The potential of using high resolution spectral data for detecting either geobiological phenomena or features in the urban landscape has grown dramatically in just the last few years.
As the availability of multi- and hyperspectral high resolution space imagery grows, the opportunities for developing more sophisticated statistical models capable of discriminating among objects and life-forms on or near the earth's surface improve.
The WorldView-2 satellite 50cm resolution image of coastal waters illustrates the clarity of output from high resolution satellites. Distinguishing one boat from another, for instance, is straightforward. Less than a decade ago, with only Landsat's moderate resolution (30m) imagery available, scientists were hard pressed to visually discriminate among objects smaller than 100 ft across. With advancements in satellite optics that same task, as shown, is easily achieved today.
Furthermore, as satellites add spectral sensors - going from 4 or 8 bands to 29, as in the case of the WorldView-3 satellite - huge amounts of important new data are generated and the opportunities for developing statistical models to detect and distinguish among life-forms or objects in an urban landscape correspondingly improve.
Two projects are underway by those working in Senecio's Remote Estimates branch. The first venture seeks a solution to the problem of locating abandoned and long forgotten oil and gas wells. This is especially a problem in oil boom areas of 100 years ago when the locations of the majority of wells were not carefully logged. In the second project, preliminary analysis using Landsat-7 output suggests it may be possible to develop statistical models capable of detecting biological toxins in desert soils. Work on this project will begin with ground-truthing conducted in the Sonoran desert over the next few years with simultaneous statistical modeling completed using Landsat-8 sensor data.
Current & Pending Projects
Pinpointing Abandon Oil Wells
There are an estimated 36,000 abandon oil wells in Wood County, Ohio, USA. The number for all 88 Ohio counties is well over one hundred thousand. Many of these wells date back to the late 1800‘s - the period of the great oil boom. According to state records, the location of approximately 20 percent of these wells are known, meaning the whereabouts of roughly 30,000 oil wells in Wood County are unknown. Determining the locations of these abandoned wells has become a high priority for County Health and EPA officials since abandoned and often unplugged or unsealed oil and gas wells pose potentially serious health and environmental risks.
Other than the haphazard and unpredictable discovery of an abandon well by a homeowner, farmer or developer, there is currently no cost-effective, systematic and widely applicable method for locating the whereabouts of these “lost” wells. However, with the successful launch of WorldView-3 (1.8m, 29 spectral bands) an experienced group of remote sensing scientists and statisticians have begun working on this problem with the goal of developing a remote estimates model capable of pinpointing the sites of many abandon oil and gas wells across the county and around the state.
Recent work at Senecio using Landsat-7 imagery to study world deserts suggests the possibility to improve detection of certain types of bacteria in desert soil. If confirmed, it argues for the broader use of remote sensing in field’s as diverse as geomicrobiology, regional ecology and atmospheric sciences. Additionally, this work points to an important new application in the field of public health since certain bacteria and their toxins are increasingly implicated as risk factors in a number of autoimmune and neurodegenerative diseases.
Detecting Toxins in Desert Soils
To illustrate the possibility of detecting bacteria in soil, a Landsat-7 image acquired in 2009 of irrigated farm circles near New Mexico's Estancia Basin were processed. These images appear below.
On the bottom left of the four is the natural color image of the Estancia Basin's irrigated farm circles. Circles with vegetation show as green and are easily distinguished from bare soil fields. At top right, a vegetation filter (vegetation shows white) has been applied to the satellite's image. The amount of vegetation is reflected in white's intensity. The bottom right image is separately filtered for water. If present, water shows as white. At top left, the Landsat-7 image uses a phycocyanin filter developed for aquatic-environments. Levels of phycocyanin are colored with the lowest levels of the pigment shown in light blue to yellow. Red indicates the highest level of phycocyanin. Phycocyanin is the pigment in cyanobacteria. Since certain strains of cyanobacteria are known to produce harmful toxins, the ability to remotely detect the presence of this bacteria in soil is important, especially as it applies to health issues for those working in desert environments where they may be subjected to frequent inhalation of desert soils due to work activity and major dust storms.
Although the statistical filter for phycocyanin was not intended for use on images of desert environments, and is at best indirect evidence of the presence of cyanobacteria since no ground-truthing work could corroborate these remotely sensed findings, it offers the promise of developing powerful new statistical models capable of detecting certain strains of terrestrial bacteria. Such models could be important new tools in the field of applied health and possibly serve as part of an environmental early warning program where large urban populations live in the path of frequent massive dust storms laden with cyanotoxins.