Jeffrey C. Bjerke

Border Mapping Project


ABSTRACT

A signed border agreement by Ecuador and Peru was mapped by the National Imagery and Mapping Agency (NIMA). The project covered an area approximately 30 minutes by 45 minutes at a scale of 1:25,000. The project required a collective NIMA approach to solve photogrammetry problems, database limitations, and vendor Commercial-Off-The-Shelf (COTS) software problems. The source materials were brought to NIMA Saint Louis, Missouri, to begin the border delineation project in August 1999. The materials brought by the Ecuador and Peru delegates included more than 50 nine inch by nine inch frame images and associated control points. The imagery was taken in different years at different scales using different cameras, and of varying sidelap and overlap. The NIMA personnel scanned the images, tied the image scenes together through sidelap and overlap, and controlled the project to form a block triangulation solution. The NIMA analysts collected over five thousand drainage vectors, roads and trails, escarpments, spot elevations, and buildings. The analysts also created a one arc second Digital Elevation Model (DEM) from each stereo model to support twenty-meter contours. The maps were finished through a process that utilized a suite of scripts to symbolize the features. The output of the script produced raster representations of the collected data for color plots and TIFF files to create film negatives for mass printing.

ECUADOR - PERU BORDER BACKGROUND

According to John Gates, NIMA's liaison to the U.S. Southern Command, a few historians believe that the territorial dispute between Ecuador and Peru can be traced back to the early 1800s. Since that time, there have been frequent conflicts and attempts to negotiate a solution. In 1941, an uneasy truce broke down and Peru invaded Ecuador in a short but bloody war. An ad-hoc coalition of interested nations, known as the Guarantor Nations, was made up of Argentina, Brazil, Chile, and the United States. They helped Ecuador and Peru resolve the conflict.

Given the world politics at the time, the Guarantor's mission was to bring about a lasting peace and stabilize the region. In January 1942, with Guarantor oversight, Ecuador and Peru signed the "Rio de Janeiro Protocol for Peace, Friendship, and Borders". The Rio Protocol is a short document taken up in large measure with delineation of the border. There are no annexed maps to the Protocol, so everything relied on an accurate description of the border.

For several years following 1942, a commission from Ecuador and Peru carried out geodetic surveys and placed more than 150 border monuments, effectively demarcating almost 90 percent of the 1,420 kilometer border.

Seeing the urgent need for better maps over the Cordillera del Condor and other areas of the frontier, Dr. George McBride, the U.S. technical advisor to the Guarantors in the 1940s, sought the help of the U.S. Army Air Force to photograph and map the border.

From 1943 to 1946, numerous sorties were flown and the first accurate topographic map of the Cordillera del Condor was compiled. However, that effort had a tragic price. Two aircraft crashed in separate incidents killing 14 U.S. airmen. The new map was a significant contribution to knowledge of the region, but did not end the controversy.

More and better maps were made in the intervening years through the efforts of the Ecuadorian and Peruvian geographic institutes in collaboration with the Inter American Geodetic Survey (IAGS). IAGS was formed in 1946 in response to Department of Defense and the Pan America Institute of Geography desires to assist Latin American mapping efforts.

"All went relatively well until they came upon the area known as the Cordillera del Condor (Condor Mountain Range)," Gates said. An extremely isolated area on the eastern flank of the Andes, it was poorly mapped due to its rugged jungle terrain, almost permanent cloud cover, and sparse population. In this area of the frontier, the border runs along a watershed line between two major river systems. Because of the lack of accurate maps, this watershed line was poorly described in the Protocol.

Ecuador Peru Conflict Area[1]

In 1960, Ecuador abandoned the Protocol process because it felt that the Protocol was not executable due to a perceived error in the description of the watershed line in the Cordillera del Condor. Since then, Ecuador and Peru have had years of uneasy peace and periods of outright hostilities. Two major outbreaks occurred in 1981 and 1995.

"The solution we have today came as a result of diplomatic negotiations begun after signing a peace declaration in February 1995," Gates said. "The final solution of the border controversy depended on arriving at a delicate balance between the spirit and letter of the Protocol, and some innovative thinking by the negotiators."

"The images, topographic maps and visualization systems were important contributions to the diplomatic process," said Gates. Products used in the negotiations and to support Military Observer Mission Ecuador Peru (MOMEP) include new 1:100,000 scale Topographic Line Maps, LANDSAT image maps, Radarsat image maps, a flythrough, and scores of customized map and image products made on the Geospatial Information workstation and Remote Replication System in Panama.

The overall peace agreement includes treaties on Commerce and Navigation, Border Integration, and Mutual Confidence. After numerous rounds of technical and juridical presentations by both sides, the Guarantors gave their recommendations as to where the border should run in May 1998. "It was accepted by both sides except in a small area on the Peruvian side of the Cordillera del Condor surrounding a small border post named Tiwintza," Gates said. Tiwintza had been abandoned after the 1995 conflict under terms of the separation of forces. However, Tiwintza holds special significance to Ecuador because Ecuadorian soldiers defended it and had buried their dead within its perimeter.

"Both sides were intransigent over this section of the border and the talks soon stalled," Gates said. "Tensions grew during a period last summer [1998], and it was feared that armed conflict was imminent."

The Guarantors and the two presidents Jamil Mahuad of Ecuador and Alberto Fujimori of Peru overcame the impasse. In a flurry of shuttle diplomacy, Peru and Ecuador agreed to accept a final Guarantor solution that includes adjacent peace parks straddling the border along the Cordillera del Condor and a Peruvian secession of a one square kilometer area encompassing Tiwintza, not as sovereign territory, but a private property transfer.

Dignitaries representing South American nations, the United States, the Vatican and Spain witnessed the long awaited signing of the peace agreement between Ecuador and Peru on October 26, [1998] in Brasilia, Brazil. [2]

On May 12, [1999] Ecuadorian President Jamil Mahuad agreed with Peruvian President Alberto Fujimori on the location of the Cordillera del Condor border, and signed the "Final Demarcation Act of the Common Border". The agreement gives Ecuador a contested area called Tiwinza. [3]

THE PROJECT

In the summer of 1999, the U.S. Department of State requested the National Imagery and Mapping Agency (NIMA) to produce maps covering the Cordillera del Condor watershed area. The NIMA production facility would host daily visits of the Ecuador and Peru delegates during map production. The task for NIMA was to create maps delineating a new International border for Ecuador and Peru as defined in the "Final Demarcation Act of the Common Border" as the watershed of the Cordillera del Condor mountain range. The project brought together Ecuador and Peru delegates, nine-by-nine aerial photography and control points from each country, innovative mapping solutions, and the NIMA mapping community to complete the historic project. The project mapped the landforms, all drainage features, any buildings, roads, spot elevations, twenty-meter contours, and border depiction. The project commenced on 23 August 1999 and completed 19 October 1999.

NIMA personnel scanned more than fifty nine-by-nine aerial film positives using a "Helava" scanner. These scanned images and the parameter files were brought to the production environment. The production environment utilized four SUN[TM] Microsystems, Ultra[TM] 2 workstations for the project. The scanned images and parameter files were loaded to the production workstation server Redundant Array of Inexpensive Disks (RAID). NIMA analysts performed the orientations of the stereo source and added tie points in the stereo overlap. Following this step, the analysts added the supplied control points and created five separate image strip projects using BAE Systems, SOCET SET® software. Each image strip contained from five to eight stereo pairs, all obtained during the same flight path. When attempting to merge the flight strips using common endlap and sidelap tie points, the residuals would exceed acceptable tolerances. This was primarily due to sidelap coverage of less than 5% in many cases.

With the delegates in attendance, the analysts began collecting data from one image strip on 31 August 1999. Each workstation exploited one stereo pair to collect three-dimensional drainage vectors by delineating the valleys. Through discussions with the delegates, intensification of drainage vectors closest to the watershed were requested. Collection was restricted to approximately 1 mile from either side of the ridgeline to reduce the project area dimension and timelines. By refining the area of collection near the watershed, this reduced the data collection in well-defined watershed regions of steep terrain where the border was not contested. Stereo collection continued until nearly all the drainage vectors were complete.

An outline of the stereo pair coverage was digitized to assist the analysts in stereo collection. The outline minimized over-collection of data in regions of stereo overlap.

To show the work completed, check plots were periodically generated and displayed to the delegates. To accomplish this, an analyst arrived one hour before the day shift to merge the drainage vector databases into one database. A plot was generated including the stereo outline along with the merged drainage feature vectors. This method was valuable in estimating the overall project completion on a model basis. By performing the feature merge, it exposed a SOCET SET® database limitation of 5,000 feature vectors. To overcome this, the analysts maintained smaller separate databases of feature vectors while attempting to minimize overlap between the feature databases.

When the drainage feature collection was nearing completion in several models on different flight strips, the analysts began to notice elevation differences between the flight strip models. At this point, the production personnel acknowledged that a block photogrammetric solution was required. A management decision was made to continue collection of drainage features since it would take away from precious collection time to stop the production while the control production environment assembled the block photogrammetric solution.

A tasking was issued to control the project as a block photogrammetric solution. Additional control points were required, so the control personnel gathered all available control points generated from previous MOMEP projects.

The project managers held meetings with the delegates to inform them of the production problem. This was not difficult as one of the delegates understood photogrammetry very well and translated the production problem to the other delegates. At this point, the delegates mentioned that additional aerial nine-by-nine photography was available that crossed some of the original flight strips. The addition of the cross-flight imagery enabled the control production analysts to insert additional tie points in the stereo bi-lap and tri-lap to generate the block solution. After multiple iterations the control analysts were successful in generating a block photogrammetric solution. The solution, the new scanned cross-flight imagery, and parameter files were saved to tape. The tape was loaded to the production environment where the project continued, with some major adjustments.

In parallel with the block control solution, the production analysts continued to collect features using the original strip projects. Additionally, the production analysts researched and found an option in SOCET SET® called Feature Registration. This option is used when improved sensor models or additional control is provided in a project. This software allowed the existing 3D extraction features to be salvaged by transforming the features from the original strip stereo models to the new block model using 3D adjustment vectors. The "adjustment vectors" shift the collected data from its original position into the improved stereo block model position. Through the use of the Feature Registration option, the production analysts were able to retain more than two weeks of intense vector collection. The adjustment process took approximately seven days, but the other solution was to rework the entire vector collection of nearly three weeks work. In some areas of the block solution, the stereo collected drainage vectors were beyond reach of the original strip collected vectors. To work through this problem, the analysts either manually adjusted the feature vector to the new surface node by node, or deleted and recollected the feature vector.

Upon completion of the feature vector adjustments, the next process was to generate a one arc-second terrain matrix to support twenty-meter contour intervals. Using the SOCET SET® software, the production analysts executed Automated Terrain Extraction (ATE) to generate the one arc-second terrain matrix with terrain posts at or near the observed reflective surface. The ATE correlator processed terrain successfully in some areas of the project and marginally in other portions. Optionally, the ATE software has the capability to implement geomorphology. Geomorphology assists the ATE process by modifying the generated terrain matrix utilizing the three-dimensional feature vectors. The geomorphology process is useful in regions of poor to marginal correlation.

Geomorphology was also utilized following ATE to adjust regions of terrain posts that were above or below the imaged stereo surface. The adjustment associates the terrain to the three-dimensional extracted feature vectors by utilizing buffer parameters relative to the feature vectors. The process adjusts the terrain posts in areas where the ATE correlator failed to place the terrain posts on the surface. Following the ATE process, the analysts manually review and adjust each terrain elevation post, post-by-post to match the observed surface. This process is known as "pounding posts" for obvious reasons.

MONUMENTS

The delegates provided differential Global Position System (GPS) survey positions of the Hito's (Spanish for monument) that were placed on Cordillera del Condor in the spring of 1999. The signed "Final Demarcation Act of the Common Border" agreement stated that the watershed defined the border and therefore would not be a straight line between the monuments. Plateaus on the mountain posed problems in portrayal of the watershed and this is where the production analysts stereo collection experience truly paid off.

The production analysts collected a "preliminary" border by traversing the stereo terrain using the drainage vectors, the contours generated from the DEM, and the highest visual elevation in the stereo imagery. The delegates reviewed the "preliminary" watershed vector and through discussions, agreed to a final border feature vector placement. This phase required true diplomacy between the managers, analysts, and the delegates. When the analysts and delegates approved all the feature and terrain collection, the data was exported from SOCET SET® using ASCII ARC_GEN format to begin the next phase of the project, map finishing.

Project finishing utilized ARCEDIT[TM] to join the feature vectors along the image stereo limits. The collected feature vectors were not continuous due to the use of individual stereo model collection and the use of multiple feature databases. A complete and thorough review of the extracted vector data exposed gaps, slivers, overshoots, and undershoots. The joining and deleting of vector dangles were edited quickly using Esri ARCEDIT[TM] tools. At times, the analyst would return to the stereo model to verify the appropriate action.

After a complete data inspection, the feature vectors were prepared and processed using the Nominally Attributed Topographic Example (NATE) software in the production environment. The NATE software is a set of Arc Avenue Macro Language (AML) scripts that portray and symbolize the data. Using the NATE software the UTM grid, UTM grid reference box, magnetic declination diagram, scale bar, and symbol legend were placed on each of the five final maps. Further processing placed internal text for river names, country names, notes, country seals, and a signature block on the maps. Periodically, plots were generated using ARCPLOT[TM] for the delegates to proof the map textual data. The five 1:25,000 map sheets were saved in Tagged Image File Format (TIFF[TM]) for mass printing at NIMA.

Parallel to the finishing process, a fly-through was prepared that covered the project area. An image mosaic of the nine-by-nine imagery was created and merged with the terrain matrix to generate a three-dimensional orthorectified model of the project. The image mosaic was resampled to five-meter pixels to reduce the overall file size.

A signing ceremony was held in NIMA Saint Louis, Missouri on October 19, 1999 using four copies of each finished "NATE" map. The ceremony brought closure to the Cordillera del Condor border project. At the conclusion of the ceremony, the fly-through was shown to the delegates.

The project's success was assured by the delegates participating in the production process to resolve any lingering issues. Having the delegates in-reach and on-sight eliminated the time lag in resolving sensitive border issues. Equally important to the success of this project were the coordinated efforts of the U.S. State Department and the NIMA organizations CO, GI, ST, IS, and MS.


References

1. http://www.lib.utexas.edu/Libs/PCL/Map_collection/americas/Peru_Ecuador_81.jpg>

2. NIMA Edge. (January 1999), "Ecuador - Peru: NIMA Plays A Major Role in Historic Peace Accord".

3. NotiSur. (14 May 1999), "Mahuad Agrees To Peru Border" , (20 May, 2000).


Jeffrey C. Bjerke
Geospatial Analyst
National Imagery and Mapping Agency
4600 Sangamore Road/OGIRA/MS D-120
Bethesda, MD 20816-5003