Briac Amundsen
Modelling Biophysical and Carbon Dioxide Exchange Properties in the Canadian Arctic using Soil-Adjusted Vegetation Indices © 2014
High resolution remote sensing data can be used to quantify, assess and predict Arctic vegetation and ecosystem carbon dioxide (CO2) exchange properties. Often, the Normalized Difference Vegetation Index (NDVI) is used for this purpose to relatively high degrees of accuracy. However, in harsh environments such as the Arctic, low vegetation cover and plenty of exposed soil have the potential to falsely increase NDVI values thereby reducing confidence in results. In this study, ten soil-adjusted vegetation indices are calculated from high spatial resolution IKONOS data covering two distinct Canadian Arctic sites. Using linear regression, soil-adjusted vegetation indices are compared to each other and to NDVI in their ability to accurately model vegetation biophysical and CO2 exchange properties. Results indicate that, in the High Arctic, soil-adjusted vegetation indices reduce soil noise effectively and outperform NDVI (r2 values were 0.87 for the best soil-adjusted vegetation index and 0.84 for NDVI when modelling percent vegetation cover). However, results from the Mid Arctic study site indicate the reverse with NDVI outperforming soil-adjusted vegetation indices (NDVI achieved r2 values of 0.80 while the best soil-adjusted vegetation index achieved an r2 of 0.77 when modelling percent vegetation cover). Thus, the benefits of using soil adjustment in remote sensing derived vegetation indices appear to be site-specific and dependant on latitude, climate and associated degrees of exposed soil.