spatial resolution. The ability to identify an area of vegetation change is improved by understanding the conditions that are normal for a given time of year and location, along with a typical range of variability in a given parameter. Statistical information acquired from a multi-year view of vegetation and land surface temperature can result in a large volume of data to manage and process, such as acquiring the period of record from sensors such as NASA's MODIS (aboard Terra and Aqua) or higher resolution imagers from the Landsat series of satellites. Google's Earth Engine offers a “big data” solution to these challenges, by providing a streamlined API and option to process the period of record of NASA MODIS and Landsat products through relatively simple Javascript coding. This presentation will highlight efforts to date in using Earth Engine holdings to produce vegetation and land surface temperature anomalies that are associated with damage to agricultural and other vegetation caused by severe thunderstorms across the Central and Southeastern United States. Earth Engine applications will show how large data holdings can be used to map severe weather damage, ascertain longer-term impacts, and share best practices learned and challenges with applying Earth Engine holdings to the analysis of severe weather damage. Other applications are also demonstrated, such as use of Earth Engine to prepare pre-event composites that can be used to subjectively identify other severe weather impacts. Future extension to flooding and wildfires is also proposed.
- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner