Chevron Petroleum Company of Colombia has been actively using remote sensing technology, aerial photographs, CAD mapping technologies and field work to improve geologic understanding and field operations within Colombia. This article presents case histories of recent activity along the flanks of the Eastern Cordillera in an exploration license on the western flank "Sumapaz" and in an active license along the eastern flank "Rio Blanco." The Rio Blanco acreage is on trend with the recent giant Cusiana discovery.
      Both Sumapaz and Rio Blanco are characterized by rugged mountainous terrain, extensive cloud cover, and areas of dense vegetation. Only regional geologic maps were available from published sources when both projects were initiated.

Landsat MSS images from the mid-1970s had been previously processed for regional mapping, but their ground resolution was insufficient and they were too out-of-date to support extensive field operations. Landsat TM and SPOT imagery, airborne radar (SAR), and aerial photographs have been integrated successfully with field work to improve geologic mapping and field operations along the western and eastern flanks of the Eastern Cordillera, south of SantafŽ de Bogota, Colombia. Along the western flank, TM imagery was used for attempting to correct geologic interpretations of SAR images because the mountainous terrain on the SAR images was distorted due to radar layover. Structural features first recognized on the SAR imagery included: a) the surface trace of a thrust fault, b) an overturned anticline, and c) a thrust structure produced by either wedging or nappe emplacement. These surface expressions of structures were substantiated during field mapping and a subsequent seismic survey. Aerial photographs were used to determine dips, resolve questions about the geology, and construct a 1:50,000 topographic map. In the eastern area of interest, a panchromatic SPOT image was used as a cartographic base. SPOT provided excellent information on roads, land use, and forest conditions. Where cloud cover degraded SPOT, SAR images were utilized for planning. As expected, SAR images provided excellent information on geologic structure and lithology across the mountainous project area. Integration of images and maps was done manually in the western area, while integration of these data in the eastern area was done digitally within a workstation environment. Although in mountainous terrain digital integration of unrectified SAR images was difficult, mapping in a workstation environment offered significant advantages, especially in a long-term field project where interpretations and maps were continually being modified and updated.

MAPPING THE SUMAPAZ AREA
Sumapaz, the area along the western flank of the Eastern Cordillera, was evaluated with Landsat TM, stereoscopic radar flight strips, and aerial photographs. The license covers approximately 1800 km2 with topographic elevation increasing from 500 m along the western margin to 3600 m along the eastern portion. Sumapaz lies approximately 30 km east of the Magdalena River in the Upper Magdalena Basin.
       Landsat TM imagery was used to check and update a CAD basemap that was digitized from published topographic maps. The Landsat image was acquired in 1988 and was extensively processed to maximize color differentiation, emphasize topography, and penetrate haze. Color saturation was increased through the IHS transformation. A black & white image emphasizing topography was created with principal components transformation. A final color composite was generated with band 7, principal component 1, and band 5 as red, green, blue. This combination displayed vegetation as shades of green, with the most dense forest as dark green. This TM color composite was easy to understand in the field and proved useful for planning and estimating difficulty and costs of field operations. For example, in the eastern portion of the Sumapaz license, the Landsat image showed where forest cover no longer was a logistical problem as it clearly displayed where the terrain rose above treeline. In addition, clear-cut areas, agricultural plots, and burn areas were distinctly shown with brownish-red colors. There were no major roads in the area, however, larger villages were displayed as light brown areas. For geologic interpretation, the color Landsat was useful for mapping the accurate location of resistant ridges and hogbacks. TM imagery could not be uniquely correlated with lithology or stratigraphy, but was useful for recognizing relative variations in lithology within local areas.
      Airborne radar data were acquired in 1987 from a speculative survey by Intera Technologies. The flight lines were north-south and the terrain was illuminated with an east-looking radar beam. The data were collected with a ground resolution of 12m. Only 1:100,000 and 1:50,000 prints of stereoscopic flight strips were used in this project. These were manually spliced together to form a regional mosaic.
      No digital correction was made to the SAR data to compensate for layover distortions due to topography. To minimize radar distortions, a basemap overlay was manually shifted while interpreting the SAR plots. The locations of rivers and crests of resistant ridges (taken from the Landsat) were plotted on the overlay and these provided the control for this approximate cartographic correction. Geologic strike and dip were initially estimated across the field area from stereoscopic interpretation of overlapping SAR flight strips. Lithology (where field work or published maps were available) was interpreted from the SAR flight strips. The advantage of SAR imagery compared to multispectral satellite imagery for geologic mapping in cloud-prone areas such as Colombia was very advantageous.
      Structural features first recognized on the SAR imagery include: a) the surface trace of a thrust fault, b) an overturned anticline, and c) a thrust structure produced by either wedging or nappe emplacement. These surface expressions of structures were substantiated during field mapping and a subsequent seismic survey.
      Black & white aerial photographs were available for much of the Sumapaz area. These stereoscopic photographs were used to determine dips, resolve questions about the geology, and construct a 1:50,000 topographic map. Bedding attitudes determined from air photos and field work were used to constrain seismic interpretation.
      Several iterations of field mapping, reinterpretation of Landsat, SAR, and aerial photographs, and redrafting resulted in a final 1:50,000 geologic scale map of Sumapaz. The amount of redrafting was substantial as overlays at 2 scales were utilized by several earth scientists for geologic interpretation during the project. This procedure led to a very high error rate that required a substantial effort to repair. However, compared to one of the published maps available at the beginning of this Sumapaz mapping project, the new geologic map generated from this integrated project was more accurate and useful for supporting exploration.

MAPPING THE RIO BLANCO AREA
Terrain in the Rio Blanco area increases in elevation from 600m in the east to 2400m in the west. The mapping effort along the eastern flank benefited from a high resolution, panchromatic SPOT image that was acquired in 1992 for verifying well locations in oil fields operated in the adjacent Llanos Basin by Chevron Petroleum Company of Colombia. This SPOT image covered the mountainous area of interest and was used as the cartographic base. The SPOT image was an excellent source of information on land use and the transportation network, and was used extensively for planning field operations except where clouds obscured the terrain. Within the mountains there was little geological structural information visible on the SPOT image due to high sun angle (minimum shadowing) and excessive cloud cover.
      Airborne SAR data were acquired in 1992 and were used with SPOT for planning seismic operations. The SAR flight lines were North-South with the terrain illuminated from the East. The data were collected with 12m pixels. As expected, SAR images provided excellent information on geologic structure and lithology. Digital SAR mosaics were acquired from the contractor.
      The SAR mosaics and individual flight strips were loaded into an image processing/interpretation workstation as raster images and registered to the SPOT image. No attempt was made to cartographically correct the SAR for distortion due to topography. Image-to-image registration was difficult because of a lack of visible control points on the SAR and SPOT imagery, especially in the mountains. Excessive shadowing obscured the valleys on the SAR images and clouds covered many of the crests on the SPOT images. After co-registration, the airborne and satellite images were rectified to a Transverse Mercator projection using visible well pads with known Latitude/Longitude locations. Proposed seismic line locations, oil seeps, tar sand deposits, national park boundaries, wells, and major roads were embedded into the images. These informative image maps were delivered to the field at scales to 1:50,000 to assist in planning field operations, environmental baseline studies, and geologic mapping. Some seismic line locations were shifted after evaluation of the up-to-date imagery, resulting in significant cost savings. Within the workstation environment, structural interpretations of distorted SAR images were digitized into more correct cartographic positions using the co-registered SPOT images as a base.
      A published geologic map and the project's geologic map (based on new geologic observations in the field) were digitized and registered to a Transverse Mercator map projection. On the workstation these geologic maps were displayed and compared with co-registered raster images (SPOT and SAR imagery) and vector maps (proposed seismic line locations, new image interpretations, etc.) to facilitate identifying, documenting, and changing inconsistencies. As new geological and geophysical information came in from the field, interpretation and/or planning changes were easily entered into the CAD mapping files. Replacement maps and updated interpretations were continuously and efficiently generated during the life of the project.
      Landsat TM, SPOT, and SAR are effective sources of up-to-date information for supporting field operations, geologic mapping, and planning of oil and gas exploration. When SAR images are not corrected for distortions due to topography, digital integration with satellite images and maps is difficult. Co-registration of raster images and vector maps in a workstation environment offers significant advantages, especially in a long-term field project where interpretations and maps are continually being modified and updated.

ACKNOWLEDGEMENTS
We acknowledge the support of Chevron Petroleum Company of Colombia and Chevron Overseas Petroleum Inc. (COPI) for these projects and for granting permission, along with Ecopetrol, to allow this publication. Luc Lalonde of Intera Technologies was particularly helpful with the SAR images. Mark Choiniere of COPI assisted with CAD and Howard Hou-Lap Wong of Overseas Petroleum and Investment Corporation assisted in geologic interpretation of the SAR images of Rio Blanco.

About the Authors:
Jim Ellis is the supervisor and Peter Goodwin is the remote sensing geologist of Chevron Overseas Petroleum's remote sensing group.
Wayne Narr is a senior geologist, specializing in structural geology. Gildardo PŽrez is the exploration manager of Chevron Petroleum Company of Colombia in SantafŽ de Bogota, Colombia. Over the past several years they have worked together on several remote sensing projects across Colombia utilizing Landsat MSS and TM, SPOT, airborne radar, space shuttle radar, published maps, aerial photographs, and field work to improve operations and reduce exploration risk.

Back