Architecting an Earth Observation Strategy for Disaster Risk Management

EarthzineEarth Science Informatics, Informatics Theme, Original, Themed Articles

Building on results of the Group on Earth Observations (GEO) Disasters Societal Benefit Area analysis, the Working Group on Information Systems and Services (WGISS) project analyzed how satellite, sensor, and modeling systems interact so that satellite data providers can better meet user needs, ultimately enabling a broad spectrum of users to more fully utilize satellite data and services.

photo 1

Composite image of Radarsat-2 water level (red) and extracted Open Street Map baseline water (light blue) as tiled layers on Google Earth depicts flooding from Hurricane Isaac in Haiti, 2012. Image Credit: Stuart Frye, SGT Inc. for NASA Flood Sensor Web Pilot.

åÊ

Satellites provide a unique perspective for monitoring natural hazards, providing both global and regional information to support analysis and forecasts as well as decision-making activities of emergency management personnel. Geosynchronous sensors can provide continuous regional information on a continental scale in near real time, as frequently seen in weather monitoring by National Oceanic and Atmospheric Administration (NOAA) operational satellites.

NASA is examining concepts for monitoring atmospheric and coastal events from geosynchronous orbit, potentially sharing a communications satellite as a host payload. Polar-orbiting satellites, such as the experimental EO-1, can provide high-resolution information for targeted events that are within its field of view and revisit frequency, typically every 16 days. However, providing relevant, actionable products to decision-makers requires many steps to be worked out in advance.

A GEOSS Architecture for Disasters report was released in December 2013.

A GEOSS Architecture for Disasters report was released in December 2013.

Enhanced satellite data support to disaster risk management requires timely delivery and streamlined access to products that are customized to specific applications and regions. The goals of the Global Earth Observing System of Systems (GEOSS) architecture for disasters management were to lower barriers to entry for users and suppliers of satellite products, to reduce redundancies and gaps in inter-organizational systems, and to assist in managing and prioritizing information and computing resources. Resulting systems must interoperate, while sustaining capabilities for the long term, including adapting to new technologies and evolving user needs.

Building on results of the Group on Earth Observations (GEO) Disasters Societal Benefit Area analysis, the Working Group on Information Systems and Services (WGISS) project analyzed how satellite, sensor, and modeling systems interact so that satellite data providers can better meet user needs, ultimately enabling a broad spectrum of users to more fully utilize satellite data and services. The main audience for the resulting architecture document includes providers of satellite and other relevant data for disaster management, value-added service providers who transform and distribute data into information products for end users, and managers who prioritize investments in remote sensing data acquisition and use.

The purpose of the resulting reference model is to provide an enterprise perspective for comprehending the role of diverse distributed systems and services for disaster management. By providing a common vocabulary to describe the system-of-systems building blocks and how they are combined to support disaster response, the reference model assists satellite system planners and managers to more clearly see what resources are shared, what is missing, and where there are interdependencies when providing useful products to the disaster-management community. Through analysis of several disaster response case studies involving satellite data, the project strove to achieve a common understanding of processes involved, information and computing resources employed, and user needs.

The WGISS project culminated in the release of an architecture document in December 2013, after review by members of WGISS, the Committee on Earth Observation Satellites (CEOS) disasters study group, and the United Nations Space-based Information Platform for Disaster and Emergency Response (UN-SPIDER). This report is currently accessible from the WGISS GEOSS architecture for disasters project website.

The report provides a balanced emphasis across all phases of disaster management and risk assessment: warning, response, recovery, and mitigation. It draws on insights gained from, and shared with disaster management and technology practitioners throughout the project. It provides guidance for creating systems with streamlined, sustainable capability, especially for access to remote and in situ sensor data, sensor tasking, data catalogs and archives, forecasting and simulation, and data analysis, reduction, and visualization. It also has begun to serve as a framework for assessing roles and impacts of emerging Earth science technologies in disaster management.

Key architecture functions and potential roles of emerging technologies in disaster management activities. Image Credit: John Evans, GST.

Key architecture functions and potential roles of emerging technologies in disaster management activities.åÊImage Credit: John Evans, GST.

The architecture bases its scope, purpose, and structure on the ISO/IEEE Reference Model of Open Distributed Processing (RM-ODP). It is consistent with GEOSS principles (System of Systems, Data Sharing Principles, Interoperability Arrangements) and extends prior CEOS work by reusing their definitions of disaster lifecycle phases and their scope and structure for hazard types such as flooding, earthquakes, volcanoes, drought, windstorms, landslides, wildfires, and tsunamis.

The architecture traces a set of activities, or business processes, shared across many examples of satellite information support to disaster management:

1. Detect and/or predict events based on global or regional monitoring, models, or reports from users

2. Monitor operations, with åÊa shared awareness of a dynamic situation, enabling timely decisions about data assimilation, analysis, and dissemination

3. Task Sensors and acquire other data for high-resolution observations of areas threatened or impacted by a disaster event

4. Model and predict to pinpoint priority times and locations for response and recovery efforts, and more fully understand the natural phenomena

5. Analyze and interpret data obtained via satellite or in situ sensors or other sources (including validating the resulting information products)

6. Disseminate visual or other products to end users, including reports or updates

7. Allow user access to all activities, so users can draw upon, or even shape, the gathering, processing, or production of information.

The architecture identifies three stages of related activities required in order to establish a disaster management capability to perform the activities listed above:

‰Û¢åÊInitiation: identify inputs for event detection, and event triggers; choose indicators for situational awareness (e.g., flood extent); define modeling elements (e.g., regional flood model); and develop workflows; data flows for processing and delivery; and automation goals (e.g., subscriptions, custom products)

‰Û¢åÊOperational event detection and response: monitor data streams, detect events and trigger workflows; track key indicators; task sensors; acquire data; run models (hindcast, nowcast, forecast); analyze and disseminate products

‰Û¢åÊDisaster recovery and mitigation: perform damage assessment, surveys of reconstruction/ recovery; review historical data to quantify, locate, and refine risk; and conduct research toward improved preparedness.

The topics discussed above constitute the enterprise view of the reference model, capturing the scope and structure of the activities, the stakeholders and the operating principles. The report also addresses the information and computational viewpoints, while only alluding to the engineering and technology viewpoints comprising the full RM-ODP definition.

The architecture is being used to assess the role and impact of potential ‰ÛÏgame-changing‰Û technology in supporting disaster management with Earth science information. Examples of such technologies include mobile devices, including crowdsourcing and Location-Based Services; adaptive sensing using small, cheap, ubiquitous sensors, sometimes known as an ‰ÛÏInternet of things‰Û, including unmanned aerial vehicles and small satellites; new approaches to data fusion such as cloud computing, big data, and model webs; and enhanced collaboration tools, including semantic reconciliation services.

In the example below, students from the University of Namibia work with Namibian hydrologists and NASA researchers to collect in situ data to calibrate radar and imaging sensor data and to improve detection of water in grassy marsh.

Namibia Department of Hydrology staff and students prepare to make GPS readings using mobile phones for the Kavango River water extent validation exercise. Image Credit: Dan Mandl, NASA Goddard Space Flight Center for NASA Flood Sensor Web Pilot.

Namibia Department of Hydrology staff and students prepare to make GPS readings using mobile phones for the Kavango River water extent validation exercise.
åÊImage Credit: Dan Mandl, NASA Goddard Space Flight Center for NASA Flood Sensor Web Pilot.

Composite image of Radarsat-2 and EO-1 vector products, and crowd-sourced data points as tiled layers on Microsoft Bing Maps depicts water extent of the Kavango River in northern Namibia. Image Credit: Stuart Frye, SGT Inc. for NASA Flood Sensor Web Pilot.

Composite image of Radarsat-2 and EO-1 vector products, and crowd-sourced data points as tiled layers on Microsoft Bing Maps depicts water extent of the Kavango River in northern Namibia.åÊImage Credit: Stuart Frye, SGT Inc. for NASA Flood Sensor Web Pilot.

The resulting image depicts water edge as detected by satellite (Radarsat/yellow, EO-1/red). On the upper shore, red Xs depict student recordings showing a good correlation with Radarsat.åÊ On the opposite shore, green Xs show student recordings on the riverbank, which the Radarsat algorithm failed to detect due to marsh grasses. A georeferenced helicopter photo augments the data record. The georeferenced photo enables experts to train classifier algorithm to detect the presence of water in grassy marsh lands from satellite data.

Complementing satellite data with targeted new technology will greatly increase its utility. Crowd-sourced data from smart mobile devices and processing the satellite data stream on cloud computing resources are two such trends improving the outcomes of the disasters management architecture.

Author bios

Karen Moe is a technology development manager at NASA’s Earth Science Technology Office in Greenbelt, Maryland. Her work has focused on information architecture and the role of information technology in achieving NASA’s Earth science objectives. She can be reached at karen.moe at nasa.gov.

John Evans is a principal computer scientist at Global Science & Technology Inc. in Greenbelt, Maryland. His primary interests are interoperating geospatial information systems and broad access to Earth science data and computing resources. He can be reached at john.evans at gst.com.