Advances on Earth Observation and Artificial Intelligence to Map Unofficial Roads in the Brazilian Amazon Biome
Abstract
Roads built by the Brazilian government in the 1960s integrated the Amazon biome to the rest of the country. These official roads cut pristine forests and connected the region to the main cities in Brazil. As a result, a new development frontier initiated with cattle ranching, agrarian settlements, logging, and gold miners. By the end of 1990s, a new deforestation frontier was consolidated along the fringe of the Amazon biome with the Cerrado grassland biome, creating the so-called arc-of-deforestation. Then, a second boom of road expansion emerged in the early 2000s with the new plans to improve the road infrastructure of the Amazon region, notably along the Santarém-Cuiabá BR-163 corridor for soy transportation, and along the BR-319 connect Manaus to Porto Velho. Much attention to studies about road impact (positive and negatives) has been on official roads. However, the expansion of spontaneous secondary roads, mostly unofficial, opened by loggers, miners, cattle ranchers, and to grab illegally public lands, represents a much larger threat to the Amazon biome, as well as to traditional population and indigenous people. Our research group has applied a satellite image protocol to map unofficial roads using satellite image interpretation. Until 2016, we have mapped 551,646 km in addition to 39,421 km of official roads (36% of federal jurisdiction and 64% state one). The unofficial roads surpass the official ones by almost 13-fold, creating an extensive human footprint in the Amazon region. Most of these new roads did not have a legal permit and lack an assessment of their environmental impact. The unofficial roads accelerated even more deforestation, illegal logging and fires, beyond the impact of official roads. However, the official roads are the main point of expansion of unofficial roads. Here, we demonstrate that the roadless landscape in the Brazilian Amazon is much smaller than what has been estimated considering only official roads, increasing the risk of deforestation, forest fragmentation and fires, and threatening local and indigenous communities. It is urgent to expand the monitoring of unofficial for the entire Amazon biome to better understand their social and environmental impact. This is technically feasible with our latest progress on applying deep learning algorithms to detect and map roads from space which will allow operationalizing road mapping and monitoring in the entire Amazon biome with Earth Observation Systems.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2020
- Bibcode:
- 2020AGUFMGC106..09S
- Keywords:
-
- 1622 Earth system modeling;
- GLOBAL CHANGE;
- 1630 Impacts of global change;
- GLOBAL CHANGE;
- 1632 Land cover change;
- GLOBAL CHANGE;
- 1640 Remote sensing;
- GLOBAL CHANGE