Forest inventories are technical studies that enable a comprehensive understanding of a forest by, identifying, quantitatively and qualitatively, which resources it has. The collected data is essential to the planning, handling and use of that area, whether for preservation, sustainable exploration or recovery of degraded areas.
However, carrying out the inventory by traditional methods is usually quite laborious. It involves at least 73 days of effort and a team of five professionals to map an area. At each day of work, around 20 hectares are mapped.
The news is that this process has become more agile, cheaper and accurate by means of artificial intelligence and drones. The Empresa Brasileira de Pesquisa Agropecuária (Embrapa) [Brazilian Agricultural Research Corporation] from Acre, has developed a technology, named Netflora, which uses technological resources to map up to 3.500 hectares per day, producing species recognition, geographic location of trees, metrics and maps, at a speed of 2 hectares per second. The initiative is led by agricultural engineer Evandro Orfanó.
The new tool has already mapped areas in Acre, Rondonia and Amazonas. Over the course of one year more than 70 thousand hectares were inventoried. The goal is to reach other states, in partnership with the other Embrapa Units in the Amazon.
CASTANHA-DO-PARÁ [BRAZIL NUT]
In the state of Amazonas, the work was carried out in partnership with Embrapa Amazônia Ocidental [Embrapa Western Amazon] and the Secretaria de Estado do Meio Ambiente [State Secretary of Environment], in the Uatumã Reserva de Desenvolvimento Sustentável - RDS [Sustainable Development Reserve]. Within this área, 604 Brazil nut trees (Bertholletia excelsa) and more than 14 thousand other arboreous species were identified in 1.150 hectares of forest. The extractivism of Brazil nuts, plays a vital role in the local bioeconomy, providing the main source of livelihood for several families in the region.
Therefore, the next step of the project, whether in the Uatumã RDS or in other mapped areas is to offer a mobile app for extractivists, who will have access to the digital inventories. The platform will allow to visualize the exact location of the trees and to find one’s way in the forest just as one “Navigates” in a city, searching for a location, considering that each mapped tree will have a unique “address”.
This will allow workers to locate the species of interest more accurately, optimizing routes, reducing physical effort on long walks and facilitating collection. Extraction areas can be expanded and exploitation can become more efficient and also more sustainable, with reduced environmental impacts and monitoring for preservation.
Marcondes Costa, extractivist of brazilian nuts from “Nova Califórnia” district in Porto Velho, Rondônia, participated in the mapping of the area, where he works with technicians from Embrapa. “I think this mapping will help us a lot, because we will find the trees that are off our trail. Knowing where the Brazil nut trees are, those where we haven’t reached yet, will make things much better. Having this mapping in a mobile app will contribute not only for me, but for everyone doing the same job, because there are many collection teams ”, he says.
Project aims to optimize extractive collection
In Uatumã RDS, the work was carried out in partnership with another project, from Embrapa Amazônia Ocidental, led by forestry engineer Kátia Emídio and sponsored by Fundação de amparo à pesquisa do estado do Amazonas [Amazonas state research support foundation]. Named “Otimização da Coleta Extrativista da Castanha-do-Brasil no Amazonas” [Brazil nut Extractive Collection Optimization in the Amazon], the initiative aims to improve nuts extraction and transportation techniques, which are generally carried out with a low level of technology and great physical effort by extractors.
The idea of the project is to evaluate the operational and economic costs of installing aerial cables, similar to zip lines, to carry out the activity in the Reserve, Thus contributing to improved ergonomy for workers and greater agility in the work. Traditionally, extractivists carry bags or baskets of nuts, which can cause harm over time, such as back pain.
LOCALIZATION
Though, to use cable technology, It is essential to know precisely where the Brazil nut trees are. At this point, the mapping made by using Netflora’s methodology comes in. “To optimize the collection, We need to know how Brazil nut trees are distributed in the space, which places are difficult to access, which areas are flooded during the extraction period, or areas with very steep slopes, or even the presence of Igarapés. Cable technology can be extremely important in this activity, reducing the efforts of extractivists. And the inventory is the starting point for everything, because, by identifying where the Brazil nuts trees are, I can optimize the route of the aerial cables”, explains Kátia Emídio.
According to the researcher, the project is in its final phase. “We have already carried out a physical inventory of the area, which was later on validated with the technology developed by Embrapa Acre. The use of artificial intelligence was extremely positive in identifying the Brazil nut trees,” says the forestry engineer.
BIOECONOMY
Kátia defends that, the facilities brought by the two methodologies have the potential to significantly boost the local bioeconomy. "These are technologies capable of engaging younger generations, who are increasingly less involved in the extractive practices of their parents. There are families that have been collecting nuts for generations, and this can contribute to the preservation of this activity, culturally speaking, but also from social and economic standpoints, by generating additional income and encouraging greater participation in the harvesting process. In this way, Brazil nut extraction in the region may be sustained over the long term," she highlights.
The researcher adds that the inventory not only allows you to plan the collection, but also to give an idea of the stock of Brazil Nuts. "It enables us to determine the quantity of this resource and assess how many identified trees are capable of producing. It allows for the quantification of forest resources, not only Brazil nut trees, but also other species of economic interest as well," she explains. In the Uatumã Sustainable Development Reserve, for example, other trees of commercial interest were located, such as breu branco [Protium heptaphyllu], baru [Dipteryx alata] and copaíba [Copaifera langsdorffii].
Forest mapping process is divided into two stages
Evandro Orfanó, the lead researcher of the project that employs artificial intelligence to conduct forest inventories, explains that the activity involves two main stages. “The first step is obtaining what are known as orthophotos, captured by drones that fly over the forest in an automated manner. Once the orthophotos are collected, these drone images are processed using artificial intelligence,” he explains.
The researcher explains the orthophotos make it possible to obtain measurements similar to those from satellite images. “However, while satellite images typically have a resolution of 30 meters, drone images offer a resolution of just 4 centimeters. With this level of detail, artificial intelligence can be trained to recognize patterns in tree canopies and, thereby, identify each species,” Orfanó remarks.
TECHNOLOGY
The tool represents the evolution of a previous technology known as ModeFlora, originally developed in 2007 by Embrapa Acre. “ModeFlora is currently used in the forest planning of virtually all management plans in Acre, Rondônia, the southern region of Amazonas, as well as in other national forests. We have continued to develop this technology, now incorporating artificial intelligence,” says Orfanó.
Artificial intelligence also offers the advantage of expanding knowledge about the biodiversity of the Amazon rainforest itself. As the database is enriched with images and the AI model is further trained, the algorithm becomes increasingly capable of accurately recognizing not only well-known species, but also many others, including various types of palm trees and even dead trees. “AI detects specific patterns, a feature that facilitates the identification of new species and the continuous enhancement of the Netflora database,” adds Orfanó.
TRAINING
The tool is publicly available and can be used upon completion of an online course offered by Embrapa. “We provide remote training and offer support with all flight parameters and the necessary technical guidance,” explains Orfanó. All relevant information can be found on the Netflora project website.
INSTITUTIONAL PARTNERSHIP
The production of Liberal Amazon is one of the initiatives of the Technical Cooperation Agreement between the Liberal Group and the Federal University of Pará. The articles involving research from UFPA are revised by professionals from the academy. The translation of the content is also provided by the agreement, through the research project ET-Multi: Translation Studies: multifaces and multisemiotics.
