The growing demand for renewable energy is increasing more and more. In this context, sugarcane is an important option to meet the need for ethanol and bioenergy, with a direct impact on sustainability. According to the National Supply Company, for the 2021-2022 harvest, the production estimate is almost 585 million tons of sugarcane.

According to NASA, Brazil has 400 million arable hectares, of which 8 million are cultivated with sugarcane. In 1950, the pedological study of soils was started in Brazil, even today we are only 60 pedologists, completely insufficient for the current demands and that of the near future.

In Brazil, there are 13 soil types at the most generic classification level and more than 100 subtypes when all pedological details are considered. In order to know the productive potential of sugarcane, it is essential to consider the climate, the type of soil and the level of management, which represents production environments. The most common soils are Latosols and Ultisols. Only the Argisols have a great difference in clay values between the arable layer and below it, giving a greater water storage capacity, which varies in the depth where they occur in the soil profile.

The effect of this difference can be clearly observed in the best sprouting of the plant in the Ultisol and in the worst sprouting in the Latosol, under the same management level, as shown in Figure 1.



In pedology, several tools are used to automate production processes and contribute to a faster response time for results and decision-making. Remote sensing is one of them and, through satellite images, it has been bringing efficiency to pedology and production environments, carried out by field teams. In this method, the analysis of the soil is done by reflectance (a form of energy reflected from the soil and obtained in the form of electromagnetic radiation), by sensors present in satellites, as long as the soil is uncovered.

Due to the different components of each soil, we can obtain different wavelength responses, leading to some interesting correlations, in the topsoil layer, as seen in Figure 2.



Soil elevation models allow assessing the topography of an area and making good correlations with possible variations in soil types, helping the field team to collect data. Several works have been addressing the importance of uniting several “layers” of information, such as satellite images and elevation models, in time series, and performing machine learning with the objective of obtaining a faster and more accurate soil classification, without the need to go to the field. Interesting results are showing some correlations that are already very useful for pedology and, consequently, for production environments.

However, it is still necessary in-depth studies and more publications in the area, so that they allow us to establish methodologies with greater accuracy, especially considering the existing soils of the arable layer, as well as the one just below it. Many soils have a strong correlation with some topography models, but this does not necessarily allow identifying the soil it really is, nor does changes in its color necessarily mean soil change.

Soil classification work in the field is always necessary, making it possible to produce faithful pedological maps of the areas, with publication scales compatible with semi-detailed or detailed levels , feeding databases that can make artificial intelligence a reality. In the form of a ruler, the different sugarcane production environments are presented, in which the respective average yields of five cuts (TCH5) are indicated, figure 3.