SOIL MOISTURE ESTIMATION THROUGH MACHINE LEARNING USING UNMANNED AERIAL VEHICLE (UAV) IMAGES

• Autor(a) principal: Safre, Anderson Luiz dos Santos [UNESP]
• Data de Publicação: 2021
• Outros Autores: Fernandes, Caio Nascimento [UNESP], Saad, João Carlos Cury [UNESP]
• Tipo de documento: Artigo
• Idioma: por
• Título da fonte: Repositório Institucional da UNESP
• Texto Completo: http://dx.doi.org/10.15809/irriga.2021v26n3p684-700; http://hdl.handle.net/11449/240944
• topic: artificial neural networks; machine learning; soil moisture; UAV

The soil moisture is an important parameter for the calculation of water depth and irrigation management since it is directly related to the soil water content. Remote sensing techniques combined with statistical models can be used to estimate the spatial variability of soil moisture, extrapolating point measurements. The objective of this study was to determine the soil moisture through machine learning algorithms such as Support Vector Regression (SVR), Random Forests (RF), and Artificial Neural Networks (ANN). High resolution multispectral images obtained by an Unmanned Aerial Vehicle (UAV) in an irrigated bean area at the Experimental Lageado Farm at Unesp in Botucatu, SP, Brazil, were used. The reflectances in the Green, Red and Near Infrared bands along with the NDVI vegetation index were used as inputs for the models. All the algorithms performed well; however, the model that best fitted the data was the SVR, with mean square error (RMSE) of 0.46% of the estimated soil moisture and determination coefficient (R²) of 0.71.