Desenvolvimento de versão digital para um instrumento analógico de medição de densidade de copa em áreas florestais

Abstract

The spherical densiometer is an instrument used to measure the density and gap of tree canopies, essential information for areas such as ecology, forestry and forest management. Although practical, the object has some limitations, which justifies the development of a digital version of the spherical densiometer, based on the need to improve the data collection and analysis process, which currently depends on the user's subjective observation and subsequent analysis using external softwares. With the advancement of digital technologies, mobile devices (cell phones and digital cameras) are being evaluated as alternatives or complements to traditional observation made with the naked eye. The development of a digital densiometer aims to integrate these steps and make the process more agile, precise and functional, in addition to enabling use on mobile devices, such as smartphones. The general objective of this study is to create a digital version of the spherical densiometer. The methodology involves programming a digital desktop version in Python (PYTHON 3.13.3) and a digital smartphone version written in Java-compatible Kotlin. A correction factor was developed to match the distortions between conventional camera lenses and fisheye lenses, which are anatomically closer to the curved mirror of the spherical densiometer. Subsequently, validation was performed through comparative and standardized field tests with the traditional analog model. Ten sampling units were measured in three types of forest stands, totaling 30 observations in the study. The canopy cover values (%) measured with different devices (mobile and desktop) were compared against the reference visual method using an analog densiometer. All measurements were compiled in Excel spreadsheets, and normality and similarity tests were applied to evaluate the agreement between methods.. Through a mixed linear model and Dunnett's contrast, it was possible to verify that one of the plots caused significant systematic variation, possibly due to interactions with the measurement context, such as lighting or canopy structure. Upon removing this location from the analysis, it was found that the cell phone cameras (front and rear) tend to underestimate the coverage compared to the reference, while the "desktop" version showed statistically compatible results, suggesting the absence of relevant systematic bias.