Monitoring the Spatio-Temporal Variations of C3/C4 Grass Species Using Multispectral Satellite Data

摘要

Grass that follows a C3 and C4 photosynthetic pathway represents a fundamental component of grass species functional type, with outstanding services in grassland ecosystems. These grasses differ in morphology, phenology and physiological features, over time, due to varying environmental requirements. These features determine the success of their discrimination using remotely sensed data and their response in Aboveground Biomass (AGB) over time. For decades, the lack of appropriate remote sensing data sources compromised C3 and C4 grasses monitoring over space and time. This has resuited in uncertainties in understanding their potential and contribution to the provision of services. The aim of this study was to determine the optimal period to discriminate C3 and C4 grass species. The study additionally estimated species AGB over space and time. This was achieved, using Sentinel 2 satellite data with Discriminant Analysis (DA) and sparse partial least squares regression (SPLSR) algorithms. The winter peak was shown to present the best temporal window for discriminating C3 and C4 grasses. This period was also associated with higher species AGB (±1.11kg/m^{2) for both species as they have reached their peak. Although summer period was associated with reasonably high classification accuracies, highest errors (±20%) were encountered and this period had higher AGB, as both species were in their early stages of growth. The discrimination of C3 and C4 and AGB variations were significantly (α = 0.05) contributed by red edge, NIR and SWIR portions of the electromagnetic spectrum. These variables managed to capture species phenological, physiological and morphological contrasts as well as spatial variations over time.}