Given the strong seasonality of Arctic vegetation (see Section 4.4),
additional consideration of the timescale of measurement and underlying
phenomena are critical to mapping efforts. Coupled observations across
spatial scales that can be conducted with high observation frequency
across seasons will help resolve this challenge (Table 2). Further, both
seasonal and interannual evaluations of change in the Arctic tundra must
consider the constraints of winter in terms of both sampling design and
physiological effects. The rapid seasonal progression (as discussed in
Section 4.4) imposes tremendous challenges for benchmarking the onset of
seasonal photosynthetic activity and tissue growth, quantifying
sensitivity to shoulder season stress, and detecting legacy effects on
productivity in subsequent seasons. In particular, the strong
seasonality of photoprotective pigments in evergreens (see Section 3.4
and 4.4), which complicate interpretation of spectral reflectance,
requires further research across the Arctic tundra domain to improve
parameterization of models of primary production. Additionally,
deciduous shrub species exhibit strong autumn leaf coloration with
photoprotective pigments present (and chlorophyll content declining)
during leaf senescence that may facilitate remotely sensed
quantification of species cover values. For example the red autumn
leaves of birch continue to actively photosynthesize even though
chlorophyll pigments may be less evident by traditional greeness-based
remote sensing (Patankar et al., 2013). Spectroscopy is well suited to
address these challenges, and could likely help disentangle the timing
of vegetation responses among plant functional types.
The use of optical remote sensing information over large regions (i.e.,
across continents) and through time (i.e., multiple decades) has
increased considerably in recent years (Ustin & Middleton, 2021). This
includes IS data in the Arctic (e.g., (Langford et al., 2019)), given
the increased availability of these datasets (Miller et al., 2019).
However, new approaches for access, use, and analysis of large IS
datasets will be needed given the growing volume of remote sensing
observations across scales. For example, fusing high volume data from
novel UAS and ground-based platforms and expanded use of datasets across
scenes and locations will greatly increase the overall volume of data
for any given project. Seasonal weather conditions and sun-sensor
geometry changes in the Arctic mean that a considerable fraction of data
may have variable data quality over scenes or across scenes in a study
area. Similarly, current methods for retrieval of IS data require manual
search, collection and combining of data across different locations by
end-users. To ease and expand use of IS data for Arctic researchers, it
is recommended that data systems provide analysis-ready (e.g.,
geo-rectified and consistent atmospheric correction) and cloud-optimized
data storage formats (e.g., cloud-optimized GeoTIFF). In addition, files
should be accessible on storage buckets (i.e., basic container that
stores bulk data, usually used for organizing combinations of similar
datasets, e.g., S3 or Google cloud bucket) through cloud-based tools to
facilitate rapid search, filter, and extraction of data across specific
locations, regions and scenes. Similarly, it is recommended that
cloud-based tools facilitate basic analyses, data transformation,
subsetting, and application of mapping algorithms without downloading
large volumes of IS datasets but instead the final derived products or
results of the cloud pre-processing. For example, this could be
facilitated through the use of a cloud storage location within Google
Earth Engine (GEE) or GEE within the Python or R environments. By moving
IS data access to the cloud would also facilitate easy combination with
other remote sensing data or even multiscale observations, including UAS
data. This would also reduce the data latency from collection to
community use and allow more users to facilitate discovery of novel and
important patterns in phenomena in the Arctic biome.
We described important attributes of tundra ecosystems that impose
challenges for conducting spectroscopy, including plant functional type
and pixel-composition characteristics, intrinsic dimensionality, and
capacity for land cover classification, change detection, time series
observations, and characterizing biophysical properties. Future
spectroscopy missions such as SBG would be well-advised to consider the
challenges of complex biomes such as the Arctic tundra during mission
development and especially for data product generation. To address these
challenges, an optimized mixture of narrow and broad bands should be
considered for SBG to accurately characterize Arctic vegetation.