Remotely piloted aircraft system (RPAS) platforms are able to optimize the process of acquiring aerial images and improve the quality of the products generated in terms of spatial and temporal resolution. The exponential advance of the use of RPAS platforms in forestry, especially from the year 2010, is noteworthy. In this review, we present the global state of the art of the development and applications of RPAS technology in forestry, structured from a systematic review. Our results reveal a trend towards the use of multirotor RPAS platforms compared with fixed-wing platforms and that sensors that register in the visible spectral range are still the most widely used. More recent research has shown applications geared especially for areas such as forest inventory, with many innovations based on the detection of individual trees. Special focus has also been given to new alternatives for pest and disease mapping and phenological phenomena that occur at short intervals, as well as the monitoring of fires and postharvest areas. Therefore, there is a great potential for the use of RPAS platforms in a wide range of forest applications, whether linked to the productive sector or to the conservation of biodiversity, with great advances for spatiotemporal forest monitoring and expectations of further progress for the coming years.

<!-- *** Custom HTML *** --> LiDAR remote sensing data combined with machine learning (ML) techniques have presented great potential for large-scale modeling of tropical forest attributes. However, the large amount of information that can be derived from an aerial LiDAR survey, summed with the intrinsic heterogeneity of tropical environments (e.g., the Amazon), makes it a challenge to accurately estimate forest biophysical variables. The aim of our work is to investigate the potential and accuracy of different ML techniques and a generalized linear model (GLM) to learn the relationships between LiDAR-derived metrics and forest inventory data for aboveground biomass (AGB) prediction in Amazon forest sites under selective logging regimes. The predictive performance of three ML techniques, namely random forest (RF), support vector machine (SVM), and artificial neural network (ANN), was compared against result from the GLM technique, across 85 sample plots. Interestingly, the GLM retrieved the most accurate estimations of forest AGB (rho Spearman’s coefficient = 0.87), compared with the ML techniques (RF = 0.77, SVM = 0.67, and ANN = 0.50). A number of possible factors affecting such results are listed and discussed in the text, including sample size and number of predictor variables. Continued research is necessary to improve the confidence of AGB estimation, especially over complex forest structures.

The tropical forest is characterized by expressive biomass and stores high amounts of carbon, which is an important variable for climate monitoring. Thus, studies aiming to analyze suitable methods to predict biomass are crucial, especially in the tropics, where dense vegetation makes modeling difficult. Thus, the objective of the present study was to estimate aboveground biomass (AGB) in a tropical forest area with selective logging in the Amazon forest using the Random Forest (RF) machine learning algorithm and LiDAR data. For this, 85 sample units were used at Fazenda Cauaxi, in the municipality of Paragominas, Pará State. LiDAR data were collected in 2014 and made available by the Sustainable Landscapes Project. The software R was used for data analysis. Among the LiDAR metrics, the average height was used as it had the greatest significance to compose the model. The model presented a pseudo R² of 0.69 (value obtained by the RF), Spearman's Correlation Coefficient of 0.80, RMSE of 47.05 Mg.ha-1 (19.84%), and Bias of 2.06 Mg.ha-1 (0.87%). With the results, it was possible to infer that the average height metric was enough to estimate AGB in a tropical forest with selective logging, in addition, the RF algorithm the biomass to be estimated, which can be used to assist in monitoring and action management in areas of selective logging and serve as a basis for climate change mitigation policies.

Developing an efficient method to accurately estimate aboveground biomass in tropical forests is critical to monitoring the carbon stock and implementing policies to reduce emissions caused by deforestation. Thus, the objective of the present study was to estimate aboveground biomass in areas of the Amazonian Forest with selective logging, using the random forest (RF), support vector machine (SVM), and artificial neural network (ANN) machine learning algorithms, using light detection and ranging (LiDAR) data and these combined with OLI/Landsat 8 variables, as well as mapping the biomass for the years 2014 and 2017, allowing one to analyze its dynamics between the years of analysis. The RF and SVM algorithms obtained the lowest error values in all datasets. The association of the variables increased the RF performance. Analyzing the dynamics of biomass, it was observed that the oldest exploration units (2006, 2007, and 2008) have lower biomass stocks. The highest biomass losses in 2017 came from units operated between 2012 and 2013 (the most recent record). Thus, with the method used in this study, it was possible to infer that the machine learning algorithms were efficient in estimating the biomass, emphasizing the RF and the SVM.

Optimization in the process of managing forest resources seeks alternatives that make data collection possible. One of them alternatives is spectroradiometry, which consists of measuring the spectral response, having as product the response of the target in relation to the incident radiation along the electromagnetic spectrum, and that, using machine learning, with pre-selected models, makes it possible to identify. Given the above, the study aimed to use machine learning algorithms to classify species by vegetation indices from reflectance data. The study was developed at the Federal University from Santa Maria, working with the species Ficus benjamina, Inga marginata, Handroanthus chrysotrichus, Psidium cattleianum, Salix humboldtiana, Corymbia citriodora and Myrcianthes pungens, and spectral readings of the leaves were taken using the FieldSpec®3 spectroradiometer connected to RTS-3ZC3 integrating sphere. The reflectance values with wavelength ranged in amplitude from 350 ƞm to 2,500 ƞm and spectral resolution of 1 ƞm. Vegetation indices were calculated using the software R Studio, being: NDVI, SAVI, RVI, GNDVI, NDWI, NDWI2, GEMI, DVI, TVI, RVI, MSAVI, WDVI. The algorithms used to develop machine learning were: Random Forest (RF), k-Nearest Neighbors (K-NN), Naive Bayes (NB) and Support Vector Machine (SVM). RF proves to be the most appropriate for data validation, with 85% global accuracy, followed by SVM, with 71%, K-NN with 64% and NB with 35%. The indices with the best performance to point the species were NDWI and SAVI.