Soil Moisture Acquisition Through Multispectral Sensors Coupled to Unmanned Aerial Vehicle (UAV)

Abstract

This study aims to evaluate the use of remote sensing technologies in the determination of soil moisture content in an exposed soil area prepared for irrigated rice cultivation, as the knowledge of this soil property is crucial for control and productivity. These values are usually determined in a conventional way, which requires a large amount of time and allows only a few points of information in the area. With the technological advances in remote sensing, several studies are in development to determine these properties through satellite imagery, which allows mapping in large scale, but with a low spatial resolution. Therefore, the possibility of using sensors coupled in an unmanned aerial vehicle (UAV) presents a lower cost and adds a gain of time in the collection of samples. In order to generate a product representing the soil moisture from the reflectance data, aerial images were obtained using the Sequoia multispectral sensor from Parrot, which was developed for use in precision agriculture, and also were collected 85 soil samples which moisture was determined by the conventional method, serving as the basis for the study. From the samples collected in the same study area, a controlled experiment was reproduced in the laboratory using 16 samples with different moisture contents. Their spectral response was measured with a spectroradiometer to generate mathematical models that could be applied in the data collected by the Sequoia sensor and define the moisture from reflectance data. It was verified that the data from the Sequoia can be influenced by external factors, which may compromise its direct application in the models generated in the controlled experiment. In addition, there is the limitation of spectral bands of the Sequoia when compared to the spectroradiometer.