Big data, Internet of Things and Artificial Intelligence tools are changing the face of industry as we know it – and the forestry sector is no exception. Far from mere gadgets. These technologies are definitely here to stay.

When we try to glimpse the future of forestry, we can certainly imagine that major technological advancements will keep driving growth and progress in the national and global markets. We think of highly technological cultivated forests, with topography and terrain characteristics fully georeferenced and mapped by tools such as UAVs carrying multispectral cameras. We think of advanced management soft- ware and platforms integrated to onboard technologies in forest machinery and equipment.

In this journey towards the future, Big Data, Analytics, Machine Learning, IoT (Internet of Things) and AI are buzzwords that will become more and more prevalent in forestry – and they are already a reality in Brazil’s largest forest companies.

“A forest’s productivity varies according to a great and complex number of combinations rang- ing from climate variables (rain, humidity, light conditions etc.), nourishment, management (activities carried out in the right time and quality level), genetic material, forest threats (fires, pests etc.) and much more. Effectively dissecting the relations between each variable with the goal and impact of each scenario combination over years and years of historical data is only possible with the use of Big Data,” explains Carlos Albuquerque, Inflor’s innovations director.

This, in a nutshell, is the economic potential of Big Data, defined as a great set of stored data, and based on 5 Vs: Velocity, Volume, Variety, Veracity and Value. IoT, on the other hand, is the interconnectivity of onboard technologies, the near omnipresence of web connectivity in every area, tool, equipment and machine, which would be able to exchange information and effectively learn from each other.

With this gigantic database transformed in information available, AI is inserted to help in decision making in such steps as: implementation or regrowth in an area according to its specific his- tory and its consequence in productivity/cost over the next period; the execution (or lack thereof) of handling activities for lowering risks; corrective actions in the case of climate variations or forest threats; anticipating or postponing harvesting; and much more.

According to Albuquerque, these are examples related only to forest productivity, but they may be broadened for different areas of forest production, such as harvesting, by defining the best harvest system and the idea  types of machinery according to productivity history and specific forest characteristics such as steepness, average individual volume, type of production and other commercial decisions such as which products must be sold and processed according to the history of sales vs. profitability.

“In a short time, with the evolution of ICT and IoT technologies, we’ll have thousands of sensors distributed in the field for monitoring our forests in real time, that is, greater availability/ quality of information and, con- sequently, ever more effective algorithms for helping the forest production chain in its entirety,” concludes the professional.

The Fibria case

The trend is clear: Brazilian forestry companies are already aware of the great potential of smart tools and are investing heavily in research, develop- ment and application of new technologies in the field.

At Fibria, AI and Big Data are pillars of Forest Management and Analysis and are tied to machine learning processes and large scale analytics software. The company has an extensive and complex structure for gathered data, stemming from its nursery, silviculture, harvesting and logistics operations and related to operational costs and quality, data volume and forest productivity. Data is also gathered from the environment, such as topography, soil type, climate conditions and interactions, which are use in highly complex analyses with the use of Big Data as a basis for decision making in productive processes.

“We’ve been working with Big Data tools and Artificial Neural Networks since 2015 in  order to generate qualitative and quantitative value in the handling of our forests through the large- scale analysis of data, as well as processing speed, analytical accuracy and variability in data origin,” outlines the manager of geoinformation and forest protection at Fibria, Luis Sabbado.

At first, the company applied Neural Networks to its forest inventory processes with the goal of decreasing the costs of measurement fieldwork. Thus, it was possible to successfully validate the tool, resulting in a 20% decrease in costs for this activity. Later, its use was expanded to basic wood density predictions, a variable of great interest in industrial and forestry processes. In this application, there was a decrease in the time needed to obtain the information, as well as sample collection in the field, aside from the possibility of increasing the number of input variables in the prediction model. Quantitative forest inventory data and environmental indexes were added to the forest’s qualitative information.

The company has also used Big Data tools to create analyses and recommendations for optimizing regrowth performance and in the evaluation of weed prevention figures. In this first application, Fibria managed to create a new scenario for its planning horizon, broadening the regrowth area with more assertive technical criteria and a positive financial return of BRL 11.9 million in NPV in the 2017/2018 period. Among the current challenges faced by Fibria is the structuring and quality improvement of its databases, as well as higher speed in process- ing for greater agility in decision making.

“With the use of new technologies, we have some goals to reach, including: optimizing water and resource consumption; better energy efficiency; optimizing the use of labor; geographic traceability of applied assets and services; increased operational security through behavioral management; ensuring information arrives at the adequate time; increasing productivity; and improving information accuracy,” concludes Sabbado.