Analysis of vegetation growth in relation to abandoned mine proximity in the North Fork of the American Fork Canyon

Randon Chesnut

CEE 6440

Final Project

December 6, 2007

Abstract

Abandoned mines have become a problem on their surrounding landscapes. This is true for the North Fork of the American Fork canyon. Mining stopped around the North Fork in 1953, yet very little cleanup of tailings and waste has occurred. Cleanup projects have commenced through collaborative efforts. GIS and remote sensing analysis might be a useful way to monitor impact of mines on vegetation, riparian buffer analysis, and success of the reclamation projects. Utilizing techniques such as unsupervised classification, buffering, Normalized Difference Vegetation Index (NDVI), and Fractional Vegetation (FV), analysis over time, pre/post reclamation changes can be evaluated. After this analysis, it was determined that abandoned mines prevent any vegetation from growing around them. It was also apparent that reclamation efforts of mines encroaching on riparian buffers have allowed for the growth of vegetation where only mine tailings previously existed.

Table of Contents

Page Number

Abstract……………………………………………………………………………………2

Introduction………………………………………………………………………………..4

Objectives.………………………………………………………………………………...5

Study Area……......……………………………………………………………………….5

Data Sources and Methods….…………………………………………………………….7

Results……………….…………………………………………………………………...13

Conclusions………………………………………………………………………………20

Recommendations………………………………………………………………………..22

References………………………………………………………………………………..23

Introduction

Abandoned mines can have detrimental long-term impacts on their surroundings. The North Fork of American Fork Canyon is no stranger to mining. Mining began in 1870 with the formation of the Mountain Lake mining district (Crosland et al., 1994). Throughout the history of the North Fork, there have been over 100 mines, mostly located on Miller Hill. Although mining in American Fork Canyon stopped in 1953 the remnants from them are still present. Mill foundations, ore chutes, ore carts, office walls, and other structures are still visible in the canyon today. Worse than these structures, are the tailing piles. Almost all of these tailing piles contain high levels of arsenic, cadmium, copper, iron, lead and zinc (Burk, 2004). The impact these tailing piles have had on the American Fork River, and local fauna has been studied. The results from these studies prompted a warning to anglers regarding the high arsenic levels in fish (Fitzgerald, 2004). A cooperative alliance comprised of the U.S. Forest Service, Trout Unlimited, the Snowbird Corporation, and Tiffany and Company, was formed on August 16^th, 2004 (Barringer et al, 2004). This alliance continues to work on restoring water quality and fish species to their natural states. While this extensive work continues on the North Fork, few studies have been done to determine the impact the re-growth of riparian buffers are having on the chemical and biological conditions in the area. The potential effluent sources from these abandoned mines are diverse and riparian buffer re-growth could assist in mitigating toxic chemical inputs of post-reclamation sites (Brady et al, 2003). The purpose of this GIS project is to focus on the vegetation and riparian buffer characteristics surrounding the Pacific, Dutchman, Bog, and other Miller Hill mines. Some of these mines are difficult to reach. GIS/Remote Sensing (GIS/RS) could be a cost effective method for analyzing these areas. Comparisons will look at the change in vegetation surrounding the mines before and after reclamation projects. Successful analysis should help in determining the benefit reclamation projects are having in buffer areas. These data could be utilized in prioritization of mine reclamation in the North Fork of the American Fork Canyon, as well as progress indicators for current and future reclamation projects.

Objectives

There are five objectives to sufficiently perform the desired analysis. Accurate unsupervised classification of study area has to be obtained. If large inaccuracies exist in classification then inappropriate riparian buffer size will cause each subsequent step in analysis to be in error. Classifications will be used to determine appropriate buffer size. Mayer et al, 2005 released a comprehensive study denoting appropriate riparian buffer size based on vegetation classifications. Mines that are encroaching into the riparian buffer need to be identified. While simply stated, without this knowledge appropriate mines for reclamation and monitoring can not be determined. Vegetation changes over time will be monitored to determine effectiveness of reclamation efforts, and re-growth of riparian buffers. This step is a useful and potentially cost effective to monitor success or failure in a project. Finally, is there any potential for this type of analysis to be used at other locations for management and monitoring of reclamation projects? Associated with this step is determining what data or analysis would improve the overall potential for success.

Study Area

The North Fork of American Fork Canyon is quite diverse in geographic, vegetation, and climatic characteristics. American Fork Canyon is located about thirty five miles southeast of Salt Lake City in the northeastern part of Utah Valley (figure 1). Geographically, it is made up of steep, narrow canyons with small

Figure 1: North Fork Subset of Utah County

open areas forming five lakes and reservoirs. The canyon is bounded by several mountains to the north and south that exceed three thousand-four hundred meters. Elevation of the canyon varies from 1,500 meters at the mouth, to 3,500 meters on the boundaries of American Fork Canyon and Snowbird Ski Resort (Crosland et al., 1994). Miller Hill, the location of the majority of mining activity, is partially encompassed by narrow, steep canyon walls. The northern half of Miller Hill opens up into Mineral Basin.

Vegetation in the area varies widely based on elevation and aspect. The lower elevations of the North Fork riparian area consist of cottonwood, box elder, and willows. Higher riparian elevations, consisting of the primary study area, are found to have mostly willows. The North Fork slopes are covered with mixed conifers, aspen, oak, and maple. The southern aspect slopes have characteristically been dominated by junipers and aspen. Above 3,050 meters, common tundra vegetation, such as shrubs, low grasses and herbs can be found (Burk, 2004).

Climactic conditions vary widely due to the topography and elevation of the North Fork. Snowfall usually begins in September and ends in June, with accumulations totaling over twelve meters annually (Burk, 2004). These conditions make for mild summers and cold winters. Annual runoff is typically only from May to June at which time canyon access becomes possible due to decreased snow levels (Burk, 2004).

Data Sources and Methods

Data to be used for vegetation analysis was acquired from the Intermountain Region Digital Image Archive Center (IRDIAC) and from the Utah AGRC. National Agricultural Imagery Program (NAIP) aerial photography from 2004 and 2006, 5 meter DEM, and Landsat Enhanced Thematic Mapper (ETM) Imagery collected was from 1986 to 2006. Landsat ETM data from June 2, 1987, June 24 and June 22, 2006 was chosen to give change over time comparison as well as pre/post phase one reclamation completion comparison. Other reasoning of ETM data choice was to have close dates for similar vegetation growth and water states, and low cloud cover. During the process of verifying projection data it was noticed that two projections were being used. The standard to be used for this project was WGS 84, UTM Zone 12 and all non-conforming images were then re-projected utilizing the re-project tools under the data prep icon in Erdas Imagine. In order to perform accurate vegetation change analysis around mining sites, data rectification needed to be verified. In the example of the June 22, 2006 image, projection was the same, but landmarks were not synchronized upon visual comparison. Four ground control points (GCP) (figure 2) and four check points were utilized for each image requiring rectification. Along with check points, post rectification visual comparison was done to confirm synchronicity in images. To correct for atmospheric, sun geometry, or instrument calibration error, image standardization was going to be

Figure 2: Ground Control Points (GCP) used for image rectification

performed. This would have utilized the the COST model (Chavez, 1996) programmed as an Erdas Imagine spatial model by the USU Remote Sensing and GIS Laboratory. Unfortunately, due to header file availability, this step was not able to be performed.

Unsupervised classification utilizing Imagine isodata method was used for vegetation classification. To determine appropriate classes for the process, signature files were created at 2, 5, 10, 15, 20, 25, and 30 cluster intervals utilizing Landsat ETM image from June 22, 2006. The mean and standard deviation were added to each signature file, and then imported into excel to calculate the average sum of variance (Table1).

Table 1: Average Sum of Variance used to determine clusters for unsupervised classification

Through variance stabilization, appropriate number of classes was set at 12. The isodata algorithm in Imagine was run one more time on all three images selected for analysis with 12 classes specified. Photographs taken in 2005 as well as Uinta National Forest vegetation cover maps were used for best identification of classes. Accuracy of the classification was determined through visual comparison of field knowledge and NAIP comparisons from 2004 and 2006. Aesthetic changes to classification appearance were performed utilizing ArcGIS operations.

Riparian buffer analysis was done with ArcGIS and the Multi-Watershed Delineation (MWD) tool. Data was acquired from the Utah AGRC, Horizon Systems, U.S. Forest Service, and USU’s Western Center for Monitoring and Assessment of Freshwater Ecosystems (WCMAFE). In order to perform the pre-processing steps with the 5m DEM data in the MWD tool, the 12 digit HUC, and intersecting NHDplus stream data were used. The merged and clipped DEM was then used for a base layer in future analysis. The stream file generated during this process didn’t accurately interpret headwaters, so a conditional statement was used to define a stream as 100 input cells instead of the default 500.

Buffer size was determined from the stream data and Mayer et al, 2005 EPA estimates of

appropriate riparian buffer based on vegetation type (table 2). This process was performed utilizing ArcGIS techniques.

Table 2: Mayer et al, 2005 EPA estimates of buffer mean and effectiveness at removing nitrogen. Items with R2 values lower than .2 were not predicted. This data is used to estimate buffer size of approximately 112 meters.

Using an estimated buffer of 112m, the mine sites encroaching upon the buffer were then selected out for further use.

An analysis of vegetation cover around mining areas was performed using NDVI (Normalized Difference Vegetation Index), FV (Fractional Vegetation) and FV temporal change comparisons. Models were constructed using Imagine’s model builder for the three analysis methods allowing comparisons of vegetation change around potential barren vegetation locations and mining areas over the interval of approximately 20 years (Figures 3,4,and 5). All color changes done for classification, NDVI, FV and FV temporal change were done using ArcGIS’s ArcMap classification methods.

Figure 3: Change in Vegetation Model

Figure 4 and 5: NDVI and FV models created in Imagine

Results

Figure 6 contains the results derived from unsupervised classification of 12 clusters with the isodata method in Imagine.

Figure 6: Unsupervised Classification

Figure 7 represents the preprocessed data from stream buffer analysis with figure 8 indicating the riparian buffer and figure 9 showing the mines that encroach into the buffer.

Figure 7: Stream and mine layout for the North Fork of American Fork River

Figure 8: Map of North Fork area with 112m buffer applied to stream areas.

Figure 9: North Fork area with mines selected that encroach inot the riparian buffer.

Figure 10 contains the NDVI output from the custom built model for all both dates.

Figure 10: NDVI output from selected dates

Figure 11 contains the output from the FV model for both dates.

Figure 11: Functional Vegetation (FV) output for selected dates

Figure 12 contains the output from the FV temporal change model for all both dates.

Figure 12: output from FV temporal change detection from both dates

Conclusions

Overall vegetation analysis appears to be useful in monitoring growth around riparian areas. Unsupervised classification determination was first done on the June 2, 1987 image. The classes were then applied to the other image. There are areas of correct and incorrect representation in both data sets. The aspen, conifer/aspen combinations, and snow areas appear to be fairly accurately classified across all both dates according to Uinta National Forest data and photographic comparisons. A general error from the unsupervised approach was no classification for water. While the North Fork River is quite small, Pittsburgh Lake, located at the top of the area of interest, was noticeable in every analysis performed. Unfortunately a class was not created for it during the process. On a more specific level, mining sites were classified relatively well in 1987 and 2006. Greater accuracy in classification can be obtained with larger clusters used in analysis.

Riparian buffer and mining site analysis appears to be successful in relation to objectives. The riparian buffer determination and identification of encroaching mine sites within the buffer worked to satisfaction. Further analysis would be possible with these datasets.

The NDVI Output helps clarify some of the classification errors, and assumptions. Vegetation patterns appear similar in 1987 and 2006. Differences in vegetation growth occur around abandoned mining sites. Dutchman, Pacific mill, Bog, and the 1^st phase of the Pacific mine reclamation projects were all finished and seeded in the fall of 2005. 1987 shows considerable no growth areas around these mines. 2006 data however, shows more vegetation occurring in those reclaimed areas, as well as other reclamation locations on Miller Hill.

This same theory between 1987 and 2006 data regarding mine reclamation sites is also supported by the FV analysis. Percentage of vegetation occurring in reclaimed mining sites may not be 100%, yet there is still fractional growth, as opposed to the 1987 response of no growth.

Further support is gained by the FV temporal change analysis performed. The most significant data is the change between 1987 and 2006. While the green, which represents loss of vegetation from 1987 to 2006, exists in the basin and the valley floor, the red, or gain of vegetation from 1987 to 2006, has substantial presence on Miller Hill and reclaimed sites. The one location of initial concern was the prior location of the Dutchman mine, which appeared to have lost vegetation. In photos that I took in the fall of 2005, this location had become the repository for all tailings from the other reclamation projects. This caused the removal of aspens and conifers, and had just recently been capped and reseeded.

In order to substantiate these GIS/RS data pre and post-reclamation aquatic invertebrate analysis was used to support the results. Ephemeroptera, Plecoptera, and

Table 3: Invertebrate analysis of EPT taxa from below Pacific mine prior to and following reclamation efforts.

Trichoptera (EPT) are known as intolerant taxa when physical and chemical changes are introduced. Table 3 shows the total family taxa from below Pacific mine before and after reclamation efforts. The significant rise in Plecoptera and Trichoptera taxa indicate an increase in biological and thus chemical health of the system.

Recommendations

Improved data would improve the results. Initial analysis and results were successful in determining those mine sites causing problems within the riparian buffer. Accuracy could be increased with the use of better data. Increased resolution of DEM, and RS data would allow for more detailed classification and vegetation analysis to be performed. Comparative chemical data from pre and post-reclamation would be useful to support GIS/RS results. Attempts to acquire this data were unsuccessful in the time permitted for this assignment. Alternative directions for classification could be to determine chemically polluted soils based on RS responses (Mars and Crowley, 2002). Use of these techniques could be used to monitor the remaining reclamation projects around the North Fork of American Fork River.

References

Crosland, R.I., and Thompson, C., 1994, Heritage resource inventory of American Fork area mine closures, Utah County, Utah, final report. U.S. Forest Service, Uinta National Forest, 103 p.

Barringer, Felicity. Unusual Alliance Is Formed to Clean Up Mine Runoff. The New York Times 18 August 2004, Late ed.: A15.

Brady, W.D., M.J. Eick, P.R. Grossl, P.V. Brady. 2003. A site-specific approach for the evaluation of natural attenuation at metals-impacted sites. Soil and Sediment Contamination 12:541-564

Burk, Neil I. Geochemistry of Ground Water - Surface Water Interactions and Metals Loading Rates in the North Fork of the American Fork River, Utah, From an Abandoned Silver/Lead Mine. Ann Arbor: ProQuest, 2004.

Fitzgerald, T., 2004. Streamlined Risk Evaluation. Created Nov. 27 2004.

Fitzgerald, T., 2004. Site Characterization. Created Nov. 19 2004.

Mars, J.C., and J.K. Crowley., 2003. Mapping mine wastes and analyzing areas affected by selenium-rich water runoff in southeast Idaho using AVIRIS imagery and digital elevation data. Remote Sensing of Environment 84:422-436.