Interactions among land cover, disturbance, and productivity across Arctic-Boreal ecosystems of Northwestern North America from remote sensing
Wang, Jonathan Alexander
MetadataShow full item record
Arctic and Boreal ecosystems are experiencing accelerated carbon cycling that coincides with trends in the normalized difference vegetation index (NDVI), a widely used remotely sensed proxy for vegetation productivity. Meanwhile, a variety of processes are extensively altering Arctic-Boreal land cover, complicating the relationship between NDVI and productivity. Because high-quality information on land cover is lacking, understanding of relationships among Arctic-Boreal greenness trends, productivity, and land cover change is lacking. Multidecadal time series of moderate resolution (30 m) reflectance data from Landsat and high resolution (<4 m) imagery were used to map annual cover and quantify changes in land cover over the study domain of NASA’s Arctic-Boreal Vulnerability Experiment. Results identify two primary modes of ecosystem transformation that are consistent with increased high latitude productivity: (1) in the Boreal biome, simultaneous decreases in Evergreen Forest area and increases in Deciduous Forest area caused by fire and harvest; and (2) climate change-induced expansion of Arctic Shrub and Herbaceous vegetation. Land cover change imposes first-order control on the sign and magnitude of NDVI trends. Over a quarter of NDVI trends were associated with land cover change. Relative to locations with stable land cover, areas of land cover change were twice as likely to exhibit statistically significant trends in Landsat-derived NDVI. The highest magnitude trends were concentrated in areas of forest disturbance and regrowth and shrub expansion, while undisturbed land showed subtler, but widespread, greening trends. Based on Orbiting Carbon Observatory-2 data, sun-induced fluorescence, a proxy for productivity, reflected relationships among land cover, disturbance age, and productivity that were not fully captured in NDVI data. In contrast with NDVI, time series of aboveground biomass provide physically-based measures of productivity in forests. Using Landsat-based land cover and reflectance and ICESat lidar data, aboveground biomass was mapped annually across the study domain. Most forests showed increasing biomass, with wildfires imposing substantial interannual variability and harvest imposing steady biomass losses. This dissertation provides new information on how disturbances are driving land cover and productivity change across Arctic-Boreal northwestern North America and reveals insights regarding the interpretation of remote sensing observations in these biomes.
RightsAttribution 4.0 International