The growth of the forestry sector and the intensive use of wood makes short, medium and long-term planning vitally important and requires that they be increasingly optimized to guarantee the flow of wood over time. That said, large-scale monitoring of forest resources is necessary and requires the employment and integration of remote sensing and forest sampling techniques in the field.

The sensor LiDAR is a cutting-edge technology that can be used in different products, such as: digital terrain models and digital surface models. It is normally used for topographic surveys in order to characterize structure and is increasingly applied to obtain volume in the Forest Inventory, both for native forests and for planted forests. The technology is characterized as an active remote sensing system.

Its operation involves the use of laser pulses directed at the ground and, through the time taken to return these pulses, an estimate is obtained, in this specific case, of the height of the forest. In general, this time will be processed to calculate variable distances between objects, or the earth's surface, making it possible to determine shapes, heights, among others, as well as revealing intricate details of the forest structure that previously remained hidden in traditional inventory methods.

This wealth of data not only improves the planning and execution of sustainable management practices, but also boosts the efficiency of forestry operations that sometimes have a high aggregate cost. Another major difference is that for images derived from passive sensors, the major limitation is that two-dimensional information can be obtained from the forests. In the case of active sensors, where it is included, we have the ability to generate three-dimensional images, making it possible to provide, in addition to area calculations, also the characteristics of the vertical structure of the sampled vegetation.

Compared to traditional methods, LiDAR stands out for its operational efficiency, its ability to cover large areas quickly and provide more comprehensive and accurate data. This advance represents a paradigmatic change, making it possible to move from sampling estimates to quasi-census assessments of forests, with the identification of intra-plot variations that were previously inaccessible.

Bracell has been using this technology embedded in drones to obtain the Pre-cut Inventory, improving the estimate of the volume that will be harvested and subsequently delivered to the factory. The idea of using this technology comes from the high sampling intensity, since when going through the continuous inventory of one plot every 10 hectares, the Pre-cut inventory requires one plot every 3 hectares, tripling the costs of data collection.

The adoption of the use of drones equipped with these sensors has been a differentiator, as it provides greater flexibility and extremely high data resolution, which not only speeds up data collection on extensive surfaces but also enables the generation of valuable by-products for the most various forest areas, such as productivity maps and detailed tree canopy models that can even be used to map gaps and planting gaps.

With this, we stopped using average values, such as individual volume, and started using strata within the plot, improving not only the accuracy of wood volume information, but we began to have a new harvest strategy, improving productivity planning by volume stratum. However, despite its indisputable effectiveness, LiDAR does not eliminate the need for traditional sampling methods; the calibration of mathematical models, often based on machine learning techniques, requires precise field data to guarantee the accuracy of the generated estimates.

Thus, complementarity between methods is essential for the continuous improvement of the models used in the forest inventory. Furthermore, it is worth noting that, despite the benefits, point clouds end up resulting in heavy and time-consuming processing that requires certain technological investments. The pilot project began in 2022 in partnership with APRIL, a cellulose and paper company from the RGE Group of which Bracell is also part. Through the technical knowledge of our “brothers” from Indonesia, we started the project with some models they applied to the eucalyptus forest that were later adapted to our forest profile, reducing testing time and improving the use of technological resources.

To ensure the success of the tool, 3 main steps must be mastered:
a) The Flight: in the pilot project, flights were carried out with LiDAR equipment attached to an agricultural plane and drones. Bracell chose to use LiDAR attached to a drone for two main reasons: First, for the autonomy in the flight plan, since we have our own team to fly the drones; Second, and most importantly, due to the quality of the point cloud that allows us to extract more precise metrics for the volumetric model, we rely on Bracell's Topography team to carry out drone flights;
b) Processing: it was necessary to invest in hardware and storage to receive a huge amount of data to process, in addition to the use of Artificial Intelligence to speed up processing. This stage falls to the Remote Sensing team, which has all the expertise to handle the data clouds and extract the metrics, and
c) Mathematical modeling: with regard to modeling, another major difference is that we move away from the strictly dendrometric approach of equations with two or three variables to robust and sensitive models, in which a relatively high number of variables are used, such as example: height percentiles, crown density, individual counts, indices calculated by RGB, etc.

Finally, this last step is the responsibility of the Forest Inventory team, which carries out the modeling and cross-referencing of data with the continuous inventory to make available the volume by plot and the maps with the productivity levels. Thus, LiDAR, at Bracell, passes through several areas until the product reaches the Microplanning and Forestry Operations teams, meeting the company's values of teams that complement each other and the client. In addition to wood volume and productivity maps, LiDAR technology has been used in other products at Bracell, such as:

• Slope maps for cutting and shifting operations;
• Soil surface maps for water analysis and recommendation of contour lines and terraces for soil conservation, and
• Mapping of the electrical network and other improvements, among others.

In summary, LiDAR has established itself as a transformative tool in the Forest Inventory, contributing significantly to the precision, efficiency and sustainability of the management of planted forests in Brazil. The integration of this technology with traditional practices and overcoming its operational challenges point to an even more promising future in the Brazilian forestry sector.