Context
The Amazon Forest is one of the largest ecosystems on Earth, spanning an area of about 2.7 million square miles, of which 2.3 million square miles are rainforest. Its boundaries encompass nine Latin American countries and approximately 3,000 indigenous territories. The majority of the land lies in Brazil, which accounts for 60% of the territory. However, despite the forest’s crucial role in mitigating global climate change, its area has been significantly deforested since the 1970s, at the risk of irreversible environmental damage.
The map focuses on the Brazilian portion of the Amazon rainforest. It aims to visually showcase the deforestation rate, the expansion of anthropic lands—including urban areas and agriculture—and regions burned by wildfires over the past 40 years, documented in five-year increments. The data and subsequent analysis are presented in a chronological series of maps, creating a timeline that overlays the three variables.
Data
The primary sources of data on Amazon’s environmental issues include official governmental agencies, international cooperation groups, and independent non-profit institutions. There are two main formats of data: numerical, either as absolute information or incremental proportions, and satellite imagery. The map was designed using tabular (text-based) data and optimized for visualization in Tableau. Furthermore, because the images show the precise location of land types, the information is spatially dispersed (shown above), not aggregated, making proportion comparisons difficult.
The datasets identified with the most relevant information were sourced from IBGE —Brazil’s national statistical institute— and MapBiomas, an organization that unites NGOs, universities, and startups to monitor environmental changes through spatial analysis using satellite-based technology systems. Other considered data sources included the independent organization Global Forest Watch, RAISG, the Amazon Geo-Referenced Socio-Environmental Information Network, and NASA.
Although considerable data is available regarding Amazon’s transformation over the years, the challenge has been finding information at the municipal level. There are nine states within the legal Amazon limits, each exhibiting distinct patterns of land use. Thus, the state-level analysis proved insufficient for understanding localized patterns. In addition, the aim was to find data from as far back as possible, dating to the 1970s, which is considered the beginning of deforestation, characterized by the inauguration of the Trans Amazonian Highway’s first section (e.g., Laurance, Patowari). MapBiomas has provided a great deal of information since 1985, and it is documented annually.
In addition, there were essentially two ways to measure wildfires, by area and by frequency. The decision was made to focus on area, facilitating easier comparisons with the other two datasets, the natural and anthropic regions. Lastly, the data chosen for analysis was derived from tabular information and wasn’t managed or aggregated at any level. The solution was to use Python to create a spreadsheet for data cleaning.
The Map
The resulting map is a chronological series that can be navigated by the years 1985, 1995, 2005, 2015, and 2022 (the last year with data). Each municipality is colored with a green gradient to indicate the extent of natural vegetation. Areas affected by wildfires are represented as pink translucent circles. The challenge was to overlay the variables, demonstrating a visual correlation without hindering one another.
User Impressions
Two users —one North American and one Brazilian— were invited to provide feedback on several versions of the map. The first user made observations about the square shapes, which suggest population density data, while the circles are more likely connected to forest fires. Also, the pink color created a more harmonious combination with the dark green than the orange or yellow, yet it remained related to fires. Moreover, she favored the option with an outlined circle representing anthropic land, which could result in a more visually appealing map and not as busy.
The second user agreed with some of the first user’s points but disagreed about the square observation. He believes it’s a better way to compare two areas. He also enjoyed the proposal to add anthropic land to the map. However, he was more concerned about having too much information overlaid in different ways of representation at the same time. It was suggested to eliminate the gradient of green in the name to keep the information represented in shapes only. “Not having green in a forest map is a deal breaker,” was the answer. He then suggested varying the color between the outlined circumference and the solid one.
Next Steps
Taking all the considerations into account, a potential solution would be to create separate maps at different scales: one for a general overview of the Brazilian Amazon and another detailed map focusing on anthropic areas and land use breakdowns. This way, a map of the entire Brazilian Amazon showcases only the wildfires by municipality, overlaid by a gradient of green. The second map could focus on a strategically notorious location, such as the states of Mato Grosso or Pará, to produce a larger-scale representation. Information about the anthropic area and its proportion of farming lands can be added to this map in a color distinct from the wildfire boundaries, also overlaid by a gradient of lighter green.
In addition, to provide a more tangible understanding of the impact of a wildfire, several charts illustrating the spatial data shown in the maps could be added as supportive information. Moreover, in conjunction with the maps and secondary charts, the study could benefit from incorporating a timeline of significant political milestones, which could enrich the analysis.
The final product will be presented in an interactive map, visually showing the changes in the ground patterns over the past four decades. It will be necessary to design the overlayer representation and determine how to control such comparisons.
References
Articles and Web Pages
Our World in Data. “Cutting Down Forests: What Are the Drivers of Deforestation?” Our World in Data. Accessed April 20, 2025. https://ourworldindata.org/drivers-of-deforestation#cutting-down-forests-what-are-the-drivers-of-deforestation.
Daley, Jim, and Andrea Thompson. “Deforestation Intensifies Warming in the Amazon Rain Forest.” Scientific American, 2019. https://www.scientificamerican.com/article/deforestation-intensifies-warming-in-the-amazon-rain-forest1/.
InfoAmazonia. “Deforestation in the Amazon: Past, Present, and Future.” InfoAmazonia, March 21, 2023. https://infoamazonia.org/en/2023/03/21/deforestation-in-the-amazon-past-present-and-future/.
Xu, Yue, Daniel B. Wright, and D. Luke Urban. “Identifying Predictors of Forest Loss in the Contiguous United States.” Environmental Research Letters 17, no. 3 (2022): 034037. https://doi.org/10.1088/1748-9326/ac4df9.
Data Sets
Dados.gov.br. “Portal de Dados Abertos.” Accessed April 20, 2025. https://dados.gov.br/dados/conjuntos-dados.
Global Forest Watch. “Brazil Country Dashboard – Fires.” Global Forest Watch. Accessed April 20, 2025. https://www.globalforestwatch.org/dashboards/country/BRA/?category=fires&location=WyJjb3VudHJ5IiwiQlJBIl0%3D.
Instituto Socioambiental. “RAISG Mapa Online.” Socioambiental.org. Accessed April 20, 2025. https://www3.socioambiental.org/geo/RAISGMapaOnline/.
MapBiomas. “Maps and Data.” MapBiomas Brasil. Accessed April 20, 2025. https://brasil.mapbiomas.org/en/downloads/.
NASA Earthdata. “Earthdata Search.” NASA Earth Observing System Data and Information System (EOSDIS). Accessed April 20, 2025. https://search.earthdata.nasa.gov/search?gdf=Excel&lat=-22.194090572094453&long=-131.34375&zoom=1.
Podcasts
Podcasts
Rádio Novelo Apresenta. Sob Estresse [Under Stress]. 2023. Podcast. https://open.spotify.com/episode/6KUwUHS5nZX1ZeAYH63Ymo?si=ODTrUjfNRXeMDK9S4BlT8g.
Rádio Novelo Apresenta. Salles, João Moreira. 2025. Podcast. .https://open.spotify.com/episode/5vHfEZoyNUGmVUsnN30rPQ?si=V4sSrrdeRtKWFF_gjAYImw.