Combining hand-held and drone-based lidar for forest carbon monitoring: insights from a Mediterranean mixed forest in central Portugal

Abstract The adoption of novel methods in forest management planning requires the incorporation of precise forest and tree data to improve scheduling and meet multi-objective criteria principles. This study evaluates advanced methods for mapping tree structural attributes to create detailed baselines for forest carbon biomass, a key indicator in environmental policies. We specifically investigate the combined use of mobile sensors (hand-held laser scanning, HLS) and airborne (unmanned laser scanning, ULS), to estimate biomass and carbon stocks in a Mediterranean mixed forest. The novelty of our study lies in the synergistic application of HLS and ULS technologies and the evaluation of different ULS flight altitudes (50, 70, 90, 110 m) and scanning modes to optimize data accuracy and coverage. The main questions addressed are: (1) How do different flight altitudes and scanning modes of ULS affect the accuracy of biomass and carbon stock estimations? (2) What is the impact of merging HLS and ULS data on the precision of tree structural attribute measurements? (3) Can the combined use of HLS and ULS overcome the limitations of individual systems, particularly in complex forest structures? Our case study is conducted in a 1-ha plot in a complex, terraced forest region in Central Portugal, chosen for its high species diversity and structural complexity, which present significant challenges for remote sensing technologies. This site represents a typical Mediterranean mixed forest, allowing us to test methods in conditions that are both typical and challenging for forest monitoring. The distribution of HLS estimates was aligned with reference DBH measurement, though systematically lower (~ 2–3 cm bias). The impact of these measurement errors on total biomass estimation was around 13%. In contrast, major discrepancies were observed in tree height estimations when comparing HLS, ULS, fused ULS-HLS point clouds, with field reference data. ULS operated effectively at heights up to 110 m, increasing coverage without compromising result quality. However, merging point cloud datasets did not significantly improve the accuracy of tree height estimates due to the complexity and high species mingling of the forest stand. We recommend caution in using field measurements for validating tree height estimates with laser sensors under these conditions.