characterizing responses of land surface phenology to urbanization, climate change, and extreme weather events using remote sensing and Bayesian models

Land surface phenology (LSP) is the intra-annual rhythm of vegetation dynamic from dormancy to activity and back to dormancy over the landscape. Shifts of LSP have cascading effects on food production, carbon sequestration, water consumption, biodiversity, and public health. There are three major knowledge gaps in understanding the impacts of urbanization, climate change, and extreme weather events on LSP. (1) Previous studies mainly focused on investigating the effects of urbanization on the spatial patterns of LSP by comparing the phenological metrics between urban center and the surrounding rural regions. However, it remains unclear how urbanization-induced land cover conversions and climate change jointly influence the temporal variations of LSP. (2) Conventional methods usually model key phenological transition dates (e.g. discrete timing of spring bud-break and fall senescence) based on aggregated climate variables (e.g. mean temperature, growing-degree days), ignoring the fact that LSP is a dynamic and continuous process which responds to daily weather conditions continuously. (3) Current projection of LSP shifts under future environmental changes relies heavily on species-level observations and degree-day models. It is challenging to produce a set of future LSP metrics in a temporally consistent manner at regional-to-continental scales. To fill these knowledge gaps, this three-part dissertation built a data-model synthesis framework by integrating remotely sensed data and climate data into a state-of-the-art Bayesian model. First, I established a framework to separate the temporal shifts of phenology driven by climate change from those caused by urbanization. Second, I developed and evaluated a Bayesian Hierarchical Space-Time model for Land Surface Phenology (BHST-LSP) to synthesize remotely sensed vegetation greenness with climate covariates at a daily scale from 1981 to 2014 across the entire conterminous United States. Finally, I used the BHST-LSP to project vegetation phenology from 2020 to 2099 under two climate change scenarios. This dissertation contributed in-depth understanding of the LSP and its environmental cues in both natural and human dominated ecosystems. Results from this dissertation provide strong evidences for adoption of climate change mitigation policies and immediate management measures to prevent severe adverse impacts of global warming and urbanization from disrupting vital ecosystem services and functions.