Spectral indices for remote sensing of phytomass, deciduous shrubs, and productivity in Alaskan Arctic tundra

Predicting aboveground biomass in Arctic landscapes using very high spatial resolution satellite imagery and field sampling

“…Examination of previous studies in sub-Arctic or Arctic environments suggests that coarse patterns in vegetation and biomass distribution are easier to detect than finescale variations. Usually moderate to high R 2 ajd values (>0.4) have been obtained in studies from plot to circumpolar scale (Hope, Kimball, and Stow 1993;Walker et al 2003;Riedel, Epstein, and Walker 2005;Heiskanen 2006;Fuchs et al 2009;Kushida et al 2009;Raynolds et al 2012;Atkinson and Treitz 2013;Buchhorn et al 2013;Doiron et al 2013;Collingwood et al 2014;Johansen and Tommervik 2014;Kushida et al 2015). The highest R 2 ajd values (>0.7) between NDVI and biomass have been obtained in studies that use moderate to coarse resolution satellite datasets (Walker et al 2003;Heiskanen 2006;Raynolds et al 2012;Buchhorn et al 2013;Johansen and Tommervik 2014;Berner et al 2018).…”

“…The highest R 2 ajd values (>0.7) between NDVI and biomass have been obtained in studies that use moderate to coarse resolution satellite datasets (Walker et al 2003;Heiskanen 2006;Raynolds et al 2012;Buchhorn et al 2013;Johansen and Tommervik 2014;Berner et al 2018). In studies that use plot-scale NDVI measurements or VHSR imagery, the R 2 ajd values have often been near 0.5 or even below it (Hope, Kimball, and Stow 1993;Riedel, Epstein, and Walker 2005;Fuchs et al 2009;Kushida et al 2009;Atkinson and Treitz 2013;Kushida et al 2015) but there are also some exceptions (Atkinson and Treitz 2013;Buchhorn et al 2013;Collingwood et al 2014). One reason behind this disparity might be that the variation in spectral reflectance patterns is more evident at coarser scales, which usually also include areas with no or little biomass and low NDVI values.…”

“…To get field validation data for remote sensing studies, harvested biomass samples at a plot scale are needed (Hope, Kimball, and Stow 1993;Walker et al 2003;Raynolds, Walker, and Maier 2006;Kushida et al 2015;Greaves et al 2016).…”

20 cm resolution mapping of tundra vegetation communities provides an ecological baseline for important research areas in a changing Arctic environment

“…Although our intensity data are not radiometrically calibrated or normalized (Yan and Shaker 2014), the intensity response showed few flight-line artifacts, and the nonparametric nature of the Random Forest modeling approach permits the inclusion of raw, uncorrected data. All predictor metrics were selected to be ecologically meaningful for Arctic vegetation distribution, as well as being likely to improve prediction of vegetation characteristics based on previous work (e.g., Ohmann and Gregory, 2002, Dirnböck et al 2003, Naito and Cairns, 2011, Kushida et al 2015, Greaves et al 2016.…”

Aboveground biomass patterns across treeless northern landscapes

“…In treeless northern landscapes, coarse-scale biomass patterns have been tracked with greenness indices, such as normalized difference vegetation index (NDVI; Rouse et al 1974) obtained from optical remote sensing data (Raynolds et al 2012;Walker et al 2003). Greenness indices quantify especially the photosynthetic tissue biomass (Kushida et al 2015) but have also functioned when estimating the total aboveground biomass (Berner et al 2018;Johansen and Tommervik 2014;Atkinson and Treitz 2013). Nevertheless, in particular at finer scales, some studies have shown that greenness indices are poorly correlated with biomass and argued that there are no generalizable NDVI-biomass connections that can be used across the circumpolar north (Cunliffe et al 2020b;Bratsch et al 2017).…”