Using Time Series of Landsat Data to Improve Understanding of Short- and Long-Term Changes to Vegetation Phenology in Response to Climate Change

Thursday, 18 December 2014: 10:38 AM
Mark A Friedl, Eli K Melaas, Damien J Sulla-menashe and Josh M Gray, Boston University, Boston, MA, United States
Phenology, the seasonal progression of organisms through stages of dormancy, active growth, and senescence is a key regulator of ecosystem processes and is widely used as an indicator of vegetation responses to climate change. This is especially true in temperate forests, where seasonal dynamics in canopy development and senescence are tightly coupled to the climate system. Despite this, understanding of climate-phenology interactions is incomplete. A key impediment to improving this understanding is that available datasets are geographically sparse, and in most cases include relatively short time series. Remote sensing has been widely promoted as a useful tool for studies of large-scale phenology, but long-term studies from remote sensing have been limited to AVHRR data, which suffers from limitations related to its coarse spatial resolution and uncertainties in atmospheric corrections and radiometric adjustments that are used to create AVHRR time series. In this study, we used 30 years of Landsat data to quantify the nature and magnitude of long-term trends and short-term variability in the timing of spring leaf emergence and fall senescence. Our analysis focuses on temperate forest locations in the Northeastern United States that are co-located with surface meteorological observations, where we have estimated the timing of leaf emergence and leaf senescence at annual time steps using atmospherically corrected surface reflectances from Landsat TM and ETM+ imagery. Comparison of results from Landsat against ground observations demonstrates that phenological events can be reliably estimated from Landsat time series. More importantly, results from this analysis suggest two main conclusions related to the nature of climate change impacts on temperate forest phenology. First, there is clear evidence of trends towards longer growing seasons in the Landsat record. Second, interannual variability is large, with average year-to-year variability exceeding the magnitude of total changes to the growing season that have occurred over the last three decades. Based on these results we suggest that year-to-year variability in phenology, rather than long-term trends, provides the best basis for predicting future changes in temperate forest phenology in response to climate change.