Title: Global Flood Risk from Advanced Modeling and Remote Sensing in Collaboration with Google Earth Engine
Presenting Author: G. Robert Brakenridge / Albert Kettner
Organization: Dartmouth Flood Observatory, CSDMS/INSTAAR, University of Colorado
Co-Author(s):
Albert J. Kettner (Univ. of Colorado) Guy Schumann (Remote Sensing Solutions Inc.) Konstantinos Andreadis (NASA JPL), Tyler Erickson (Google Inc.) Paul Bates (Univ. of Bristol, UK)

Abstract:
We present initial results of an Advanced Information Systems Technology (AIST) project. Its objective is to develop a remote sensing/modeling/data processing system that produces global flood hazard geospatial information (computer models and EO data) and distributes them to the international public through Google, Inc.'s Google Earth Engine (GEE). Specifically, a global flood hazard model is created by calibrating and validating existing state-of-the-art hydrology and hydrodynamic models (Fig. 1), and is calibrated by integrating NASA satellite observational data of actual floods (Fig. 2). . The Dartmouth Flood Observatory (DFO) at the University of Colorado has collected NASA EO data during large floods worldwide since 1993, and especially following Terra launch in late 1999. A global archive now exists (several hundred major events occur each year). DFO classifies water areas in the image data, and transforms this information into geocoded GIS files showing flood inundation limits. This information can be compared to the increasingly accurate, hydrologic- and hydrodynamic-model based simulation of floods of particular estimated recurrence intervals (e.g. the 100 yr or the 25 yr flood). Such simulation became possible as high quality global topographic data became available; for example, from the NASA SRTM mission and techniques were developed to 'hydrologically condition' these data. The AIST project also makes use of the GEE technology and computation capability to create and distribute the validated flood hazard information to users around the world. Advanced geospatial technologies used include cloud computing, big data analytics using GEE, and innovative solutions for building and integrating the two hydrologic process models. The initial objective is to deliver high resolution flood hazard maps (spatial resolution of 100 m) for three continents (Australia, Africa and CONUS), and as based on current and historic flood event hydrology going back 50 years in some cases. This is a novel combination of 1) synergetic remote sensing techniques (including optical sensors such as MODIS for flood mapping, and microwave sensors such as AMSR-E, AMSR-2, and GPM) for at-a-site flood hydrographs; 2) the linked hydrologic (water runoff volume) and hydrodynamic (flow routing) models, and 3) transformation of the mapped and modeled flood hazard areas to a uniform spatial scale within GEE. Here we show first results over all of Australia and for selected 'supersites' in Africa and the US. With the use of advanced NASA observational data, geospatial technology and data analytics, we can now deliver unprecedented information on floods for local decision-making and over spatial scales that were not possible a few years ago. Such information is especially critical in rapidly developing regions of the world where there is a rapidly rising toll of economic losses and human suffering caused by major flood events.