Analysis of Pre- and Post-remediation of Jasper County, Missouri, part of the Tri-State Mining District

Elizabeth Hagenmaier

Spring 2017 ES 775 Advanced Image Processing
Dr. James S. Aber, Instructor

Table of Contents
Abstract
Project Area
Methodology
Results
Conclusions
References

Return to top of page.


Abstract

The Tri-State Mining District covers 2,500 square miles in Missouri, Kansas, and Oklahoma. Lead and zinc mining was conducted in the area from the 1850s to 1970. As a result of the mining, the soil, sediment, surface water, and groundwater was highly impacted with heavy metals. Large piles of waste material from the mining operations scatter the landscape throughout the Tri-State Mining District. This project examines land cover in Jasper County, Missouri, focusing around Webb City. Jasper County is included in the Oronogo-Duenweg Mining Belt Superfund site (Site) by the U.S. Environmental Protection Agency (USEPA). The Site covers about 20 square miles near Joplin, Missouri, and over 10 million tons of surface mining wastes contaminated about 7,000 acres of the Site. USEPA began the cleanup of the mine waste areas in 2008. As a component of the remediation, USEPA excavates the mine waste and contaminated soils, regrades the area, and revegetates the disturbed area. The goal of this project was to identify change in land cover pre- and post-remediation through processing of Landsat TM data with TerrSet geographic information system (GIS) software. Landsat TM data was retrieved from GloVis Next at no cost. Analysis of vegetation and land cover through remote sensing is a common component of most land use studies. As expected, this project proved successful in classifying areas of exposed mine waste versus successful revegetation.

Return to top of page.

Google Earth imagery of pre-remediation (left image) and post-remediation (right image) of project area showing Joplin and Webb City, Missouri and Galena, Kansas. Maps created by author Hagenmaier on May 4, 2017 from Google Earth available imagery.


Project Area

Site Location and Background

The Tri-State Mining District covers 2,500 square miles in Jasper and Newton Counties, Missouri, Cherokee County, Kansas, and Ottawa County, Oklahoma. Lead and zinc mining was conducted in the area from the 1850s to 1970. As a result of the mining, the soil, sediment, surface water, and groundwater was highly impacted with heavy metals. Large piles of waste material from the mining operations scatter the landscape throughout the Tri-State Mining District. The Missouri portion of the district covers the southwest corner of Jasper County, Missouri and scattered throughout Newton County, Missouri. Jasper County is included in the Oronogo-Duenweg Mining Belt Superfund site (Site) by the U.S. Environmental Protection Agency (USEPA). Mining and smelting activities began around 1830 and continued through the 1970s. Ore production in Jasper County included mining, milling and smelting operations. At one time, there was approximately 200 mines found in and around the Oronogo and Duenweg areas. Approximately 60,000 people live within the Site boundaries in Jasper County. The city of Joplin houses a majority of the population along with the surrounding communities of Webb City, Carterville, and Duenweg. Several other small communities are scattered throughout the Site.

EPA Site History

The Site was listed on the National Priorities List (NPL) in 1990. The Site covers about 20 square miles near Joplin, Missouri, and over 10 million tons of surface mining wastes contaminated about 7,000 acres of the Site. For administrative purposes, the Site was broken into five operable units (OUs). OU 01 covers the mine and mill waste. OU 02 and OU 03 address the lead contamination in residential yards. Groundwater and drinking water was covered in OU 04 while surface water and sediments are under OU 05. USEPA began the cleanup of the mine waste areas under OU 01 in late 2007. As a component of the remediation, USEPA excavates the mine waste and contaminated soils, regrades the area, and revegetates the disturbed area. As of 2012, EPA has completed the cleanup of approximately 1,500 acres of mine wastes. Approximately, four million cubic yards of mine wastes and contaminated soils have been excavated and disposed of at the Site (USEPA, 2012).

Return to top of page.


Methodology

The original purpose of this process was to attempt to quantify areas that were devoid of vegetation pre-remediation and post-remediation. Areas of exposed mine waste lack active vegetation due to the level of heavy metals and physical material not allowing plant growth (USEPA, 2012). In the pre-remediation images, the areas of ‘bare’ soil would include the mine waste areas and the other non-vegetated areas such as cityscape, roads, water bodies, etc. In the post-remediation images, the areas of ‘bare’ soil should be limited to the other non-vegetated areas and the former mine waste areas should be classified as active vegetation. Using a vegetation index in TerrSet, the images were processed to quantify the increase in active vegetation in the scene as an ultimate result of the EPA remediation and revegetation activities.

The pre-remediation imagery was acquired under the Landsat 5 mission using the Thematic Mapper (TM) sensor in June 1989. And the post-remediation imagery was acquired in August 2016 under the Landsat 8 OLI mission. All imagery was downloaded from the USGS Glovis Next webpage at no cost, see References. Imagery has a 30-meter resolution and the scenes cover a size of 170 kilometers by 185 kilometers. Due to the size of the scene, each image file was resized to a smaller select scene using the WINDOW function in TerrSet and selecting appropriate rows and columns. The total rows in both sets of imagery was 770 and columns at 848. The total ground cover is rows multiplied by columns multiplied by resolution squared. For this imagery, this totals approximately 588 square kilometers. False-color composites were created by the author to allow for visual observation of areas of active vegetation in red color and areas of no vegetation/’bare’ soil as a cyan color.

before
after

False-color composite of 1989 pre-remediation (left image) and 2016 post-remediation (right image) of area in Jasper County, Missouri. The city of Joplin is in the middle of the image in cyan. Maps created by author Hagenmaier on May 5, 2017 from Bands 2, 3, and 4 from Landsat TM imagery.

NDVI ratio

The Normalized Difference Vegetation Index (NDVI) can be utilized to determine the density of green on a patch of land and observe unique wavelengths reflected by plants. NDVI is calculated from the visible and near-infrared light reflected by vegetation. The formula for the NDVI analysis is the near-infrared (NIR) radiation minus visible (VIS) radiation divided by near-infrared radiation plus visible radiation. (NASA, n.d.)

NDVI = (NIR – VIS) / (NIR + VIS)

After calculations, each pixel in the image results in a number between -1 and +1. High density of vegetation results in a value closer to +1 while the low density/no ‘green’ leaves results in values closer to 0. Using TerrSet VEGINDEX module, the NDVI analysis was conducted on Bands 3 (visible radiation) and 4 (near-infrared) for both the pre-remediation and post-remediation imagery. The following images are the result of this calculation.

NDVI images for both the 1989 pre-remediation (left image) and 2016 post-remediation (right image). Maps created by author Hagenmaier on May 5, 2017. Values for the pre- and post-remediation images range from -0.38 to 0.78 and -0.34 to 0.67, respectively.

Following the creation of the NDVI images, it is appropriate to convert the raw NDVI values into a byte-binary scale. (Aber, 2017) Using the SCALAR module, +1 was added to the NDVI values to shift the value range to 0.63 to 1.78 for pre-remediation and 0.66 to 1.67 for post-remediation. Again using SCALAR, the NDVI values were multiplied by 125 to range the values from 78 to 223 for pre-remediation and 82.5 to 208 for post-remediation. Finally, using CONVERT, the real number values were converted to byte-binary format. Based on the review of the final images, water bodies are in a black color with values closer to 0 while the active vegetation is a red color at the highest values over 200. Areas of bare soil are the in yellow color to green color. After evaluation of the pre-remediation and post-remediation NDVI results, there were noticeable differences between the coverage of active vegetation to no vegetation.

Using the ISOCLUSTER analysis, the images were examined to define areas of similar cell attributes. The ISOCLUSTER analysis is an iterative self-organizing unsupervised classification. Based on the review of both analyzed images for pre- and post-remediation, the ISOCLUSTER analysis was determined to be the most desirable for developing more unique groups within the project area. One cluster, Cluster 2 in the pre-remediation imagery, resulted from the ISOCLUSTER analysis that appeared to represent the assumed limits of mine waste areas based on reference to Google Earth imagery (Google Earth, n.d.).

ISOCLUSTER images for both the 1989 pre-remediation (left image) and 2016 post-remediation (right image). Maps created by author Hagenmaier on May 5, 2017.

Cluster 2 does include areas outside the mine waste areas due to their similar cell attributes to other types of land cover. Another ISOCLUSTER analysis was conducted but allowing for more clusters to be created. This was an attempt to minimize the combination of mine waste areas with other areas.

ISOCLUSTER images for both 12 clusters of the pre-remediation image (left image) and 16 clusters of the pre-remediation image (right image). Maps created by author Hagenmaier on May 5, 2017.

A larger number of clusters appeared to better classify the areas of mine waste. Cluster 3 was selected as the better representation of the mine waste areas for the pre-remediation image. To aid in visualization, a custom color palette was utilized to isolate the selected cluster. For the post-remediation image, it was assumed that the remediation was complete in the areas seen in the 1989 image. However, it is evident from the natural color imagery of the area that the former mine waste areas are remediated but are not fully revegetated. Cluster 14 matches the areas that fit that description. To aid in visualization, a custom color palette was utilized to isolate the selected cluster.

ISOCLUSTER images with custom color palettes for both the 1989 pre-remediation (left image) and 2016 post-remediation (right image). Mine waste in the pre-remediation is indicated by a gray color while un-vegetated former mine waste areas are colored orange with all backgrounds in green. Maps created by author Hagenmaier on May 5, 2017.

The use of Google Earth was paired with the Landsat TM data to operate as another source of easy-to-use natural-color imagery. Historical imagery was also available through Google Earth at no cost. Based on visual review, the selected clusters successfully compared to the Google Earth imagery in both 1989 and 2016. The total area of the clusters representing mine waste areas or un-vegetated mine waste areas was calculated for both the pre-remediation and post-remediation images.

Return to top of page.


Results

The final land cover dataset developed for the project area, with qualitative assessment, represents an estimated reduction of exposed mine waste areas. The total area of mine waste in the 1989 pre-remediation image is 60.6 square kilometers. And the total area of former mine waste left un-vegetated in the 2016 post-remediation image is 13.6 square kilometers. There an assumed 100% reduction in area of exposed mine waste from 1989 to 2016. But a 77.5% reduction in un-vegetated areas was calculated from 1989 to 2016.

Return to top of page.


Conclusions

Overall, the land cover analysis utilizing the Landsat TM data did an adequate characterization of the pre-remediation and post-remediation impacts within the project area. The use of Google Earth allowed for better resolution on smaller mine waste features. It was evident from the natural color imagery that the exposed mine waste had been remediated since 1989. Although the mine waste was removed from the surface, there was still large areas left un-vegetated. This is due to the removal of contaminated topsoil and transition zone soils during the remediation. USEPA is evaluating soil amendments to recreate the soil horizons and allow for adequate revegetation. (USEPA, 2012)

Return to top of page.


References

Aber, J.S. 2017. ES 775 Lab 4 Resampling and Change Detection. Accessed May 5, 2017. http://academic.emporia.edu/aberjame/es775/lab04/lab04.htm

Before-and-after JavaScript plugin.

Google Earth, n.d. Imagery dated December 1989. Accessed May 4, 2017.

NASA, n.d. Earth Observatory, Measuring Vegetation (NDVI & EVI). Accessed May 5, 2017. https://earthobservatory.nasa.gov/Features/MeasuringVegetation/measuring_vegetation_2.php

USEPA, n.d. Superfund Site: Oronogo-Duenweg Mining Belt Joplin, MO. Accessed May 3, 2017. https://cumulis.epa.gov/supercpad/cursites/csitinfo.cfm?id=0701290

USEPA, 2012. Third Five-Year Review Report, Oronogo-Duenweg Mining Belt Site, Jasper County, Missouri. Accessed May 4, 2017. https://semspub.epa.gov/work/07/30246134.pdf

USGS, n.d. Accessed May 3, 2017. http://glovis.usgs.gov/next/

USGS, 2015. Landsat Thematic Mapper TM. Updated May 2015. Accessed May 4, 2017. https://lta.cr.usgs.gov/TM

Return to top of page.

Last updated May 5, 2017