Vegetation phenology serves as a primary ecological indicator of climate change and has numerous ecosystem and climate impacts including nutrient cycling, energy budgets, and annual primary productivity. Phenology models, especially ones of autumnal processes like senescence, are typically based on correlations between environmental threshold triggers and transition dates and less is known about the specific mechanisms behind phenological events. Higher temporal resolution satellite data is needed to continue to identify the mechanisms at larger scales. It is unclear if a start of senescence (SOS) trigger is needed in mechanistic models and if decreased photosynthesis drives senescence. In this dissertation, I have two main themes: the first (Chapters 2 and 3) is to investigate the potential of the Geostationary Operational Environmental Satellite (GOES) to track changes to the phenology-sensitive Normalized Difference Vegetation Index (NDVI) and the second (Chapters 4 and 5) is to develop dynamic mechanistic models to predict senescence in cold-deciduous broadleaf forests.
In Chapter 2, I created a novel statistical model to estimate daily NDVI with uncertainty from high temporal resolution (five - ten minutes) GOES-16 and -17 data. In Chapter 3, I used this data to track forest phenology by fitting double-logistic Bayesian models and comparing transition dates to those obtained from PhenoCams (digital cameras) and the Moderate Resolution Imaging Spectroradiometer (MODIS). Compared to MODIS, GOES was more correlated with PhenoCam at the start and middle of spring.
In Chapter 4, I developed a dynamic Bayesian model based on the physiological process of chlorophyll cycling that assumes a constant chlorophyll breakdown rate and synthesis dependent on temperature and photoperiod to predict senescence without including a SOS trigger or degree-day memory. I fit the model to greenness time series from 24 PhenoCam sites and found that for 49% of the site-years the model could predict SOS using only pre-SOS data. Furthermore, the model could regularly predict greenness at other sites better than their climatologies.
In Chapter 5, I investigated if including photosynthetic feedbacks could improve the chlorophyll synthesis model at the canopy and leaf-levels. Testing this against leaf-level measurements of photosynthetic capacity and changes in chlorophyll concentrations of Fagus grandifolia and Quercus rubra demonstrated that the model fit improved at the canopy level, but not at the leaf-level. This dissertation illustrates that GOES can track phenology and that senescence in cold-deciduous broadleaf forests might not be initiated with a threshold-based trigger.
| Identifer | oai:union.ndltd.org:bu.edu/oai:open.bu.edu:2144/48019 |
| Date | 07 February 2024 |
| Creators | Wheeler, Kathryn I. |
| Contributors | Dietze, Michael |
| Source Sets | Boston University |
| Language | en_US |
| Detected Language | English |
| Type | Thesis/Dissertation |
| Rights | Attribution 4.0 International, http://creativecommons.org/licenses/by/4.0/ |
Page generated in 0.0026 seconds