Plenary Talks

Talk recordings are now available here!
Monitoring Drought with Google Earth Engine: From Archives to Answers

Dr. Justin Huntington, Desert Research Institute / Western Regional Climate Center

Drought has adverse effects on society through reduced water availability, ecological biodiversity, agricultural production, and increased wildfire and crop-failure risk. Satellite imagery can be used to monitor historical and near real-time drought conditions by visualizing vegetation, snow, water, fire, and thermal indices computed from optical and thermal imagery. Gridded observations of climate and meteorology can also be used and paired with satellite imagery to provide information about the causality and intensity of drought. However, despite the wealth of earth observations, tools to quickly access, compute, and visualize archives, and provide answers at relevant scales for improved decision-making are lacking. The Desert Research Institute and University of Idaho, in partnership with Google, has developed, a web application that uses Google’s Earth Engine platform to enable users to quickly compute and visualize drought products for improved place-based monitoring and early warning of drought. The application is currently being used by U.S. federal agencies and researchers to develop baseline conditions and impact assessments related to agricultural, ecological, meteorological, and hydrological drought. Internationally, the team is working with the Famine Early Warning Systems Network (FEWS NET) to develop fast and versatile tools that use Earth Engine for monitoring agricultural drought over broad areas at risk of food insecurity.

About the speaker: Justin Huntington is an associate research professor of Hydrology at the Desert Research Institute and Western Regional Climate Center, Reno, Nevada. His research interests are focused on remote sensing, land surface energy balance measurement and modeling, drought monitoring, and integrated hydrologic modeling. He is one of 25 members of the Landsat Science Team, and is a member of Nevada Governor Brian Sandoval’s Drought Forum.
Agriculture in the Sentinel era: scaling up with Earth Engine

Guido Lemoine, European Commission's Joint Research Centre

The European Copernicus program has launched 4 Earth Observation platforms since April 2014. Operational data streams from the Sentinel-1A, a C-band SAR, and Sentinel-2A, an optical and infrared sensor are provided under a full, free and open license since October 2014 (S1A) and August 2015 (S2A). These streams will soon be complemented with those from Sentinel 3A (a medium resolution sensor launched in February 2016) and Sentinel-1B, the twin of S-1A (launched April 2016). In the last year, the Earth Engine team has started with the ingestion of S1A and S2A data streams into the Engine, exposing it as 2 hosted data catalogues to the growing community of Earth Engine users. The unique combination of frequent, all-weather SAR S1 acquisitions and 10 multispectral band S2 at 10-20m resolution, and their combination with existing Landsat data sets and lower resolution MODIS time series provides exciting new opportunities for applications in crop monitoring. In the presentation, we discuss some dedicated processing methods for S1 data (e.g. speckle filtering, time series analysis) and S2 (e.g. red-edge analysis) and their combined use in large classification and validation experiments. On the basis of our results, and the further expansion of the Sentinel fleet in 2017 with Sentinel-2B and -3B, we propose some directions for future development (constrained segmentation, multi-annual change analysis, integration of crop rotation knowledge, links to ‘deep learning’ issues, etc.).

About the speaker: Guido Lemoine is a Principal Scientist in the Monitoring Agricultural ReSources (MARS) Unit of the European Commission’s Joint Research Centre (JRC). He is an agricultural engineer by training (1987, Soil Science, Wageningen University, The Netherlands). His main expertise is in applied remote sensing, first developed as a research topic (microwave backscattering of soils) and later as a commercial activity in a co-founded remote sensing and GIS consultancy. After he joined the JRC, in 1997, he has further developed his remote sensing and informatics expertise in agricultural statistics and subsidy control applications, in fisheries monitoring, and in civil security applications (Copernicus emergency services). In 2014, he returned to the MARS unit to help develop the uptake of the European Copernicus Sentinel in global agricultural monitoring applications.
Accelerating Rangeland Conservation

Dr. Brady AllredUniversity of Montana

In 2010, the USDA Natural Resources Conservation Service (NRCS) launched the Sage Grouse Initiative (SGI; to voluntarily reduce threats facing sage-grouse and rangelands on private lands. Over the past five years, SGI has matured into a primary catalyst for rangeland and wildlife conservation across the North America west, focusing on the shared vision of wildlife conservation through sustainable working landscapes and providing win-win solutions for producers, sage grouse, and 350 other sagebrush obligate species.

Moving forward, SGI continues to focus on rangeland conservation. Google Earth Engine provides new opportunities for outcome monitoring and conservation planning at continental scales. The SGI science team is currently utilizing Google Earth Engine to develop assessment and monitoring algorithms of key conservation indicators. The SGI web application ( will utilize Google Earth Engine for user defined analysis and planning, putting the appropriate information directly into the hands of managers and conservationists.

About the speaker: Dr. Brady Allred is an assistant professor of rangeland ecology at the University of Montana. His research interests include rangeland dynamics, processes, and structure at broad scales. He uses earth observation data to advance the science and practice of rangeland conservation.
Flood Vulnerability from the Cloud to the Street (and back!) powered by Google Earth Engine

Beth TellmanArizona State University and Cloud to Street

21 million people are exposed to flooding every year, and that number is expected to more than double by 2030. This talk focuses on approaches to make satellite data meaningful to understand flood vulnerability in a changing world by leveraging Earth Engine. Cloud to Street, a mission driven science organization, has used Earth Engine to develop vulnerability assessments for New York State, Uttarakhand India, and Senegal with non-profits and development banks. I’ll discuss the science we’ve been developing in these projects, from automated flood detection that powers machine learning flood models in the cloud, to collecting crowd sourced flood observations on the street, combined with social vulnerability analysis to know who is most at risk. I’ll also share experiences engaging development practitioners through events, trainings, and contributions to development reports, to increase data literacy and inclusion in a climate changed world.

About the speaker: Beth Tellman is a PhD student in geographical sciences at Arizona State University. Her research focuses understanding how environmental change shapes flood vulnerability and how illicit human activity drives land use change. She employs tools and methods from land system science, hydrology, and the social sciences to model social and natural systems across time and space using satellite, cadastral, and ancillary socioeconomic data. Current projects include examining political drivers of urbanization and water infrastructure in irregular settlements, and the historical pathway of adaptation to flood risk (in Mexico City with NSF project- MEGADAPT), estimating the return on investment of watershed conservation for flood mitigation in Latin American cities with SNAP, quantifying the effect of the cocaine commodity chain on deforestation patterns in Central America, and modeling global biophysical and social flood vulnerability using Google Earth Engine with a company she co-founded, Cloud to Street. Beth received her M.S. from Yale School of Forestry in Environmental Science and B.S in Sustainable Globalization from Santa Clara University.
Automated methods for surface water detection

Gennadii DonchytsDeltares

Surface water detection using multi-spectral satellite imagery is a relatively trivial process. However, a very accurate estimation of water mask is still a challenge and fully automated methods are very scarce. A fast-growing satellite industry necessitates such generic automated methods. Main sources of noise in multispectral imagery include clouds and cloud shadows, systematic sensor or data processing errors, but also, varying climate and topographic conditions across the globe. At research institute Deltares, we have started using Google Earth Engine platform (EE) very intensively for many applications, including but not limited to the estimation of flood extent, surface water mask detection, and hydrological analysis. This presentation demonstrates some of these projects as well several new datasets being developed actively using EE (such as global height above the nearest drainage: and a new web tool to analyze surface water changes globally at 30m resolution ( 

About the speaker: Gennadii Donchyts is a senior consultant at Deltares. He has many years of expertise in multidisciplinary software development and research related to the numerical simulation of surface and subsurface water flow processes. He obtained an MSc in radio physics, electronics and computer systems from the Kyiv National Taras Shevchenko University and is currently pursuing a Ph.D. in at the Delft University of Technology focusing on the development of new methods and datasets for the automated surface water extraction and change detection using Earth Observation datasets.