Responding to natural disasters with satellite imagery
Formosat-2 is an Earth imaging satellite launched on May 20th 2004 and operated by Taiwan's National Space Organization (NSPO). It is the only high-resolution satellite with the ability to acquire repeated imagery of an area of interest every day (daily revisit). Its images can be used for land distribution, natural resources research, forestry, environmental protection, disaster prevention, rescue work, and other applications. Most recently, Formosat-2 and its automatic image processing system (F2 AIPS, implemented in late 2005) proved to be useful in responding to Japan's earthquake and tsunami event.
On March 11th 2011 at 2:46 p.m. local time, a 9.0-magnitude earthquake struck Japan, and the resulting 10-meter-high tsunami swept the country's northeast coast within 15 minutes. The United Nations Platform for Space-based Information for Disaster Management and Emergency Response started the International Charter on Space and Major Disasters to coordinate satellites of different countries in acquiring images of disaster affected areas. We collaborated with Taiwan's National Space Organization to rapidly respond to Japan's event with Formosat-2. Once the first image of the affected area after the disaster was acquired and downloaded at noon on March 12th, it was processed within 2 hours and published on the Global Environment Monitoring and Disaster Assessment System (GEMDAS).1
GEMDAS, powered by the cloud server at the National Cheng Kung University in Taiwan, provided a channel for people to know more about what had happened in Sendai after the disaster. From that day on, images of the Chiba refinery fire (see Figure 1), Sendai Airport (see Figure 2), Iwate of Miyako (see Figure 3), and the Fukushima I nuclear power plant (see Figure 4), among others, were updated frequently on the 13th, 14th and 17th of March.
The role that GEMDAS played in the Japanese disaster can be attributed to three factors. The daily-revisit orbit and high agility of Formosat-2, the automatic image-processing system of the satellite (F2 AIPS), and the web-based geospatial information system used to make the images public all contributed to the effective disaster response.
The Formosat-2 satellite, being equipped with a CCD array with 12,000 pixels, is a high-spatial-resolution sensor. In addition, the mission proves the concept that the temporal resolution of a remote-sensing system can be improved by deploying such a system in a daily revisit orbit. Each recorded scene can be systematically observed from the same angle under similar sunlight conditions.2 This, together with its fine-pointing capability (±45° across and along track) provided by three sets of two-axes high-torque reaction wheels,3 makes Formosat-2 a satellite appropriate for continuous monitoring and rapid response to natural disasters around the world.
However, the satellite's uniquely adjusted orbit and the impact of sensor alignment also raise new challenges in image processing. These include the band-to-band misregistration problem, which is caused when individual color lines in the CCD arrays scan the ground sequentially with a slight lag time, rather than sequentially.3 We developed F2 AIPS to process a large amount of Formosat-2 imagery on a daily basis. The system uses the panchromatic band (a greyscale image covering the red, green, and blue parts of the electromagnetic spectrum) as the base image. The other four multi-spectral bands are co-registered to the base image, using a technique called fast normalize cross coefficient. The general image co-registration is a time-consuming process, but with F2AIPS we are able to process one standard scene (12×12km) in approximately twenty minutes on an ordinary personal computer equipped with a Pentium IV 2.4-GHz processor. In addition, the system fixes all problems of image processing caused by sensor alignment.2 With the successful implementation of F2 AIPS, we are able to continuously monitor the global environment and rapidly respond to disaster events, such as the 2007 California wildfires,4 the 2008 Sichuan Earthquake in China, and the 2009 Typhoon Morakot,5 the deadliest typhoon to impact Taiwan in recorded history.
Another important factor in the effective response to natural disasters with satellite imagery is the distribution of the geospatial information. Among the various commercial platforms available for this purpose, we chose to use Google Earth, which has the largest number of users. To support the requests from the general public and government officials, we employed the super-overlay technique to convert each Formosat-2 image to a set of tiled images with different levels of details. Depending on the region browsed by the user, only those tiled images that fall within the region of interest need to be loaded and displayed on the web. With this type of processing, a large satellite image can be browsed by a number of Internet users simultaneously.
Rapid response to a global disaster event with satellite data requires timely image acquisition, fast data processing, and broad information distribution. The characteristics of Formosat-2 imply that images of a specific area, anywhere in the world, can be taken within one day, while the imaging processing system we developed, F2 AIPS, enables us to quickly process a large amount of imagery. The broad information distribution is achieved with GEMDAS powered by cloud service. The system, which is now fully operational, is an efficient approach to support the general public and decision makers by exchanging large amounts of remote-sensing imagery through the Internet. In the future, we will continue to work closely with NSPO to use F2 AIPS and GEMDAS to support the rapid response to global disaster events with Formosat-2 imagery.
Cheng-Chien Liu is a professor in the Department of Earth Sciences. He is also the director of the Global Earth Observation and Data Analysis Center, which is in charge of receiving, processing and archiving Formosat-2 imagery on a daily basis.
