Me chame no WhatsApp Agora!

Daniel Henrique Breda Binoti

Forest Measurement Coordinator at Eldorado


Weed competition mapping technology

Co-authors: Jonas Luís Ortiz and Adalto Moreira Braz, Geoprocessing Specialists at Eldorado

The effective management of forest plantations involves several factors. Some are technical, such as the best time to apply silvicultural treatments; biological, such as the productive potential of a given species or clone in a given soil or type of terrain; others are economic, such as wood prices or production costs; and still others are social, such as laws and taxes that affect marketing and the labor used in various stages of the production process.

In addition to the complexity of the management process of forest plantations, managers are exposed to fierce competition among companies in the sector, which requires quick and effective decision-making when choosing prescriptions and silvicultural treatments. In this scenario, remote sensing has gained significant progress due to the low cost and the increasing availability of new imaging technologies.

Remote sensing consists of a set of techniques aimed at obtaining information about objects, by capturing electromagnetic radiation, so that there is no contact between sensor and target. Since its inception, this geotechnology has spread in numerous activities, given its power to capture images, from small areas to continental dimensions, in multiple scales and different levels of detail.

Currently, forest managers have different platforms for acquiring images from national and international sources, free and commercial, which are complementary in terms of applications. The wide availability of images favors complete coverage of the area of activity, at any time of the year, which has intensified the detailed monitoring of recurrent phenomena on forests, in a historical and even daily context.

The development of alternatives for monitoring young eucalyptus plantations based on remote sensing products and geographic information systems becomes viable when automated by programming languages, and their management is done by structured databases.

The monitoring of young eucalyptus forests can be done using several techniques, however the main one is the result of mapping the estimated level of leaf biomass in the areas of interest. This estimate is identified from the spectral vegetation index of the Normalized Difference Vegetation Index.

The calculation of the normalized difference vegetation index is performed using the ratio between spectral bands that consider the near infrared and visible red spectrum. In general, the higher the value of the Normalized Difference Vegetation Index, the greater the area occupied by photosynthetically active vegetation in a given location. And, the lower the value of the Normalized Difference Vegetation Index, the greater the area occupied by exposed soil in the plot.

After processing the satellite images, the information is recorded in databases, generating the amount of absolute area (hectare) and relative (percentage) of each class of the level of biomass found in each plot and project of the area of interest. Along with the biomass level map, the image map in false color composition must be analyzed, resulting from a mosaic composed of three spectral bands, which contrasts and enhances color tones associated, in most cases, with patterns of land cover present in each biomass level class.

As a rule, areas represented by shades of fluorescent green are related to the occurrence of highly severe infestations by weeds and regrowth between rows under renovation; and closer shades of pink to purple, exposed soil. Subsequently, all these products are analyzed by the operational, support and decision-making teams, requiring field validation of the estimates generated.

The products can be analyzed in two ways: for alterations, which are compared between different image dates, and for differences, in which the variability of the level of biomass in a single date is detected. The biomass variations evidenced in the plots may be due to the presence of weeds (implantation regrowth reform) and/or regrowth between rows under renovation, differences in the growth development of the eucalyptus, disease, pest; sinister; and all of this is indicative of intervention in the forest, helping management and decision-making.

The implemented products systematically guide the areas of silviculture, quality and forest technology in the field, in addition to generating evidence for management reports. Only in the field will the forester be able to identify the reason for these alterations, differences and variations; and if weed and/or regrowth, if this presence is really competing with planting and the need for intervention with mowing operation and/or application of post-emergent.

As a development, the studies for modeling the estimate of the level of leaf biomass, considering weights to factors that interfere in the value of the Normalized Difference Vegetation Index, such as the time of year in which the plot was planted, the time of year of the date of image acquisition, planting ages and management regime, combined with artificial intelligence algorithms, can further improve these products.

In addition, the growing market for remote sensing allows us to “envision” the integration of other sensors for this type of monitoring, such as passive (optical) and active (radar) sensors. Imager radars allow for greater temporal resolution, since, unlike optical sensors, they allow data to be obtained regardless of weather conditions, such as the presence of rain, smoke, cloud, shadow, etc., and have the ability to obtain data regardless of solar energy, that is, obtaining data is possible both during the day and at night.

We believe that the main challenge is to keep up with the rapid development of storage, communication, information and processing technologies, combined with the implementation of remote sensing solutions, with the integration and availability of results through cartographic products, reports, consumer services on WebGIS (local or mobile).