This border area, also known as the Triple Frontier, is unique for harboring two of the largest remaining stretches of Atlantic Forest: the Iguazú National Park, created by Argentina in 1934; and the Iguaçu National Park, established by Brazil in 1939.

 

With about three million annual visitors, the two parks are among the most visited national parks in Latin America. Visitors are mainly attracted by the 2.7-kilometer-wide Iguazu Falls located at the Argentine-Brazilian border and shared by the two protected areas.

Together, the two parks protect over 223,657 hectares (about 863 mi²) of forests. These numbers, however, conceal a history of deforestation and reforestation that affected both parks since their establishment in the 1930s.

 

Until the early 1970s, the Iguazú National Park in Argentina included an urban settlement, known today as Puerto Iguazú, inside its boundaries. The park was gazetted not only to protect the Iguazu Falls and the forests surrounding it, but also to help the colonization of the border. Thus, the Argentine national park agency maintaned settlements inside the park territory until the early 1970s, when a change in the paradigm of national parks led the government to retrace Iguazú's boundaries to exclude the town.

Across the river in Brazil, the Iguaçu National Park experienced a decades-long legal dispute between federal and state governments over ownership of the public land in the park. The legal impasse prevented the federal government from taking control of sections of the park, which then became a destination for settlers moving from other areas of the country. In the 1970s, the Brazilian federal government finally took control of the land and initiated a process of eviction of the 2,500 people living inside the park territory.

Today, images of the border landscape, such as the one on the left, can lead observers into seeing parks composed of pristine forests. It can be misleading, as these present-day images conceal the historical record of land use inside the parks. This project, therefore, utilizes the three sets of sources—Fixed Wing, Corona, and Landsat imagery—to unearth the history of land use in the two national parks.

This map shows a mosaic composed of 235 images taken from airplanes in 1953. The images were produced by the State of Paraná's Land and Colonization Agency (Brazil) as part of a survey that covered the entire state. A second layer with the land cover classification overlays the aerial images, diferentiating cleared land, fields, forested area, urban, and waterbodies. 

 

Almost two decades after the establishment of the parks, the entire area was still heavily covered by forest, making it difficult to identify the parks’ boundaries. On the Brazilian side, roads had recently being opened by federal and local governments to promote colonization, which brought a first wave of settlers into the region.

This area north of the river is part of the Brazilian national park, Iguaçu. At the time, it was a heavily forested area. Close to the northern limit of the park, one can see a few clearings, a sign of the first settlers arriving in the area.

 

<--- Using panning and zooming, navigate through the area of the park. Zooming closer allows you to see the original aerial images without the colored layer with the land use classification. You can explore in this way all the maps in this visualization.

This second set of 334 airplane images were produced by the government of the State of Paraná (Brazil) as part of another state-wide survey in 1980. Twenty seven years later, the landscape presented a radical change. The area outside the Brazilian national park had already being completed colonized by settlers from southern Brazil. Forest was converted into farmland, and towns sprouted along the roads opened in the 1950s outside the park. Settlers, however, stopped at the border as there are still few clearings in Argentina.

 

This frame shows the same section shown in 1953. Now the area of the Brazilian Iguaçu National Park (north of the river) is full of clearings. By 1980 most settlers had already been evicted by the Brazilian federal government, but the marks of farming on the park landscape were still present. Outside the Brazilian park, the transformation is radical—what was mostly forest in 1953 became is farmland in 1980. In Argentina, however, little was happening outside the boundaries of the Iguazú National Park, as colonization there still had to catch up with Brazil.

 

<--- Explore other areas of the parks.

Figure 1 shows the change between 1953 and 1980 within each category of land classifications, in the area of the park and selected outside areas covered by the two sets of aerial imagery. The chart shows a clear increase in the cleared area, which combined with the information from the maps, indicates most of the change occurred outside the national parks.

 

Figure 2 is the aggregate of land classifications and the change in the classes of land cover between the two periods. Check the interactive Tableau version of these charts.

This Corona image comes from a different source, the U.S. spy satellite program active between the late 1950s and early 1970s. This image, from 1967, records an intermediary moment betwen the two periods seen so far, 1953 and 1980. The mid-1960s brought about an influx of farmers to the area. A large amount of clearing was occurring outside the boundaries of the Iguaçu National Park in Brazil, with pockets of development inside the park. However, the settlements in Iguaçu had still a few years before reaching the scale of the late 1970s, when the government evicted the last settlers from the park.

 

The imagery from the Corona spy satellite collection was declassified in the early 1990s and acquired through the USGS. It boasts resolution nearly as high quality as the fixed wing imagery, an impressive performance for a 1960s satellite. Details of the Corona imagery can be found here. 

For the Brazilian Iguaçu National Park, encroaching from nearby adjacent towns was another problem. The northeastern corner of the park was an example. As settlers started to arrive in the region, many began to clear areas inside the park. The town of Santa Tereza started to grow inside the park around this time.

 

<--- Explore other areas of the two parks in 1967.

These land classification polygons were manually traced, based on the Corona spy satellite imagery dated 1978, which were also declassified in the early 1990s.

 

Inside the Brazilian park, the area known for the more dense populations is notably larger than in the 1967 image. Also, there is an increased number of pockets of cleared area along riverbanks. Outside the Iguaçu, most of the land was already converted to farmland.

 

In Argentina, the government changed the boundary of the park in 1972 to exclude the town of Puerto Iguazú from the territory of the park. There are also the development of clearings in the eastern and southern boundaries. In the east, farmers coming from Brazil and from other parts of Argentina constitued settlements in the area. In the southern edge of the park, outside its limits, one can see the marks of the new tree plantations that began operating there.

 

Eleven years later, this northeastern corner of the Brazilian national park has seen two developments. Part of the clearings were halted, allowing the regrowth of vegetation. On the other hand, the town of Santa Tereza has grown, which would lead to the retracing of the park boundaries in 1981 and the exclusion of this urban settlement from the limits of Iguaçu.

 

This follow a pattern analogous to what happened in the Argentine park, Iguazú, a decade before, when the borders of the park were retraced to exclude the city of Puerto Iguazú, originally located inside the park.

 

<-- Explore the map.

The comparison of Corona imagery displays marked changes from 1967 and 1978, when the majority of the eviction of settlers was taking place from the two national parks. 

 

Figure 3 displays the comparison of cleared land, field or agricultural land, urban areas, waterbodies, and forested land. While the designation of land types was difficult at times with the black and white imagery, the overall rising and falling patterns of land classification changes corresponds with what we know to be happening in the area. 

 

Figure 4 is the overview of all of the classification types collected from this imagery in the two years. Check the interactive Tableau version of these charts.

Landsat imagery has much lower resolution than the fixed-wing or Corona imagery. However, Landsat features multiple layers representing spectral bands over the same area. Landsat resolution also varies between the three images used here. Landsat 1 (1973) and Landsat 3 (1986) use a four band sensor, MSS, with a 60 m resolution. Landsat 8 uses a nine-band sensor, OLI, with a 30/15 m resolution. While the years of imagery collected seem like they could be compared from image to image, the difference in resolution between fixed wing, Corona, Landsat 1 and 3, and Landsat 8 is too great, so it was decided not to compare directly the results.

 

The fixed-wing and Corona images are monospectral—they have only one band, grayscale. Landsat images, on the other hand, have multiple bands, including near infrared and thermal infrared (in the case of Landsat 8), and allow for various types of supervised and unsupervised land cover classification. This classification performed in the 1973 image captured the cleared lands and fields inside the Iguaçu National Park. The resolution is coarse, as it comes from the first generation of Landsat images.

The 1986 land classification analysis was based on Landsat 3 imagery. A supervised classification method was used to define the parameters of the classification and create the output. 

 

The image is over five years after the final evictions took place in the Brazilian National Park. It is evident the disappearance of the cleared field areas through the advanced encroaching of forest.

 

In Argentina, the eastern area outside Iguazú started to show a scattered pattern of clearings, resembling the Brazilian side in the late 1960s.

In the 2010s, with the recovering of the vegetation in the areas inside the Iguaçu National Park cleared in the 1960s and 1970s, the difference between park and non-park landscape is accentuated. While the Brazilian park is now mostly covered by forest, the area outside it in Brazil is completed dominated by farmland.

This series of land cover assessments was based on three different types of sources not commonly used by historians: aerial imagery (fixed wing), spy satellite imagery (Corona), and scientific satellite imagery (Landsat). By combining the three types of source is possible to reconstruct the evolution of landscape in this borderland space. Forests were cleared and regrew according to social and political processes playing in and around the national parks created in the region in the 1930s. This is a story where borders—national and park ones— played a crucial role into shaping the geography of the region. The landscape of the two parks, and of the remnants of Atlantic Forest countained by them, is the result of changes in the planning, demarcation, implementation, enforcement, and transgression of these boundaries.

 

About this research

 

Frederico Freitas (Principal Investigator)

Emily McNamar (Research Assistant)

Juliana Quist (GIS Specialist)

 

Past Research Assistants (Spatial History Project at Stanford University)

Peter Salazar, Eli Berg

 

Previous Iterations of this Visualization

Boundaries of Nature (2014)

 

Selected Sources

• Freitas, Frederico. “Boundaries of nature [electronic resource]: national parks and environmental change at the Argentine-Brazilian border, 1890-1990.” PhD Dissertation, Stanford University, 2016. (Online at https://searchworks.stanford.edu/view/11686029)
• IBDF. Plano de Manejo Parque Nacional do Iguaçu. Brasília: IBDF, 1981
• IBAMA. Plano de Manejo do Parque Nacional do Iguaçu. Brasília: IBAMA, 1999.
• Luis A. Giúdice. Plan de Manejo Parque Nacional Iguazu - Proyecto Planificación Y Gestion de Los Parques Nacionales (APN-FAO). Buenos Aires: Administración de Parques Nacionales, 1988.