Post-wildfire surface deformation near Batagay, Eastern Siberia,
detected by L-band and C-band InSAR
Kazuki Yanagiya1 and Masato Furuya2
1Department of Natural History Sciences, Graduate
School of Science, Hokkaido University.
2Department of Earth and Planetary Dynamics, Faculty
of Science, Hokkaido University.
Corresponding author: Kazuki Yanagiya
(k.yanagiya@frontier.hokudai.ac.jp) and Masato Furuya
(furuya@sci.hokudai.ac.jp)
Key Points:
- Post-wildfire surface deformation on the northwest of Batagay, Eastern
Siberia, was detected by two independent Interferometric Synthetic
Aperture Radar systems.
- L-band long-term and C-band short-term interferograms indicate the
spatial and temporal complexity of the deformation in terms of both
subsidence and uplift.
- Spatial heterogeneities of the subsidence magnitude were clearly
correlated to the gully development, whereas the burn severity was
rather homogeneous.
- Detection of enhanced uplift signals at the fire scar and its
interpretation based on a physics-based frost heave theory.
Abstract
Thawing of ice-rich permafrost and subsequent ground subsidence can form
characteristic landforms, and the resulting topography they create are
collectively called “thermokarst”. The impact of wildfire on
thermokarst development remains uncertain. Here we report on the
post-wildfire ground deformation associated with the 2014 wildfire near
Batagay, Eastern Siberia. We used Interferometric Synthetic Aperture
Radar (InSAR) to generate both long-term (1-4 years) and short-term
(sub-seasonal to seasonal) deformation maps. Based on two independent
satellite-based microwave sensors, we could validate the dominance of
vertical displacements and their heterogeneous distributions without
relying on in-situ data. The inferred time-series based on L-band ALOS2
InSAR data indicated that the cumulative subsidence at the area of
greatest magnitude was greater than 30 cm from October 2015 to June
2019, and that the rate of subsidence slowed in 2018. The burn severity
was rather homogeneous, but the cumulative subsidence magnitude was
larger on the east-facing slopes where the gullies were also
predominantly developed. The correlation suggests that the active layer
on the east-facing slopes might have been thinner before the fire.
Meanwhile, C-band Sentinel-1 InSAR data with higher temporal resolution
showed that the temporal evolution included episodic changes in terms of
deformation rate. Moreover, we could unambiguously detect frost heave
signals that were enhanced within the burned area during the early
freezing season but were absent in the mid-winter. We could reasonably
interpret the frost heave signals within a framework of premelting
theory instead of assuming a simple freezing and subsequent volume
expansion of pre-existing pore water.
Plain Language Summary
Wildfires in arctic regions not only show an immediate impact on nearby
residents but also long-lasting effects on both regional ecosystems and
landforms of the burned area via permafrost degradation and subsequent
surface deformation. However, the observations of post-wildfire ground
deformations have been limited. Using satellite-based imaging technique
called Interferometric Synthetic Aperture Radar (InSAR), we detected the
detailed spatial-temporal evolution of post-wildfire surface deformation
in Eastern Siberia, which helps in understanding permafrost degradation
processes over remote areas. Post-wildfire areas are likely to be focal
points of permafrost degradation in the Arctic that can last many years.
1 Introduction
Wildfires in boreal and arctic regions are known to have increased over
recent decades in terms of both frequency and areal coverage (e.g.,
Kasischke & Turetsky, 2006; Hu et al., 2010), and have had significant
impacts on permafrost degradation (e.g., Jafarov et al., 2013; Zhang et
al., 2015; Gibson et al., 2018). Although fires do not directly heat up
the subsurface space deeper than 15 cm (Yoshikawa et al., 2003), severe
burning decreases surface albedo, and removes vegetation and the surface
organic soil layer that previously acted as insulators buffering from
changes in air temperature. Subsequent increases in both soil
temperature and thickness of the active layer, a near-surface layer that
undergoes a seasonal freeze-thaw cycle, have been documented for up to
several years after the fire (e.g., Yoshikawa et al., 2003). Meanwhile,
in ice-rich permafrost regions, the thawing of permafrost and the
melting of massive ice can lead to formation of characteristic landforms
such as thaw pits and ponds, and retrogressive thaw slumps. While there
are a variety of classifications in terms of morphological and
hydrological characteristics (Jorgenson, 2013), those thaw-related
landforms and the topography they create are collectively termed as
“thermokarst”. However, the role of wildfires in developing
thermokarst terrain remains quantitatively uncertain. Moreover, in
comparison to the controlled warming experiments in Alaska (Hinkel and
Hurd Jr, 2006; Wagner et al., 2018), wildfires in arctic regions may
also be viewed as uncontrolled disturbance experiments that aid in
understanding the permafrost degradation processes.
Ice-rich permafrost deposits, known as the yedoma ice complex (yedoma),
are widely distributed in the lowland of Alaska and Eastern Siberia
(Kanevskiy et al., 2011; Schirrmeister et al., 2013). The greatest
subsidence within the 2007 Anaktuvuk River tundra fire scar was
identified in the yedoma upland by LiDAR (Jones et al., 2015). Yedoma is
a unique permafrost deposit in terms of its extraordinarily high volume
of ice (50-90 %) and organic-rich sediments. While the organic carbon
trapped in permafrost regions is estimated to be twice that in the
current atmosphere, permafrost thawing and related thermokarst processes
may release the carbon as greenhouse gasses (CO2 and
CH4) via microbial breakdown, which may further promote
global warming (Mack et al., 2011; Schuur et al., 2015). Thus, in order
to estimate the volume of greenhouse gasses released, it is important to
evaluate the volume of thawed ice associated with thermokarst processes
in yedoma-rich areas.
Near the village of Batagay, Sakha Republic, Eastern Siberia (Figure 1),
there exists the Batagaika megaslump, known as the world’s largest
retrogressive thaw slump, exposing roughly 50-90 m thick yedoma deposits
on the north-east facing slope (e.g., Kunitsky et al., 2013; Murton et
al., 2017). Thaw slumps are characterized by a steep headwall
surrounding a slump floor and develop as a result of rapid permafrost
thawing. The Batagaika megaslump was initiated at the end of the 1970s
by deforestation but still appears to be growing (Günther et al., 2016).
Considering this feature, it is worth considering whether new
disturbances in the proximity will result in the formation of similar
landforms. A wildfire incident occurred in July 2014 near Batagay,
which, like deforestation, will change the ground thermal regime.
Therefore, it is important to examine whether future catastrophic
thermokarst development could be similarly initiated at the fire scar,
whose area is much larger than the Batagaika megaslump (Fig 1b).
The first objective of this study was to assess the effectiveness of
satellite Interferometric Synthetic Aperture Radar (InSAR) in detecting
surface deformation signals due to wildfire-induced thermokarst over
different temporal scales. InSAR has been used to detect long-term and
seasonal displacements over several thaw-related landforms in permafrost
areas (e.g., Liu et al., 2010, 2014, 2015; Short et al., 2011; Iwahana
et al., 2016; Molan et al., 2018; Antonova et al., 2018; Strozzi et al.,
2018; Chen et al., 2018). Although subsidence signals as a result of
thermokarst associated with Alaskan wildfires have been detected using
InSAR (Liu et al., 2014; Iwahana et al., 2016a, 2016b; Molan et al.,
2018; Michaelides et al., 2019), no such studies have been conducted on
Siberian fires, to our knowledge. Also, all previous InSAR-based
post-wildfire deformation mapping has been performed over relatively
flat terrains, but no reports over hillslopes have been shown. Moreover,
in contrast to previous studies, we employed two independent SAR
imageries with distinct carrier frequencies and polarizations, L-band
(1.2 GHz) HH- and C-band (5.4 GHz) VV-polarized microwave. Because the
imaging geometries were different and had different sensitivities to the
3D displacement vector, we could not only take advantage of the
performance of each sensor in mapping deformation signals but could also
cross-validate the measurements by
two InSAR data sets.