Embrapa researchers are looking for new possibilities for applying geotechnologies to the digital model of forestry exploration in the Amazon. The project “Geotechnologies applied to forest automation and spatialization of carbon stocks in native and modified land use in the Western Amazon (Geoflora)”, carried out in partnership with the JBS Fund, aims to improve technological tools for greater automation of planning and execution of inventories forestry for precision management.
Geoflora, the second stage of research begun in 2015, will be carried out in six states in the Amazon (Acre, Rondônia, Roraima, Amapá, Pará, Amazonas). The activities also involve studies to measure carbon stocks and emissions in different land uses in the Amazon (forest physiognomies, forest landscapes). The creation of a database of images obtained using RGB cameras on board drones (orthophotos) is among the planned actions.
According to researcher Evandro Orfanó, the initial challenge is to structure a solid, consistent database that reflects the reality of what is in the field, based on images generated in drone flights over treetops of different species and botanical identification in field. “Based on this information, we will train algorithms that, in the near future, will function as “electronic woodsmen”, capable of identifying, locating species in the forest and taking canopy measurements that will allow us to estimate the volume of wood, with the possibility of being continuously improved", explains the researcher.
In addition to this advance, Orfanó believes that with the use of the equipment in a fifth generation network, 5G, it will be possible to insert this algorithm into a drone or cell phone to identify species, making it possible to carry out a forest inventory in real time. .
An algorithm is a mathematical model that works as a kind of "recipe" for performing a task and can be applied for different purposes. In the case of the Geoflora project, the algorithms will be trained to differentiate the characteristics of forest species of commercial interest - timber and multiple use - and carry out their identification. The objective is to make these tools experts in carrying out inventories in forest areas with açaí, murmuru, copaíba, castanheira, cumaru-ferro, garapeira, caucho, copaíba and cedar and other Amazonian species.
According to Orfanó, the more data collected, the better the algorithm's performance. “This means flying with a drone to obtain increasingly detailed information about the visual characteristics of species, by region, time of year and fruit production season, as the algorithm needs to know all the visual variations that occur in the trees to carry out work with greater safety and precision", he reinforces.
The Yolo (You Only Look Once) algorithm, developed in 2015 in the United States, will be the basis for creating new algorithms for identifying species. Open source, the technology has an infinite potential for applications, including the military, the automobile industry (smart cars), medicine, social media monitoring, among other possible uses.
Another focus of the Geoflora project is the production of models for estimating carbon stocks and biomass dynamics in the forest. The research will be carried out by combining estimates obtained from permanent plots in the field, through data collected using LiDAR (Light Detection and Ranging), orbital images (generated by satellites) and machine learning algorithms. .
Embrapa researcher Marcus Vinício Neves d’Oliveira explains that the use of this set of technologies will enable data collection in large areas of forest. Currently, only data from sampling units (permanent plots) are used to estimate the behavior of managed forests and define their growth, production potential and fallow time necessary for their recovery. However, due to the high cost of implementation, measurement and demand for highly qualified labor to analyze the collected data, these plots are normally restricted to management plans monitored by educational and research institutions, in public forests.
LiDAR is an active remote sensing system that performs laser scanning of forest areas with high precision. The use of the LiDAR sensor on board drones will allow biomass and carbon estimates to be scaled over large areas and, in some cases, could cover the entire managed forest. This technology, until then, was used attached to piloted aircraft, a factor that made its adoption more expensive and limited. "With LiDAR we will be able to verify the growth dynamics of the forest and more accurately estimate the stocks (volume, biomass and carbon) of the species identified by the trained algorithms. This will enable a forest inventory carried out exclusively by remote sensors, which represents a scientific gain immeasurable", he adds.
“In addition to producing high-resolution biomass maps for areas of strategic interest, such as protected lands, the estimation models used in the project can be applied to forests with similar structures, such as those in Acre and Rondônia and other locations in the Amazon, without the need of permanent plots in the areas. The results of these studies will support public policies for sustainable use (forest management) and payment for environmental services (carbon market)”, highlights d’Oliveira.
According to researcher Falberni Costa, associated with biomass dynamics, soil is another component of the forest system that functions as reservoirs-drains or emitters of carbon (CO2). Stored in organic matter in the soil, part of this carbon can be released by microbial activity, a process considered the second largest source of CO2 emissions into the atmosphere, surpassed only by the burning of fossil fuels. Monitoring and calibrating these emissions are essential for greater reliability of estimates.
“Knowledge of carbon flows (soil stocks and CO2 emissions) and forest biomass stocks (roots, litter and trees) will allow us to identify whether the forest system is sequestering or emitting equivalent carbon into the atmosphere. The project will also measure soil carbon stocks and CO2 emissions, which will allow calculating the balance of the forestry systems assessed”, he highlights.
Brazilian non-profit civil association dedicated to promoting and financing initiatives and projects, the JBS Fund for the Amazon aims to promote the sustainable development of the Amazon Biome, for the conservation and sustainable use of the forest and improving the quality of life of the population that resides there, with use of technology and applied science.
Andrea Azevedo, director of the Fund, explains that due to the extension of the Amazon biome and the potential of the carbon market and other environmental services, it is essential to use technological tools that, in addition to reducing operational costs, allow obtaining data with greater precision in measuring biomass and carbon stored in the forest.
"There is still little guidance on the sustainable exploitation of many forest products and on the location of species. So, when you have an efficient system, it is possible to know, using artificial intelligence, where these resources are. This makes the extractivist's life easier and other actors involved with management”, he emphasizes.
Receive the latest agriculture news by email
Receive the latest agriculture news by email
In 2022, the Postgraduate Program in Agricultural Engineering, at the Federal University of Santa Maria (UFSM), celebrates 50 years of uninterrupted operation
The 3rd Brazilian Symposium on Biological Coffee Management, which takes place in Franca (SP), on the 5th and 6th of October, will feature the participation of the coordinator of the State Program for Research in Agroecology at Epamig, Madelaine Venzon
