Hazardous Waste Program
ARCHER collected imagery of 14 sites with a variety of contamination conditions during the two-day deployment. All of the sites were reviewed by the department's site Project Managers (PM) and University Missouri Columbia (UMC) analysts, and signatures developed, based on PMs' knowledge of site conditions, where feasible. No signature data was obtained through Phase I research. Because of limited funding and time, only those sites with the highest potential for HSI application were examined in detail. Initial processing of the ARCHER output for use in commercial imagery analysis software (ENVI and ERDAS IMAGINE) was more time consuming than anticipated and reduced analytical time. Several site conditions showed immediate potential. However, the initial results need ground truth confirmation before further use. The following are two examples of the analytical work. More detail on these sites and the remaining analysis is provided in the final report.
Potosi Ground Truth
Analysis of the ARCHER data continued throughout the quarter, with one field trip to Potosi to collect ground truth data. ARCHER normally flies at 2,500 ft above ground for one-meter resolution imagery. One pass at a lower altitude was planned for the project to evaluate higher resolution imagery. Sixty-cm was obtained.
Figure 1 - Potosi Low Altitude Collection/Concentration Level Analysis
The objective was to determine if the higher resolution would help distinguish between low and high concentrations of contaminants, e.g., between 400 and 4,000 parts per million (ppm) lead. Figure 1, left side, is a clip of the single low altitude pass along hwy 185 in Potosi. The right side, is a clip of detailed analysis of the Potosi fairgrounds. Instead of the single yellow color used in other analyses, highlighting the closest correlated signatures, a range of colors was used to show the gradual widening of the angle of correlation between exact and no correlation. Red shows areas that are the closest match to the signature, developed from the source identified by the PM, with yellow, green, cyan and purple showing areas that were progressively further from a matching signature.
Additional graphics of other areas in Potosi and the concentration range analysis are provided in the Final Report.
On Nov. 17, Shawn Muenks and Nick Carbone, Federal Facilities Section, took the imagery of Potosi, both low and high altitude images, and a Field Portable Xray Fluoroscope, or FPXRF, to collect readings of surface metals in the source area and the areas analyzed by UMC. UMC also provided coordinates for red, yellow, cyan and purple areas for ground truth. Unfortunately, some of the areas with coordinates provided by UMC were not accessible or had changed due to construction since the August flight. Figure 2 is the fairgrounds at the intersection of highways 185 and 8. The left clip is the imagery analysis, and the right clip is the results of FPXRF sampling. The main fairground area, where UMC had provided coordinates, was not accessible.
Figure 2 - Potosi Low Altitude Collection/Ground Truth Data Collection
FPXRF readings were then taken from either side of the fairgrounds, the fairgrounds parking and an Ameren UE equipment storage area, using land features and the color-coding. The right clip shows the readings for iron (Fe) and lead (Pb) in ppm. Most of the areas that should have been high in lead were not above 1,000 ppm as expected although the paved entrance to the fairgrounds parking was 1,400 and 1,800 ppm, probably from the use of tailings for aggregate in the asphalt and roadbed. However, we did record high iron readings, in access of 10,000 ppm. An iron level of 23,000 ppm or above is of concern. Subsequent collection at the source also provided readings that were low in lead and high in iron. This may indicate that, although the source area was a mining waste pile, the surface of the source area was not reflective of the buried waste. The results further emphasized the need to collect spectrometer readings of the contaminated areas of interest concurrently with the airborne collections, so that the results are reflective of the contamination of interest. Additional graphics and analysis of the Potosi ground truth work are provided in the Final Report.
Figure 3 - Praxair Endmember Analysis
Praxair Fire Analysis
Several other sites were analyzed using different methods of imagery processing, which are presented in the Final Report. An endmember analysis was used on the Praxair tank farm fire in St. Louis. The Praxair tank farm fire provided an opportunity for ad hoc collection against an Environmental Emergency Response (EER) related site. Many of the compressed gas tanks that exploded contained asbestos. EPA's ASPECT aircraft tracked the plume of chemicals from the fire, and the aircrew provided EER personnel a plot of the area covered by the plume. The ARCHER HSI was collected 45 days later with the objective of assessing the sensor's ability to determine the extent of asbestos dispersal. By the time ARCHER flew, much of the asbestos had already been cleaned up in the area plotted by ASPECT. UMC conducted a preliminary analysis of the affected area using an endmember classification routine. Figure 3 shows the results of the endmember analysis. The classification identified areas within the known contaminated area that matched the predominant signature from the tank farm. It also identified areas outside the search area. John Whitaker, EER, reviewed the Praxair imagery on September 29. Due to the progress of the cleanup, no sampling or further ground truth efforts were undertaken. However, the process does indicate a potential application of HSI for this type EER response, particularly if conducted in conjunction with EPA's ASPECT real-time support.
Additional analysis of Pilot Project HSI is provided in the Final Report.