High-resolution satellite imagery and computer image processing techniques recently developed and tested by the California Integrated Waste Management Board (CIWMB) and the NASA Ames DEVELOP Research Center (NASA) may be the latest in leading edge technology designed to locate and identify illegal waste tire piles scattered throughout California.

By Rosalie Mulé

Initiated as a pilot project in 2005, this technology has provided the CIWMB with up-to-date information on the existence of previously known and unknown tire piles in select areas of California. Generally these sites are located in remote locations such as farmlands, hillside gullies, and desert regions that are not visible or accessible.

Waste Tires in California
Approximately eight million waste tires are illegally landfilled or stockpiled in California annually (California Integrated Waste Management Board. 2003. California’s Waste Tire Challenge. Publication #400-94-012. Sacramento: State of California). These stockpiles, left unchecked, pose a significant threat to the environment and to residents in neighboring cities and towns that are affected by fire, smoke inhalation, or road closures. Illegal tire sites are also known to harbor disease-carrying rodents and serve as a breeding ground for mosquitoes that can spread the deadly West Nile virus (Beavers, G.M. 1996. Utilization of Low Altitude Remote Sensing to Detect Larval Habitats of Tire-Breeding Mosquitoes. Ph.D. dissertation. Lexington, Kentucky: University of Kentucky).

According to the Centers for Disease Control and Prevention there have been 873 reported California cases of West Nile virus (WNV) in humans as of Jan. 10, 2006. The West Nile virus presents a “well-documented threat to public and animal health; with the most serious symptoms resulting in encephalitis (inflammation of the brain) in humans and horses, as well as mortality in certain domestic and wild birds” (West Nile Virus: Fight the Bite, www.cdc.gov/ncidod/dvbid/westnile/, Centers for Disease Control and Prevention, Division of Vector-Borne Infectious Diseases, Atlanta, Ga., Feb. 14, 2006).

California’s pioneering efforts to develop a multifaceted strategy to keep the state’s increasing number of waste tires from accumulating in unregulated piles was the impetus for the collaboration between the CIWMB the NASA Ames DEVELOP Research Center and San Jose State University Foundation. Utilizing the latest in high-resolution satellite imagery and computer image processing techniques that are currently available, this team of scientists and researchers developed a pilot project that will not only benefit the CIWMB, but other local, state, and federal agencies as well. This superior surveillance technique will ultimately improve existing efforts when targeting larger regional areas throughout the state. The CIWMB also contracts with the California Highway Patrol in a complimentary effort to locate and target waste tire piles in smaller geographic areas. Both surveillance strategies will offer a comprehensive and up-to-date technology for tracking the accumulation of illegal waste tires, and other illegally disposed materials throughout the state.

In California’s Five-Year Plan for the Waste Tire Recycling Management (2003), the CIWMB allocated $350,000 to support the CHP in providing not only aerial surveillance operations, but also ground based enforcement, roadside check points, training, and educational support. These efforts have been formidable in location and remediation strategies that have targeted California’s illegal tire piles.

While the cost of the Satellite Imagery Pilot Study, which targeted four California regions totaled a mere $30,000 over a seven-month period, it is anticipated that the continued use of this high-resolution satellite imagery and computer image processing technique, along with improvements in data reporting, will enhance the state’s overall surveillance activities for the CIWMB in upcoming years.

The use of this satellite technology will provide a new tool in the CIWMB tire-prevention toolbox. It will also reduce staff time, protect personnel from hazardous field conditions, conserve resources, and provide local governments with more accurate information on any illegal disposal sites. Finally, it will provide the CIWMB with a legally defensible strategy that will validate the need for private property site access in order to inspect and remove any illegal tire piles.

Currently, satellite and aerial imagery is being used by many private and governmental organizations in and outside of California to streamline operations and provide improved data-gathering techniques for analytical and planning purposes. Select government agencies that currently use satellite data include the Office of Emergency Services, CalTrans, California Department of Forestry, Resources Agency, and the State Water Resources Control Board (NASA Ames/DEVELOP Research Center, 2005. Detecting Waste Tire Piles Using High-Resolution Satellite Imagery and an Image Processing Model in Two Regions of California. Report for CIWMB, California. Publication Number Pending).

Previous Research
Prior to the CIWMB study, there had been only two published reports on the detection and mapping of waste tire piles in the United States that utilized image interpretation. What is unique about the CIWMB approach is the utilization of computer technology to quickly model the images, thereby assisting in the location of waste tires.

The first of these was the “Utilization of Low Altitude Remote Sensing Imagery to Detect Larval Habitats of Tire-breeding Mosquitoes” (Beavers, 1996). The second is a 2002 report published by the USEPA titled, Aerial Photographic Inventory of Tire Piles: United States-Mexico Border. (USEPA, Environmental Sciences Division. Las Vegas, Nevada. TS-PIC-20209947S).

The Beavers study used a computer-assisted classification protocol, whereas the EPA study used manual image interpretation techniques for the location of tire piles. Beavers identified tire piles in Kentucky, Indiana and Ohio using low-altitude, 0.5-meter-resolution aerial videography. Data was acquired using spectral signatures that were measured with a spectroradiometer—designed to identify a variety of tire brands and types. In this study, the researcher found no significant difference in signatures between tires; whether they were in different piles, of different brands, or at different heights. However, what was determined was a spectral variability between clean tires and those covered in soil; leaving soil to become the dominant spectral signature in the analysis.

The US Environmental Protection Agency visually analyzed over 1,000 color-infrared and medium-altitude aerial images provided by the USGS and the Mexico Instituto Nacional de Estadistica Geografia e Informatica (INEGI) archived collections. The aerial images covered 824 kilometers, or 518 miles, of terrain along the border between the United States and Mexico. Using manual classification techniques, researchers identified approximately 32 sites located primarily along transportation corridors and urban areas. Tire piles were based on their signature identifiers and included dark tones, coarse-textured surfaces, shape, pattern, and height distinctions in relation to the surrounding terrain. Tire piles were divided into certain, probable, and possible identifications for purposes of this study.

Study Area
The CIWMB study targeted four sites in two climatic regions, including northern California’s coastal region and southern California’s desert region. Although tire piles were known to exist in each of the targeted regions and were thought to total over 62 square-miles, this information was not made available to the NASA research team for the duration of the study.

In the coastal region, tires were known for their use in gullies as an erosion control mechanism. In the desert regions, tires have been used as windbreaks on agricultural land and as fences on abandoned property. In both regions, it was believed that a fee was charged to illegally store the tires on private property. Identifying waste tire piles using satellite imagery is challenging, because tire piles absorb light and have low reflectance characteristics. This contributes to the confusion between tire piles and shadows or deep bodies of water.

Methodology
In order to secure data from the field, a synced Global Positioning System/Personal Digital Assistant (GPS/PDA), a handheld device used to interpret data in the field, was programmed with EcoNab, a field data classification and organization program. The data was then uploaded into a Geographic Information System (GIS) software package and used as reference data to identify waste tire locations. IKONOS imagery from space was used due to its relatively low cost and 4-meter spatial resolution in the multi-spectral bands. As with the EPA study, tires, water, and adjacent shadows were found to have low reflectance values and were identified by spectral similarities. In order to maintain the radiometric integrity of the images, they were geo-referenced to USGS Digital Raster Graphics (DRG). Each geo-referenced image was then processed separately to avoid redundancy of image overlap, after which the images were combined to create the final map used for analysis.

To create an automated model that would isolate the tires in both regions, the Leica Geosystems’ ERDAS Imagine Model Builder was used. Next, the geospatial data was translated to the Tire Identification from Reflectance (TIRe) model using the geo-referenced image. Determining how to spectrally identify tires within the imagery was essential during the analysis. “Thresholding,” a common image-processing technique for separating features, could not be used, because the digital number (DN) value of tires varied between images. Isolating the tires from other features targeted a variety of band indices; and a land-and-water index found on the Leica Geosystems Web site was modified for use with IKONOS imagery: (100*((G/NIR) + 0.0001).

This index consistently isolated tires from the original data and eliminated all significant vegetation and dark water features. In order to further separate tires from other features, hues targeting blue, green, and red bands were used. Hue separation successfully eliminated non-tire pixels, including water and dark soil, from the images. The two functions were then combined and used to subset the original image, thus eliminating between 83% and 99% of each image. As a result, the model removed all features except tires and select dark objects, such as dark agricultural lands. The first resulting output was a binary mask used for visual interpretation.

Once the images were processed through the TIRe model, less than 1% of the original image remained—consisting of only dark pixels containing tires or spectrally similar objects. The second output, a subset of the original image, required further classification to assist in the visual interpretation. In images that retained non-tire pixels it was determined that they were dark or burned soil. The first five classes of a 20-class ISODATA input were found to contain tires. This was verified by using known waste tire sites for calibration purposes. Once done, the binary mask output of the TIRe model was placed over the original image for analysis and final interpretation.

Ultimately, the TIRe model and subsequent visual analysis identified 13 waste tire sites in the four study areas. Although CIWMB knew the locations of these sites prior to the study, the NASA researchers did not. Of added benefit to CIWMB researchers was the identification of two new sites found by the NASA team that were previously unknown to the CIWMB. One previously unknown tire site was located in Sonoma West, while the second site was found in the Coachella Valley. Other research results determined that the TIRe model produced false-positives at each study site. False positives were commonly attributed to shadows, water, debris piles, tirelike tarps, polyethylene tubing, and parking lots. With continued communication between researchers and field inspectors, it is estimated that these false positives will be dramatically reduced over time. CIWMB withheld information on all of the waste tire sites in each study area in order to test the accuracy of the modeling outputs, except in the case of the calibration sites. In all cases, it was found, the TIRe model identified the undisclosed tire sites, further validating this state-of-the-art technology.

One of the challenges researchers faced in the development of the TIRe model was the image disparity taken between winter and summer months. In winter months, lower sun angles produced long shadows, decreasing the ability to effectively identify tires in hilly regions. During final analysis of the project, it was determined that the TIRe model would perform best when consistent images obtained during the summer months were utilized. The recommendation for consistent seasonal imagery combined with radiometric requirements produced the most accurate results from the TIRe model. It was also determined that large bodies of water had a tendency to skew the radiometric balance of an image and impede results. Standard deviations of bands and indices in the model that were used to separate dark features from one another reduced the coverage of large bodies of water in the imagery. This allowed the remaining dark features to accentuate, leading the researchers to believe that if an image was greater than 25% water, the area should be digitally masked during the preprocessing phase. The only exception cited was if the coastline were under review, in which case it was recommended that the image be reduced to greater than 75% non-water. Furthermore, in order to uncover smaller tire sites, researchers recommended that imagery with finer spatial and spectral resolution be utilized to improve interpretative efficiency.

Conclusion
The development of high-resolution satellite imagery and computer image processing techniques for the identification of known and unknown illegal tire piles in select regions of California proved successful at the close of this seven-month study. Recognition goes to the CIWMB for concept development and funding; to the NASA Ames DEVELOP Research Center, which conducted the research and analysis, and, finally, to San Jose State University Foundation, for its participation and collaboration on the final report.

The development of this technology has resulted in a significant tool in the prevention of tire fires and the remediation of hazardous health conditions resulting from illegal tire piles. One of the main advantages this technology provides is the high degree of reproducible accuracy that can be achieved when analyzing spectral data.

The CIWMB now has the latest state-of-the-art surveillance tool that provides a number of benefits when tracking and eliminating illegal tire piles, including improved staff response time, identification of the best and shortest access routes, centralized monitoring, the protection of staff from hazardous field conditions, and the conservation of natural resources.

Since the satellite imagery required to facilitate this project is commercially available, it is anticipated that further use of this technology will become valuable in locating and mapping oil spills, landfills, or other environmental nuisances. By spearheading this technological use, the CIWMB is preparing to make this technology available for use by other agencies. With new projects on the horizon, the CIWMB is pursuing efforts to provide this technology to other organizations by releasing it into the public domain, for the benefit of all agencies, both inside and outside California.

Currently, the CIWMB is planning to use this technology for the development of another high-profile waste tire study, to take place along the California-Mexico Border Region. Ultimately this will prove beneficial to California and our neighboring nation, Mexico.

Rosalie Mulé is a member of the California Integrated Waste Management Board.

For More Information
For further information regarding the report entitled Detecting Waste Tire Piles Using High-Resolution Satellite Imagery and an Image Processing Model in Two Regions of California, please contact Darryl L. Petker at the California Integrated Waste Management Board at his e-mail address: dpetker@ciwmb.ca.gov or by phone (916) 341-6704.

The California Integrated Waste Management Board is the State’s leading authority on recycling and waste reduction. It promotes a zero-waste California in partnership with local government, industry, and the public. This includes reducing waste whenever possible, promoting the management of all materials to the highest and best use, and protecting public health and safety and the environment. For more information, visit www.ciwmb.ca.gov.

MSW - September/October 2006