Phenological change in the tundra is characterized by rapid transition seasons with volatile weather patterns. Snow cover over the winter months and along the transition seasons complicates our ability to use remote sensing metrics to detect such phenological change. Vegetation indices that track both chlorophyll content (e.g., NDVI, NIRv, and EVI) as well as photosynthetic capacity (e.g., PRI and CCI) are all sensitive to the presence of snow cover and emergent senescent vegetation (Gamon et al., 2013; Luus et al., 2017; Pierrat et al., 2021) (Figure 7). Further, photosynthesis of Arctic tundra vegetation may occur under snow cover (Parazoo et al., 2018; Starr & Oberbauer, 2003), thereby severely hindering the utility of spectroscopy for even detecting primary productivity throughout the year. Reliance on these measures without appropriate snow cover correction significantly inhibits their utility to determine phenological change over winter and transition seasons. For many tundra species, especially lichens, bryophytes, and evergreen shrubs and trees exhibiting limited intra-annual biomass production, changes in structural indices such as NDVI, NIRv, and EVI may better capture changes in snow on/off periods than actual changes in biomass (Figure 7) (Gamon et al., 2013; Luus et al., 2017; Pierrat et al., 2021). Cold temperatures and the lack of liquid water can force dormancy and limit photosynthesis, but if the vegetation remains green, changes in NDVI may be nominal. Tundra species have been shown to acclimatize to winter conditions by increasing the size of their pool of xanthophyll cycle pigments and by maintaining that pool largely as antheraxanthin and zeaxanthin (Verhoeven, 2014), which manifests as an increase in total carotenoid pigments, and can be captured by the CCI (Gamon et al., 2016; Wong et al., 2020). In evergreen needleleaf trees, strong seasonal variation in photoprotective pigments can be detected via PRI and CCI - attuned to variation in xanthophyll and bulk carotenoid pigments, respectively (John A. Gamon et al., 2016; Wong & Gamon, 2015b, 2015a). Strong linkages between sensitivity of cessation of radial stem growth in TTE spruce trees to end-of-season meteorology is also detectable by changes in PRI (Eitel et al., 2019, 2020). Similar investigations of PRI/CCI-growth and photosynthesis relationships on (non-tree) tundra vegetation would help advance understanding of Arctic tundra phenology. In addition, plant pigment composition serves as an important indicator of the timing of autumn entry into this seasonally downregulated (i.e., dormant) state (Figure 7). Hence, phenological analysis of tundra vegetation requires integration of multiple spectral metrics, preferably including narrowband measurements related to photoprotective pigment variation, to isolate seasonal change in plant structural and functional dynamics from confounding variation in snow cover.
Figure 7 . a) Shows phenocam images from different points during the year with varying degrees of snow cover on understory/tundra vegetation at NEON Delta Junction, AK. b)-c) Shows commonly used vegetation indices (NDVI, NIRv, PRI, and CCI) measured from a tower-based spectrometer system PhotoSpec (Grossmann et al., 2018) observing three understory tundra targets at a 30-minute resolution. d) Shows daily average PAR and SZA. For b)-d), plotted is the 5-day moving mean of the measured quantity. Shaded error bars indicate the standard deviation of diurnal variability. Shaded blue regions indicate the presence of snowfall on the understory as determined visually from phenocam images.
Many spaceborne instruments are temporally incompatible with the rapid phenological progression of tundra within a compressed growing season. Historically, analyses of seasonal change across the Arctic may leverage time series observations by the Landsat missions. However, the 16-day revisit frequency precludes accurate detection of timing of important events to quantify interannual variability in phenology. The similar temporal resolution (14-day revisit) proposed for SBG may yield similar challenges for phenology applications. Furthermore, due to the prevalence of cloud cover, infrequent observations reduce the opportunity for clear-sky imaging resulting in seasonally sparse or irregular observations. Both these issues are made apparent by tower-based observations (Figure 7), which enable continuous or high frequency observations but lack the spatial coverage of spaceborne observations. Tower-based observations in the boreal forest showed a 29 day difference in the timing of the springtime onset of photosynthesis between evergreen and deciduous tree species (Pierrat et al., 2021) Such temporal asynchrony - including among evergreen and deciduous tundra plants - may not be adequately captured by spatially and temporally coarse resolution spaceborne data. Thus, tower-based observations will play an integral role in understanding Arctic phenological change. Co-incident UAS observations can help bridge the spatiotemporal gap through repeated measurements at a lower temporal resolution than tower-based but at a much higher spatial range.

4.5. Diurnal variation

The primary intrinsic mechanisms driving diurnal changes in spectral reflectance are related to plant pigment composition, which regulate the efficiencies of photochemistry through dynamic changes in photoprotective pigment pools (xanthophylls, lutein) via sustained and rapidly reversible non-photochemical quenching (Adams et al., 2004), and hydration status for non-vascular vegetation. Dynamics among a cycling group of carotenoids, violaxanthin, antheraxanthin and zeaxanthin (V, A, and Z, respectively), known as the xanthophyll cycle, are especially informative in this regard (Demmig-Adams et al., 1996). During the photosynthetically active season, the state of the xanthophyll cycle responds to diurnal variation in incoming light via enzymatically regulated conversions between Z + A and V. These dynamics are often captured using spectral indices sensitive to changes at 531 nm (the photochemical reflectance index, PRI, (Gamon et al., 1992)). However, most other vegetation spectral changes are not associated with diurnal physiological dynamics; hence, these spectral indices (i.e., NDVI, NIRv, and CCI) can remain relatively invariant (Figure 8) with the exception of changes in moisture status for non-vascular vegetation (Figure 4). Most spectral changes in the VIS-SWIR range throughout the course of the day are associated with changes in viewing-illumination geometries, as illustrated in subplots of NDVI, NIRv, CCI in Figure 8.