Improving accuracy and precision of TLS-derived gap fraction: Comprehensive evaluation of acquisition and synthesis parameters

Gap fraction serves as a fundamental variable for quantifying forest light availability and deriving key canopy structural metrics. Hemispherical photography has been the standard method for gap fraction estimation, but has limited capacity to quantify three-dimensional stand architecture and light gradients. The advance of terrestrial laser scanning (TLS) overcomes these limitations by capturing complete 3D structural information. However, TLS-derived gap fraction is sensitive to acquisition and synthesis parameters, and wind-induced canopy movement during scanning may further compromise the reliability of these estimates. We evaluated how these parameters influence the accuracy and precision of TLS-derived gap fraction across a canopy density gradient. We generated 19,968 synthetic hemispherical images from eight TLS configurations, including angular resolution (ar12: 175 million points; ar14: 44 million points), scanning mode (single vs. multi-scan), scan quality (q1–q6, corresponding to 3–30 minutes per station), beam divergence (0.01–0.21°), and image resolution (2000–20000 pixels). We compared TLS estimates against reference hemispherical photography to quantify bias (accuracy) and variance (precision). Point cloud density of single-scan largely determined multi-scan effectiveness, with ar12 achieving significantly higher accuracy than ar14 (error: 0.12 vs. 0.20). Beam divergence dominated estimation accuracy, with 0.09° as optimal. High scan quality eliminated precision benefits due to wind-induced canopy movement. Multi-scan under moderate quality settings showed the strongest variance reductions, especially at high zenith angles. Our findings demonstrate that adequate single-scan point density amplifies multi-scan benefits rather than simply relying on multiple scan positions to compensate for sparse sampling. We recommend prioritizing angular resolution over scan duration and applying moderate beam divergence to optimize estimation accuracy, precision, and field efficiency. These results provide practical guidance for TLS data acquisition and highlight the necessity of parameter configuration.