Mapping the Leaf Economic Spectrum across West African Tropical Forests Using UAV-Acquired Hyperspectral Imagery
An article of Eleanor Thomson, Yadvinder Malhi, Harm Bartholomeus, Imma Oliveras, Agne Gvozdevaite, Theresa Peprah, Juha Suomalainen, John Quansah, John Seidu, Christian Adonteng, Andrew Abraham, Martin Herold, Stephen Adu-Bredu and Christopher Doughty: Mapping the Leaf Economic Spectrum across West African Tropical Forests Using UAV-Acquired Hyperspectral Imagery, has been published in Remote Sensing, 2018, 10(10), 1532;
In 2016, Harm Bartholomeus and Juha Suomalainen, used the Hyperspectral Mapping System (HYMSY) for a large campaign to map ecosystem carbon cycling and plant traits of the tropical forests in Ghana. This dataset has been one of the main sources for a paper which has been recently published in Remote Sensing entitled: Mapping the Leaf Economic Spectrum across West African Tropical Forests Using UAV-Acquired Hyperspectral Imagery.
The leaf economic spectrum (LES) describes a set of universal trade-offs between leaf mass per area (LMA), leaf nitrogen (N), leaf phosphorus (P) and leaf photosynthesis that influence patterns of primary productivity and nutrient cycling. Many questions regarding vegetation-climate feedbacks can be addressed with a better understanding of LES traits and their controls. Remote sensing offers enormous potential for generating large-scale LES trait data. Yet so far, canopy studies have been limited to imaging spectrometers onboard aircraft, which are rare, expensive to deploy and lack fine-scale resolution. In this study, we measured VNIR (visible-near infrared (400–1050 nm)) reflectance of individual sun and shade leaves in 7 one-ha tropical forest plots located along a 1200–2000 mm precipitation gradient in West Africa. We collected hyperspectral imaging data from 3 of the 7 plots, using an octocopter-based unmanned aerial vehicle (UAV), mounted with a hyperspectral mapping system (450–950 nm, 9 nm FWHM). Using partial least squares regression (PLSR), we found that the spectra of individual sun leaves demonstrated significant (p < 0.01) correlations with LMA and leaf chemical traits: r2 = 0.42 (LMA), r2 = 0.43 (N), r2 = 0.21 (P), r2 = 0.20 (leaf potassium (K)), r2 = 0.23 (leaf calcium (Ca)) and r2 = 0.14 (leaf magnesium (Mg)). Shade leaf spectra displayed stronger relationships with all leaf traits. At the airborne level, four of the six leaf traits demonstrated weak (p < 0.10) correlations with the UAV-collected spectra of 58 tree crowns: r2 = 0.25 (LMA), r2 = 0.22 (N), r2 = 0.22 (P), and r2 = 0.25 (Ca). From the airborne imaging data, we used LMA, N and P values to map the LES across the three plots, revealing precipitation and substrate as co-dominant drivers of trait distributions and relationships. Positive N-P correlations and LMA-P anticorrelations followed typical LES theory, but we found no classic trade-offs between LMA and N. Overall, this study demonstrates the application of UAVs to generating LES information and advancing the study and monitoring tropical forest functional diversity. View Full-Text
Keywords: leaf traits; leaf economic spectrum; UAV; hyperspectral; spectroscopy; tropical forest; PLSR; Ghana; West Africa