FORESTRY
Managing Biodiversity
GeoTechnologies assist with Amazon oil exploration impact study
By Fred H. Groth

Deforestation in the Amazon is a growing environmental concern. The exploration for and production of oil and gas in the rain forest leads to both primary and secondary deforestation impacts. When oil companies clear land for the construction of seismic lines, production platforms, and other infrastructures, it is considered a primary impact. Secondary impacts occur when colonists enter the rain forest via seismic lines, roads, and pipelines to clear the land for subsistence agriculture. One oil company was interested in tracking and determining their impact on the rain forest, and used remote sensing and GIS (geographic information system) technologies to model deforestation and its relationship to their activities in the area.
    Occidental Exploration and Production Company (OEPC) wanted to determine what environmental impacts existed in their oil and gas lease block prior to 1985, when they obtained concession, and what impacts occurred during their tenure. This was a difficult task since an environmental baseline study was not performed prior to obtaining concession. OEPC wanted to identify current and past deforestation in their block in order to establish the extent of their own potential impacts. OEPC also thought this deforestation pattern could determine future deforestation trends as well. By understanding the dynamics of past activities, OEPC could implement preventive measures to minimize future deforestation. Walsh Environmental Scientists and Engineers, Inc. (WALSH), an environmental consulting firm based in Boulder, Colorado, with specialists in remote sensing and GIS, was contracted by OEPC to conduct the deforestation study.

A Sensitive Area
OEPC's oil lease Block 15 is located in eastern Ecuador, situated in the upper Amazon Basin known as the Oriente. Dense, impenetrable tropical rain forest has left this biodiverse area unexplored and unchanged for hundreds of years. The Oriente is culturally diverse as well, inhabited by indigenous peoples separated from modern society until recent years.
    Oil exploration and production is a major influence on the changing landscape and culture in the Oriente. The influx of agriculture and farming has also had a major visible impact. Vast areas of deforestation are the result of colonists moving into the Oriente and establishing residency. With the construction of roads, seismic lines, and pipelines, oil companies have opened up previously inaccessible areas to settlers.

OEPC Involvement
OEPC has leased Block 15 in the Oriente since 1985 for the exploration and production of oil, and has a 20-year contract with the Ecuadorian government. The property covers 200,000 square kilometers of rain forest, and contains three national parks and a protective forest. Major rivers flowing through the area are the Napo and Aguarico. Native communities along these rivers include the Quichua, Shuar, and Secoya peoples. The block is noted for its biodiversity. The eastern half of the block has had little human influence or impact, while significant human activity has occurred in the western portion. Roads, pipelines, large-scale commercial agriculture, and oil production facilities are present in the western portion of Block 15. In addition, the area adjacent to Block 15 on the northwest has undergone heavy deforestation from agricultural activities.
    With so many influences affecting land use, OEPC wanted to identify the changes and determine those attributable to their activities. Together, OEPC's Health Environment and Safety (HES) Manager, Patricio Rivera, and WALSH decided that a historic land use analysis based on LandSat TM data would be the ideal method for studying deforestation in Block 15. Main objectives for this project are as follows:
¥ Identify the location and extent of deforestation prior to OEPC tenure.
¥ Identify the location and extent of deforestation during OEPC tenure.
¥ Identify the location and extent of regrowth.
¥ Identify the location and extent of deforestation prior to OEPC tenure that are still deforested.
¥ Identify deforestation rates.
¥ Identify factors promoting deforestation.
¥ Determine which factors have a greater influence on deforestation, and lead to increased deforestation.
¥ Reduce deforestation and minimize environmental impacts by identifying areas prone to deforestation.

Getting to Work
The area of the project is vast and inaccessible, making a detail ground survey difficult, time-consuming, and expensive. Since a baseline audit prior to OEPC's lease was not conducted, historical data were not available. To overcome the lack of data, OEPC and WALSH used historic satellite imagery to evaluate past and current conditions. LandSat data from the past would provide a point-in-time for delineating the extent of deforestation.
    Two LandSat Thematic Mapper TM data sets were purchased from EOSAT to help identify and map the land use. Satellite images were acquired for the years 1986 and late 1995. Both data sets were imported into TNTmips, a remote sensing and GIS software package. The study area, consisting of Block 15 and a buffer zone around the block, was extracted from all bands. The 1986 image was georeferenced to existing CAD and topography maps provided by OEPC. The 1995 image was georeferenced to the 1986 LandSat image, then resampled to the same projection and cell size. Visual and numeric analyses were performed on all spectral bands. Histograms of brightness values were generated to identify the distribution of deforested and naturally vegetated sections. Areas of deforestation and agriculture were easily discerned from native vegetation in Band 7 (mid-infrared) of both data sets. Cleared areas and natural vegetation have different moisture levels and plant structure, which is visually evident in the frequency.
    Based on the brightness values of mid-infrared Band 7, a training set was compiled for both images. A filter was written in TNTmips Spatial Manipulation Language (SML) to process the images and extract those areas in the training set range. The results were processed with a filter to remove noise and delineate boundaries. Standard statistics were run on the processed images to calculate the area of deforestation in 1986 and 1995. The block boundary was intersected with the data to obtain statistics for the study area and Block 15. An evaluation of the conditions before 1986 and in 1995 was conducted. The first objective of the project was met, i.e., to determine the total area of deforestation prior to 1986 and up to 1995.
    To conduct a historical comparison, results of the two sample years were combined in an attempt to find areas of overlap. Previous results were merged using a raster add process in TNT, resulting in a new raster image. Areas of overlap identified those regions deforested prior to OEPC tenure, and which remained deforested in 1995. This process also delineated areas of regrowth, as well as sections deforested in the last 10 years. Results of this final intersection process were vectorized using a raster-to-vector conversion process in TNTmips. Standard statistics were calculated for each of the 23,000 polygons. Total area of each class of polygon was calculated, both in the block and study areas. It was determined that during the 10-year period, deforestation occurred at a rate of 6% outside the block, and 3.2% inside the block.

Why was Deforestation Occurring?
OEPC wanted to determine the extent of their responsibility for deforestation in lease Block 15. In addition, they wanted to identify long-term impacts associated with seismic activity. Colonization has occurred in close proximity to natural and manmade features acting as migratory pathways for settlers. The six features identified as promoting colonization are rivers, roads, seismic lines, commercial agriculture, wells, and villages.
    These features all act as migratory pathways for colonization resulting in deforestation of the region. Colonists took advantage of these pathways to infiltrate the area. LandSat analysis and a review of photographs identified "fingers" of deforestation perpendicular to linear features such as rivers, roads, and seismic lines. Also seen was a radial pattern of colonization and deforestation around wells, native villages, and areas of commercial agriculture. These fingers were measured to identify the average distance of travel from transportation routes. This calculated area was used to create a buffer around each feature. The buffer areas and land use change analysis were intersected using the TNT MIPs software. Each polygon or area of deforestation was classified based on the buffer areas where they were located or touched. Six hundred combinations of the 6 main features were possible. The final database contained 28 common combinations of features contributing to deforestation. Results indicated that roads and seismic lines greatly influenced new deforestation.
    The next step was developing a model that could help determine future impacts associated with activities in Block 15. Based on historic trends, the model estimated the extent of deforestation that could occur for a single unit (kilometer) of road or seismic activity. A value (Ha/Km) for combination of influences, e.g., a road near a river and a seismic line, could now be calculated.

Minimizing Future Deforestation
OEPC wanted to use this data to help predict deforestation associated with exploration and production. A proposed seismic line was imported into the GIS and intersected with the buffer areas. Each kilometer of seismic line was classified according to its proximity to those features promoting deforestation and colonization, such as roads and seismic lines. Each kilometer of proposed seismic line was multiplied by the historic deforestation rate derived in the study. The result produced a total area of expected change over the next 10 years. The data were used to identify seismic line locations prone to deforestation. In places where high deforestation is likely, mitigation measures are being developed. For example, seismic lines near a road will be monitored with a manned checkpoint. This would reduce the number of colonists moving into the area. OEPC was granted the first permit in Ecuador for a 3D seismic program in the limits of a national park with this predictive use of the model. "We demonstrated to the government the expected impacts caused by the 3D seismic program. With our strong Environmental Management Plan design, the permit was granted," said Patricio Rivero, HES Manager, OEPC.
    To better use the data collected during this study, WALSH delivered all data into an Environmental Management System (EMS). The system allowed OEPC to view, examine, and query data with a customized ArcView interface. Environmental managers can now model different cases by placing proposed infrastructure on the imagery to identify possible problems and alternative scenarios. Patricio Rivera added, "We can plan for the future. More importantly, we can write the history of the project as we go. This is a very important tool, especially for the turnover audit at the end of the lease."
    The use of satellite imagery and remote sensing provided a cost-effective solution to examining deforestation in the Amazon jungle. Archived imagery allowed the client to look at historical environmental characteristics and compare them to existing conditions. GIS tools provided fast classification and calculation of areas, and acted as a medium to examine the environmental data. The change detection analysis identified trends and patterns of deforestation that would have been impossible with traditional ground-based surveys. OEPC now has detailed baseline environmental data that will allow them to properly manage ongoing projects and plan for the future.

About the Author:
Fred H. Groth is the manager of WALSH's GIS and remote sensing group in Boulder, Colorado. For more information on this, and other worldwide projects, visit the group's home page at HYPERLINK http://www.walshenv.com/gisweb/ or contact us at (303) 443-3282.  

Back