Publications

Hu, KT; Huang, CY (2019). A metabolic scaling theory-driven remote sensing approach to map spatiotemporal dynamics of litterfall in a tropical montane cloud forest. INTERNATIONAL JOURNAL OF APPLIED EARTH OBSERVATION AND GEOINFORMATION, 82, UNSP 101896.

Abstract
Litter production, or litterfall, is a predominant part of the carbon and nutrient cycles of forest ecosystems. As a major and relatively invariable aspect of primary production, litterfall could be a predictor of total plant production. However, the conventional field approach for its estimation is laborious and costly. High temporal resolution optical satellite imagery may strengthen our ability to estimate litterfall over a vast region, but this remote sensing method is limited in some tropical and near-tropical mountainous regions, mainly due to coarse spatial resolution and frequent cloud coverage. The metabolic scaling theory (MST) states that forest productivity (including litterfall) is directly proportional to the mass of the photosynthesis tissue (leaves), and spatially, it may be isometrically scaled using the size of the largest individual in the system. In this study, we investigated the temporal dimension of the MST and hypothesized that the maximum leaf abundance over a year could also be a key determinant of annual litter production. The relationship between the leaf mass during the peak growing season and annual litterfall could facilitate large-scale litterfall mapping in mountainous terrains, since vegetation indices such as the Enhanced Vegetation Index (EVI) or the Normalized Difference Vegetation Index (NDVI), commonly utilized as a surrogate for leaf abundance, can be derived from relatively high spatial resolution satellite imagery. We correlated the summer growing season Landsat Enhanced Thematic Mapper plus (ETM+) EVI and NDVI for 2016 and 2017 and the temporally corresponding annual accumulated field litterfall data acquired from hinoki (Chamaecyparis spp.) dominant tropical montane cloud forests in northeastern Taiwan (23.98 N, 120.97 E). We found that the summer growing season EVI and NDVI may be salient variables to explain the spatial variation of annual litter production (r(2) = 0.39-0.67, p <= 0.01). The results have profound implications for applying MST to the regional mapping of ecosystem production across space and time in humid tropical regions, and these implications may facilitate the remote assessment of the terrestrial carbon budget.

DOI:
10.1016/j.jag.2019.06.006

ISSN:
0303-2434