A multitemporal 30m land cover map for the ABoVE domain using dense Landsat time series and high resolution imagery.

Jonathan Wang, Boston University, jonwang@bu.edu (Presenter)
Damien Sulla-Menashe, Boston University, dsm@bu.edu
Curtis Woodcock, Boston University, curtis@bu.edu
Oliver Sonnentag, University of Montreal, oliver.sonnentag@umontreal.ca
Mark Friedl, Boston University, friedl@bu.edu

The rapid climate change in arctic and boreal ecosystems is resulting in drastic changes to land cover composition, including woody expansion in the arctic tundra, successional shifts following boreal fires, and permafrost thaw-induced wetland expansion. The impacts on physical climate and the carbon cycle of these land cover transformations are well-documented in field and modeling studies, but there have been few attempts to estimate overall rates of land cover on decadal and continental scales. Previous studies were either too coarse in spatial resolution or too limited in temporal range to analyze relevant rates of change. As part of NASA’s Arctic Boreal Vulnerability Experiment (ABoVE), we employ dense time series of Landsat remote sensing data to map disturbances and classify land cover types across the ABoVE domain, spanning western Canada and Alaska, over the last three decades (1982-2014) at 30m resolution. We utilize regionally-complete and repeated acquisition high-resolution (<2 m) remote sensing imagery to build a database of training data that samples from geographically diverse regions and follows a nested, hierarchical classification scheme encompassing vegetation form, wetland status, vegetation cover density, and land use. We use the Continuous Change Detection and Classification (CCDC) algorithm to estimate change dates and stable temporal-spectral features in the Landsat data for training and prediction using random forest classifiers. We predict land cover types in order to estimate the distribution of land cover and land cover change rates, focusing primarily on woody expansion in the tundra, post-fire succession, and wetland expansion in the boreal forest. The hierarchical classification scheme provides flexibility in assessing different relevant plant functional types in different regions of the arctic and boreal ecosystems. We resample from the high resolution imagery to assess the accuracy of the land cover change maps.

Presentation: ASTM4_Poster_Wang_86_17.pdf (5654k)

Associated Project(s): 

• Woodcock-02

Poster Location ID: 86

Session Assigned: Vegetation Dynamics and Distribution