Image Analysis Provides the Intelligence Necessary to Prepare, Respond and Heal After Natural Disasters
Change detection is an exact science when using properly registered imagery. Here before and after images of the Red River flooding show areas that have been impacted.
This image shows a 3D overlay of burn severity. A DNBR slice image is draped over a DEM in Esri’s ArcScene to begin visually exploring the different regions that have been impacted by fire. Red areas have steep slopes and severe burning and are at risk for soil loss.
Users can interactively select an area and run a custom image processing tool that detects a burn zone inside the GIS environment. In this case, an ENVI tool is used within the ArcGIS environment.
Above average fire activity has plagued the southern United States since February of this year due to above normal temperatures and below normal precipitation, drying out vegetation and creating fuel for fires. Photo credit: Joe Zamudio, ITT VIS.
By Karen Richardson Esri Writer Redlands, Calif. www.esri.com
Extreme weather is wreaking havoc in North America this year. The 2011 tornado season was record-setting, beginning in April with 438 confirmed twisters. The same extreme weather patterns causing tornadoes are also leading to heavy rainfall in the midwestern region of the continent. Runoff led to serious flood conditions, blocking roads and highways, inundating thousands of acres of agricultural land and leaving buildings saturated with water. While the North flooded, winds and drought struck the South. West Texas experienced multiple fires that burned more than 1.6 million acres of brush and forest land.
Complex disasters like these require quick response. Earth observation through remote sensors can play a key role in managing information. Data can be collected in near real-time and analyzed for all aspects of disaster management, from preparation to response and remediation. In dire situations, imagery can become more than just a basemap or backdrop. When integrated with GIS, imagery’s key benefit becomes providing a rich set of derived information for advanced analysis.
Knowledge from Accurate Image Registration
Esri (Redlands, Calif.) is working with its partners to bring remotely sensed data and image processing into the GIS environment. This integration is making Earth observation imagery more easily accessible to a wider user community, which stands to benefit greatly from improved understanding of the dynamic conditions of the earth through repeat data collection and monitoring.
Esri partner PCI Geomatics (Ontario, Quebec, Canada) provides automated satellite sensor data integration into Esri’s ArcGIS environment. PCI’s GeoImaging Tools for ArcGIS automate the methods for correcting imagery, ensuring seamless integration within the GIS platform. Traditionally, these tasks were performed by scientists and others with specific expertise in remote sensing. A lack of tools and knowledge of how to integrate imagery into GIS has long been a barrier to accessing imagery.
“Even with the increased availability of remotely sensed imagery, downloading and making use of the imagery remains a daunting task for the nonexpert,” says Kevin Jones, director of marketing and product management at PCI Geomatics. “What should I do with it now? Where do I bring it in? What format is it in? How do I access the different bands? These are some of the common questions and reasons why non-remote sensing experts hesitate in downloading and making more use of satellite imagery.”
Historically, GIS users seeking to overlay imagery for their analysis needed to pre-process the imagery to ensure it would line up with existing features for a given study area. Collecting ground control points (GCPs) manually was the order of the day. Correcting imagery to remove distortions and ensure proper alignment meant manual collection of common features on referenced and non-referenced datasets. “Poor imagery registration limits a GIS user’s ability to efficiently perform simple change detection and feature extraction tasks,” says Lawrie Jordan, Esri director of imagery. “By achieving high levels of image registration accuracy, GIS users can extract features and focus on more detailed analysis.”
Image registration is the first step to seamlessly integrating data into any organization’s GIS workflow. Having the right tools to perform rigorous corrections and to ensure that multiple types of data are properly aligned for further analysis allows users to tap the knowledge they need from their imagery. Now, decision makers can use multi-temporal, multi-spectral and multi-sensoral imagery to derive information through advanced analysis of Earth observation in support of better business decisions.
How the Red River Flows
Up-to-date derived information from imagery can help disaster management authorities perform rapid response activities. Authorities not only are able to use massive amounts of remotely sensed imagery efficiently, but also have the ability to use the data for change detection, providing scientific, or evidence-based, decision making.
Two years ago, during the flood season of 2009, RADARSAT- 2, with synthetic aperture radar (SAR) imagery, was used by authorities in Manitoba, Canada, to map the extent of the Red River flood and its progression on a daily basis. SAR imagery is extremely valuable, since it can detect the presence of overland flooding even through adverse weather conditions such as cloud cover and darkness.
“Multiple satellite sensors collect imagery on a daily basis,” explains Jones. “These satellites can collect vast amounts of imagery that are transmitted to the ground and disseminated for analysis within minutes of acquisition. Integrated analysis and decision making in an integrated GIS platform like ArcGIS provide the ideal environment for making informed decisions that can help save lives and limit damage to infrastructure and property.”
An international watershed, the Red River flows north toward Manitoba along the Minnesota-North Dakota border into Lake Winnipeg. The Red River basin is a flat and highly productive agricultural area spreading across 116,500 square kilometers. Rapid changes from cool winters to warm summers, coupled with the terrain and climatic conditions, contribute to an area that is highly sensitive to spring flooding.
Geospatial Analysis Provides a Complete Picture
Many organizations were involved in monitoring and responding to the 2009 flooding. Response efforts included the use of an ArcGIS software-based decision support system that proved to be a very useful tool for collecting, storing, accessing and distributing the information pertaining to the floods. At the provincial level, the Manitoba Water Stewardship group embraced the use of Earth observations and in particular, RADARSAT to detect and monitor flood conditions on a routine basis due to the reliability of the information and timeliness of its collection and delivery.
The provincial authorities established links with the Emergency Response team at the Canada Centre for Remote Sensing (CCRS). Satellite images were collected in real time at a ground receiving station that produced an initial imagery product within minutes of acquisition. At CCRS, experts carefully analyzed the imagery to delineate flooded areas in the updated imagery. The information was then converted to a vector format and transmitted to the Water Stewardship group. See Figure 1.
“Critical decisions that affect our citizens are carefully considered by examining the information derived from the imagery,” says Michelle Methot, who is a water stewardship specialist in the Water Stewardship group. “Imagery played a critical role to help the Manitoba government make the difficult but responsible decision to make a controlled opening in a dike along the Assiniboine River. Geospatial analysis allowed us to take many factors into consideration related to the affected areas, the local population and potential for greater damage. This analysis led to informed decision making and, ultimately, to the breaching of a main dike at the Hoop and Holler bend of the Assiniboine River.”
In the case of this year’s Red River floods in Manitoba and North Dakota, Earth observation coupled with GIS is again playing a critical role in providing near real-time information for disaster response. Using repeat pass collection and all-weather, day/night SAR satellites, near real-time monitoring is possible in the flood-affected regions of Manitoba. The imagery is used to support daily analysis and to enhance situational awareness by identifying the rapid changes in the landscape as a result of flooding. Derived information is used to support decision making on assessing potential damages to communities and infrastructure. Determination of the risk for failures of dikes, and help in determining where existing flood diversion and protection infrastructure needs reinforcing, are again being provided.
Derived Information Helps Mitigation Efforts
The use of Earth observation extends past the disaster response stage. Imagery provides a record of conditions on the ground that can be used after the fact to confirm the location of flood-affected homes and businesses. This information is useful for individuals, businesses and insurance companies that process claims.
“The usefulness of imagery extends beyond providing a backdrop. Users increasingly appreciate the rich set of information that can be derived from mutli-temporal, multi-spectral, multi-sensor and multi-resolution imagery,” says Esri’s Jordan.
Education and outreach are critical for truly discovering the potential for the operational use of imagery. Through a joint partnership, McDonald, Detwiller and Associates (MDA), based in British Columbia, Canada, and the owner/operator of RADARSAT-2 and PCI Geomatics, has been developing a series of instructional webinars that focus on the utility of SAR imagery. The webinars are based on operational uses such as the Manitoba case. In addition, a dedicated website was created by PCI Geomatics to provide basic information on SAR imagery, its characteristics and utility; the site can be accessed at www.pcigeomatics.com/sar.
From Floods to Fires
While locals in Canada and the Northern United States have been battling flooding, those living in the Southwest have been putting out fires. This year, a drought in West Texas has led to large areas with extremely dry underbrush and ground cover. Current weather conditions have included persistent winds of 20 miles per hour or faster that have contributed to severe fire conditions in the area. Total burn area in the beginning of May was more than 256,000 acres from 511 wildfires, according to the United States Department of Agriculture. While the areas of the fires are sparsely populated, the disruption to local farming and livestock production is expected to be significant.
Esri partner ITT VIS (Boulder, Colo.) provides GIS analysts with the ability to visualize imagery and exploit information to face challenges like wildfires, which burn millions of acres of land each year. Working over time, ITT VIS has brought its advanced image processing software, ENVI, into the ArcGIS environment via the ENVI Tools for ArcGIS, a fully integrated ArcGIS Toolbox. Today, the two products work seamlessly together to share data and the ENVI image processing and analysis tools can be used within the ArcGIS environment. Using the toolbox, GIS users can build models that can be applied to any application on a server, and then published throughout an enterprise with Web applications anyone can use. “This capability gives GIS users access to the more real-time information from geospatial imagery for use in GIS workflows,” says Lori Thompson, vice president of marketing and support services, ITT VIS.
Staff at ITT VIS created a fire fuel tool model built for ArcGIS using ENVI image processing algorithms. The model maps the distribution of fire fuels and burn hazards using spectral imagery. Fire managers need to provide effective methods for mapping fire fuels accurately, since fuel distribution is very important for predicting fire behavior. Looking at elevation of the land, slope, aspect, and canopy cover, and developing a surface fuel model coupled with weather and wind data can provide those fighting the fires with the most accurate information in order to plan their response.
Surface fuels are the greatest concern, since they are major contributors to the intensity and spread of fires. The fire fuel tool developed highlights areas with high fire fuel by identifying areas with dry or drying plant material, a high risk factor for wildfire, as opposed to low fire fuel areas containing mainly lush, green plants.
In addition to this model, multiple ENVI tools exist to help assess vegetation damage caused by fires and extreme weather events. Standard classification and feature extraction tools help analysts to map areas of damage. Change detection tools can also automatically locate and measure damage. When damaged areas are small relative to the size of the scene, an anomaly detection tool can help quickly identify damaged areas that are different from those in the background.
Taking Back the Land
While burning fires need to be dealt with quickly and efficiently, the effects of the fire after the burn also cause significant issues. Covering hundreds of acres of land stripped of natural protection and reestablishing conservation measures, especially on rangeland such as in Texas, can be an expensive task. The importance of installing measures that reduce post-fire damage and aid in rehabilitating the area is high, as the USDA recognizes. This federal agency has made $400,000 available through its Environmental Quality Incentives Program to help Texans reestablish the land. This funding can be used to rebuild fences, defer livestock grazing and bring stability to farmers’ sometimes precarious operations, allowing the land to heal.
Imagery can again assist in allocating resources when erosion remediation is necessary after a severe burn such as that in Texas. Since all agencies operate under cost controls, derived information from remotely sensed data can be used to prioritize which regions receive post-fire mitigation first.
Using the ENVI image processing tools available for ArcGIS Server, ITT VIS also created another model, the Burn Severity Toolkit. Taking images before and after fires that consumed the area around Boulder, Colorado, earlier this year, staff calculated a normalized burn ratio of the damage. This was accomplished by comparing the two images to see how severely burned the area was.
The U. S. Geological Survey and the National Park Service developed a burn severity index based on Landsat Thematic Mapper and Enhanced Thematic Mapper (TM/ETM) bands 4 (near-infrared) and 7 (mid-infrared) that is called the Normalized Burn Ratio (NBR). NBR imagery allows federal land managers and fire ecologists to evaluate and compare burn severity within individual fires and between fires across various ecosystems. The formula for the NBR is very similar to that of the Normalized Difference Vegetation Index (NDVI), a simple numerical indicator that can be used to analyze remotely sensed measurements. In this case, the formula uses band 4 and band 7. The Differenced Natural Background Radiation (DNBR) is computed by subtracting the post-fire NBR from the pre-fire NBR. The higher DNBR values are correlated with more severe burns.
The next step is to integrate the DNBR information with slope information calculated from existing digital elevation models. Areas of high DNBR and high slope are merged into a single result and flagged as having a high erosion potential. All of these processing steps are combined using Esri’s ModelBuilder technology, along with surface information such as soil type, surface cover, impermeability, and many other properties, to derive a full erosion probability model that can be used to help prioritize remediation efforts. ENVI image analysis tools such as the fire fuel load tool and the burn severity toolkit being integrated with ArcGIS are two examples of how imagery brings a layer of information to GIS workflows that helps in better critical decision making. See Figures 2-4.
Importance of Imagery in the International Community
Impacts from disasters in North America such as the Red River flooding and Texas wildfires, as well as others throughout the world, are considerable. Geospatial information can improve our understanding of impacts and risks, leading to improved preparedness, prevention and mitigation of impacts. The operational use of satellite Earth observation and the increase in the sources available have made a crucial difference to how the earth’s resources are managed, where people choose to live, and what steps are taken to protect against the impacts of future disasters. Detailed analysis in GIS systems has made this possible – integrating the dynamic information derived from Earth observation adds the crucial time element.
The international community has recognized the valuable information that can be extracted from Earth observation imagery and places a high degree of importance on the creation of political mechanisms and agreements to facilitate the acquisition, processing and dissemination of imagery. In July 1999, the European and French space agencies (ESA and CNES) initiated the International Charter “Space and Major Disasters” with the Canadian Space Agency (CSA) signing the charter on October 20, 2000.
The goal of the “Disaster Charter” is to provide access to satellite imagery when disasters strike to help with assessment and relief efforts – imagery from optical and SAR satellites can be requested in a rapid manner. Satellite resources are made available through data sharing agreements set up with the space agencies of the respective governments. According to the latest annual report, the charter was activated 39 times during 2011, an average of three times per month, up sharply from previous years. The information derived from Earth observation through the charter activations has redefined people’s understanding of disasters, how to respond and what preventive actions to take in the future.
Integrated use of imagery in geospatial analysis has made a tremendous difference – further integration and streamlining of workflows will lead to even better results. The information imagery brings and the information that can be derived give an added advantage by opening up another dimension to really understand what is going on in a geographic area of interest. To learn more visit www.esri.com/imagery.