PHENOLOGY OF LEAF, FLOWER AND PHOTOSYNTHESIS: DELINEATION AND QUANTIFICATION BY TIME SERIES OBSERVATIONS FROM MULTIPLE SATELLITE-BASED SENSORS

Abstract

Vegetation phenology, including both leaf phenology and floral phenology, is a fundamental component of terrestrial ecosystems and an important indicator of ecosystem responses to environmental change. Seasonal transitions in leaf phenology, such as leaf emergence, canopy development, and senescence, regulate carbon uptake and water exchange at the ecosystem scale. Floral phenology also plays an important role in plant reproduction, species interactions, and crop productivity. Accurate detection of these phenological events is therefore essential for understanding ecosystem functioning under climate variability and global change. Satellite remote sensing has become a major tool for phenology monitoring because it provides repeated observations with broad spatial coverage and high temporal frequency. However, important challenges remain. Most satellite-based phenology studies have focused primarily on leaf phenology, while floral phenology remains much less explored. In addition, many commonly used leaf phenology methods rely on smoothed annual vegetation index curves, which may weaken the direct biological meaning of the derived start of season (SOS) and end of season (EOS) metrics. Phenological interpretation is also scale dependent, yet the relationships among plant phenology, vegetation phenology, and land surface phenology across spatial resolutions remain insufficiently clarified.This dissertation investigates how satellite observations can be used to detect both floral and leaf phenology across spatial scales, while adopting different observational strategies for each. Floral phenology is more difficult to detect because flowering events are often short in duration and highly sensitive to mixed pixel effects, especially at coarser spatial resolutions where flower signals are diluted by surrounding canopy and background components. This challenge becomes more pronounced when the flowering period is shorter than two weeks, in which case near-daily 3 m PlanetScope observations provide strong support for detection. In contrast, leaf phenology persists longer and produces more stable canopy signals, making it less sensitive to mixed pixel effects and more suitable for consistent analysis and scale extension from single-tree-level to ecosystem-level applications. This dissertation addresses three central questions. First, can a Chlorophyll and Green Leaf Indicator (CGLI)-based method, built on the spectral relationship BLUE < GREEN > RED, effectively detect the start and end of the green-leaf season across high-spatial-resolution satellite observations? Second, can daily 3 m PlanetScope observations, together with newly developed spectral indicators, capture complex floral phenology that is not adequately represented by traditional leaf-related vegetation indices? Third, how do the seasonal dynamics of climate variables, surface reflectance, vegetation indices, and gross primary productivity (GPP) relate to one another, and how can GPPEC-based SOS and EOS be used to characterize photosynthetic phenology? The results show that the CGLI-based method can effectively identify green leaf emergence and senescence at the single-tree level using 3 m, 10 m, and 30 m satellite observations, providing a physiologically grounded alternative to conventional vegetation index-based phenology methods. For floral phenology, daily 3 m PlanetScope observations captured the spring phenology of Callery pear trees that bloom white flowers before leaf emergence, and a White Flower Index (WFI) was developed to describe the dynamics of light-colored flowers. PlanetScope observations also supported the analysis of yellow-flower phenology in canola using a yellow-band-based spectral indicator, demonstrating the value of near-daily high-resolution imagery for detecting short-duration flowering events. In contrast, this dissertation uses leaf phenology as the focus for scale extension. To address the phenology of photosynthesis, it examines the seasonal dynamics of ecosystem function through GPP and derives GPPEC-based SOS and EOS as indicators of photosynthetic phenology. The dissertation also examines seasonal dynamics of surface reflectance and vegetation indices across spatial resolutions from 3 m to 500 m and across multiple vegetation types, showing that different vegetation indices provide complementary information under varying canopy and growth conditions. Overall, this dissertation advances satellite phenology research by showing that floral phenology and leaf phenology require different remote sensing strategies across scales. Short-duration flowering events are best investigated with near-daily 3 m observations, while leaf phenology provides a more suitable basis for physiologically grounded method development and scale extension to ecosystem-level applications. By linking leaf phenology, floral phenology, and photosynthetic phenology, this dissertation provides an integrated framework for understanding vegetation dynamics across structural and functional dimensions.