Satellite Data/Aerial Photo/GPS: Integrating Flood Response in the Pacific Northwest
State and federal agencies, along with local govenments, employ RADARSAT aerial videography and GPS to aid in disaster response.
By Charles L. Rosenfeld, Greg G. Gaston and Monte L. Pearson

The recent flooding in the Pacific Northwest was the worst in decades. State and federal agencies, along with local governments, employed a variety of new technologies to assess the damage and manage emergency response operations. Among the technologies being employed within this region for the first time during a major flood event were satellite radar images, aerial videography and GPS tracking. When integrated into a computerized emergency management system, these technologies offered some significant advantages in disaster response.

The Situation
Although winter rains are no stranger to the Pacific Northwest, the nearly 15 inches of rain that fell on the northern Willamette Valley in January was nearly double the average, bringing most agricultural soils to the point of saturation, and dumping an additional 6 feet of snow in the Cascade mountains. When a prolonged series of squalls brought in warm rains from an air mass formed near the Hawaiian islands the residents of Oregon, Washington and Idaho feared that this latest 'Pineapple Express' could rapidly melt the existing snowpack adding to the drenching rains to produce a deluge in a manner similar to the disastrous floods of December 1964. They didn't have to wait long for their fears to be realized. Nearly 8 inches of rain, followed by record high temperatures, swelled the Willamette basin and the lower Columbia to more than 7 feet above flood stage in many areas. The saturated soils on the hillsides could not resist the additional runoff and over 400 landslides were reported, often producing torrents of mud and debris along stream channels. Major connecting highways between the densely populated Willamette Valley, the Oregon coast, the Cascade mountain passes, and even the vital I-5 corridor connecting to Seattle and California were cut off by floodwaters. Nearly 10,000 northwest residents were evacuated to hastily prepared shelters as waters rose.

The Response
As swollen rivers drowned sections of U.S. 101 along the Oregon coast, and local officials received calls for assistance from stranded residents of coast range communities on the evening of February 7, the full dimensions of the flooding was unknown. Over four additional inches of precipitation would fall during the next 24 hours, and the situation was deteriorating rapidly.
      Colonel Charles Rosenfeld reported for duty, directing the National Guard Emergency Operations Center on Oregon's north coast. Normally a geosciences professor at Oregon State University, Dr. Rosenfeld's research activities in remote sensing includes a RADARSAT1 applications project focused on flood hazard reduction in Bangladesh. By the morning of February 8, the similarity between the flooded Ganges Delta and Oregon's Willamette Valley was striking. He placed a call to the Coordination Office at the Canadian Space Agency (St. Hubert, Quebec) to request emergency imagery support from the RADARSAT satellite - then in a pre-operational 'commissioning' phase. Salvatore Carboni and Ken Link of the Applications Development and Research Opportunity Coordination Office rushed to obtain a C-band radar image as soon as the next orbital pass would permit.
     On February 13, at 7:31 local time, RADARSAT-1 acquired a C-band (25 meter resolution) image just following the peak flood crest. To deliver the acquired images quickly, the processed imagery was converted to a TIFF format and placed on the RADARSAT Home Page on the Internet [http//radarsat.space.gc.ca]. Figure 4 shows a portion of the RADARSAT image collected over the lower Columbia River on February 13. The flooded areas are portrayed as smooth black features which cover many areas of the floodplain and the lower reaches of tributary rivers such as the Willamette. The 25 meter resolution of the imagery was sufficient to provide a quick assessment of the extent of flooding, status of flood protection dikes, and location of ships in the navigation channel. Figure 5 shows the same area imaged by an X-band Side-Looking Airborne Radar (SLAR) system2 flown on an OV-1 (Mohawk) aircraft by the Oregon Army National Guard in August of 1977, during a non-flood period. Although the older SLAR technology displays the scene in a 'slant range' format (compressing the image closest to the aircraft, left side of Figure 5) as opposed to the map-like ground range display of RADARSAT, the standing water is clearly displayed as smooth, black areas on both images. The contrasting images permitted mapping of the flood-affected areas, and allowed the National Guard Emergency Operations personnel to evaluate the probability of road flooding and access to affected areas. Likewise, buildings and bridges are depicted as 'bright' returns on both systems, and structures are clearly shown, especially in flooded areas.
      In theory, radar systems illuminate the terrain with microwave energy directed at an angle across the image area. The microwaves are either absorbed by the surface, backscattered by objects, or reflected away from the receiving antenna. In terms of interpreting radar imagery of flooding conditions, the 'brightest' image returns are produced by dense, angular objects that are either located on dry land or at least 'protrude' from the flood waters. Flood waters affect the complex dielectric constant, a property of materials that allows microwave penetration when moisture levels are low, resulting in absorption and darker appearing imagery. As moisture content is increased, there is a nearly linear corresponding increase in the dielectric constant, and subsequently brighter returns. However, in the case of open flood water almost all of the microwave energy is reflected away from the radar antenna, and the resulting radar signature is very dark.
      Figures 4 and 5 illustrate these imagery characteristics very well. The dark flooded areas shown on the RADARSAT image (Figure 4) show a marked increase in open water areas, when contrasted to the dry season SLAR image, in Figure 5. Several distinctive differences in the appearance, or radar signature, of the terrain are apparent between these images. The orbital characteristics of the RADARSAT satellite illuminates the scene with microwave energy of 20 to 27 degrees from the vertical, insuring that much of the energy backscattered from surface targets will be acquired by the antenna. Thus, subtle surface characteristics such as soil moisture and riparian vegetation are more apparent and depicted by a variety of gray tones. The SLAR image is illuminated from an aircraft flying about 10,000 feet above the terrain, and the microwaves graze the surface at very low angles (75 to 85 degrees from the vertical). This reduces the differences caused by moisture content of the surface, but greatly amplifies the backscatter caused by angular objects such as buildings - thus the overall appearance of the terrain is more homogeneous, with built-up areas and ships in the channel appearing as very bright returns.
      Another difference between the two radar systems is that the aircraft mounted SLAR system may be used to acquire an image 'on demand' within its local area of operations, while the RADARSAT image can only be acquired once the orbital path aligns with the target area. Neither clouds, nor darkness inhibit active microwave radar imaging in either case, however the imagery acquired from aircraft platforms is easily degraded due to atmospheric turbulence and is limited by the local availability of SLAR-equipped aircraft. The image characteristics compared in this study clearly favor the higher illumination angles of the satellite sensor, which provide a greater range of information on surface conditions related to flood events.
      Once fully commissioned, RADARSAT assures a day/night and cloud-penetrating acquisition capability that will be used to monitor major floods in remote areas such as Bangladesh, and the imagery and 'ground-truth' acquired during the northwest floods will add insight to future interpretations. The Columbia River image was acquired within four days of the initial request, and the orbital parameters of RADARSAT-1 permit site sampling at 24 day repeat cycles, with seven day and three day sub-cycles. This should meet both reconnaissance and monitoring requirements for many natural hazard events.

Follow-on Verification
Following the torrential rains, the skies cleared on February 10 and 11. Numerous aerial photo missions, as well as the media and curious aviators, took to flight. Among these aircraft was a Cessna 182 operated by the Oregon State University Geosciences Department, equipped with aerial video mapping cameras, and a computerized Global Positioning System to map the flight track. The objective of the videography missions were to collect flood coverage along the banks of the Willamette and lower Columbia Rivers in a high resolution video format, easily converted to digital images. The flight track information is gathered and displayed using FliteMap software from Mentor Plus, which also provides a text archive of heading, position, altitude and time of the flight.
      The OSU videography system is installed in any Cessna 172 or 182 aircraft using a fiberglass nacelle which replaces the standard baggage door. This permits the installation of the entire mapping system into any similar aircraft in less than 30 minutes, without any airframe modification. The video cameras are controlled from the cockpit where the vertical image is displayed on an LCD color monitor. The GPS receiver, a Garmin GPS-55 AVN, sends position updates to an on-board computer2 shown in Figure 6, supporting the FliteMap software which displays the position and track of the aircraft on an aeronautical sectional map at 5 second intervals.
      The map shown in Figure 1 is a cockpit display from the Mentor Plus FliteMap computer of one of the videography flights over the lower Columbia River. The videography also offered the opportunity to compare large scale images to the RADARSAT coverage, for extensive ground truth verification. These video images are also being used by the U. S. Army Corps of Engineers to delineate the flooded areas and compare them to normal bankfull conditions of the channels, and the location of flood protection dikes and structures for maintenance, dredge materials disposal, and planning for future flood control strategies. The computerized flight track may be used to re-fly the video coverage after the flood waters have dissipated, accurately locating damage, facilitating assessment, and even verification for flood loss claims.
      As flood waters subsided, and clean-up efforts began, the RADARSAT images provided sufficient detail to assess the extent of flood damage to cropland and floodplain areas. Aerial videography was used to monitor the immediate post-crest conditions of many of the inundation areas displayed on the RADARSAT imagery.
      FieldNotes software from Penmetrics was used to produce this data integration between an image and Intergraph drawing files from the Portland District Corps of Engineers. In addition to providing a visual record of inundation, the FieldNotes software represents a dynamic link to relating images and maps to databases such as Emergency Operations, Disaster Assessment, and Recovery Operations.
      FieldNotes displays positioning on a portable computer containing images and/or maps of the area, when attached to a GPS receiver. The FieldNotes software also links to databases for collecting information by disaster response teams in the field. This capability was evaluated for use by Oregon National Guard emergency operations command posts to track emergency response missions through the receipt of the request, the authorization process, site reconnaissance, mission planning, operations and completion phases of disaster response operations, providing an interactive display of the situation map, the FieldNotes software is used as a management tool recording the personnel and equipment assigned, field communications, supplies required, status of each mission. These functions are dynamically linked to a database which provides a summary of activities, costs, and results at the conclusion of the operation. This data can accelerate cost recovery and claims processing, and greatly reduces subsequent administrative costs associated with traditional 'after action' reporting.

Conclusions
In future operations, RADARSAT imagery acquired during a flood event could be merged with existing map and image data, down loaded to a field-portable computer, linked to a GPS receiver, and fielded to assist in a variety of emergency service operations from rescue and recovery to damage assessment. The ease and speed with which this demonstration project was accomplished, allows us to confidently envision many practical applications with a minimum of additional development.

1 The AN/APS-94D Side Looking Airborne Radar system, manufactured by Motorola, was carried aboard a Grumman OV-1 (Mohawk) twin turbine surveillance aircraft.
2 Computer used is a Compaq 'Concerto' 486-based pen computer running Microsoft © 'Windows for Pen Computing.'

About the Authors:
Charles L. Rosenfeld is a professor in the Geosciences Department at Oregon State University. He may be reached at 541-737-1208. Greg G. Gaston, Ph.D. is a national research council resident research fellow at the U.S. EPA Laboratory in Corvallis, Ore. He may be reached at 541-754-4496. Monte L. Pearson, Ph.D. is an adjunct professor in the Geosciences Department at Oregon State University. He may be reached at 541-737-2722.

Back