Optimizing forest canopy structure retrieval from smartphone-based hemispherical photography

Abstract

Hemispherical photography (HP) is a long-standing tool for forest canopy characterization. Currently, there are low-cost fisheye lenses to convert smartphones into highly portable HP equipment (smartphone-based HP, hereafter SHP). However, there is an obstacle to having a close-to-ideal method for citizen science and large-scale or opportunistic sampling: the known sensitivity of HP to illumination conditions. The purpose of this paper is to test a ready-to-use approach based on previous research and to contribute to quantifying the errors associated with choosing SHP in non-recommended light conditions over well-established HP practices. In 30 locations distributed in broadleaf and coniferous woodlands, a total of 1080 photographs were taken with two smartphone models, manipulating the exposure, and under varied sky conditions. After image binarization, accurate reference data was employed to evaluate the reliability of extracting canopy parameters from SHP. The proposed methodology can reliably quantify canopy openness (RMSE ~0.04) and plant area index (RMSE ~0.4). Results suggest that SHP, when used following the recommendations of the present study, allows retrieval of reliable canopy metrics independently of sky conditions and forest type.