Permafrost and Freshwater Ice

Surface deformation caused by changing permafrost conditions poses a threat to existing and planned infrastructure in many areas within the Arctic and sub-Arctic.  Satellite based synthetic aperture radar interferometry (InSAR) is a method that measures surface deformation at continuously repeating intervals over wide spatial scales.  This provides quantification of ongoing change and also has potential to map areas at risk of future change by characterizing the surface expression of seasonal changes to the permafrost active layer.

Robust monitoring of permafrost areas with InSAR is faced with several challenges related to the temporal and spatial characteristics of the measured deformation.  Seasonal snow cover reduces coherence over many months of the year and results in wide scale phase biases.  Complex deformation dynamics arise due to the combination of seasonal active-layer induced heave and long term subsidence caused by permanent changes in ice content.  Spatial variability in permafrost conditions frequently occur due to variations in surficial geology and other factors.  This can result in high spatial phase gradients which make phase unwrapping difficult.

We present an InSAR monitoring solution designed to provide sufficient spatial and temporal sampling to adequately characterize the surface deformation in these areas.  This method, termed Multi-Track Homogenous Distributed Scatterer InSAR jointly processes multiple same-side SAR data stacks from one or more sensors which improves the achievable temporal sampling rate.  Spatial resolution is preserved while supressing phase noise by the use of spatially adaptive multi-looking.   The method also includes a model for the surface expression of seasonal active layer changes as an aid to robust phase unwrapping and as a means for quantifying seasonal deformation.

SAR datasets for this study have been acquired over five areas in northern Canada: 1) city of Yellowknife (62° 27′ N, 114° 24′ W), 2) village of Umiujaq (56° 33′ N, 76° 33′ W), 3) village of Salluit (62° 12′ N, 75° 39′ W),  4) section of the Alaska highway in the Yukon territory (62° 20′ N, 140° 50′ W) and 5) section of the Inuvik to Tuktoyaktuk highway (68° 22′ N,  133° 44′ W).  These sites were all chosen because they are within sporadic or discontinuous permafrost zones and are known to have ongoing permafrost related surface displacement threatening existing or planned infrastructure.  These data include ~ 740 SAR images from a total of 22 RADARSAT-2, TerraSAR-X and COSMO-SkyMED interferometric stacks.  This diverse set allows for demonstration of both single sensor Multi-Track analysis with RADARSAT-2 and also joint C-/X-band analysis with all three sensors.  They also allow for ascending-descending decomposition of deformation into vertical and east-west directional components.

Surface deformation maps decomposed into linear (long-term) deformation and seasonal (active layer/ frost heave) deformation components are presented along with deformation time series for a series of natural and infrastructure areas of interest.  Spatial deformation patterns and locations of coherent targets are shown to correlate well with independently derived surficial geology maps.  The proposed method is shown to provide sufficiently high spatial and temporal resolution to allow robust characterization of permafrost induced surface deformation in both natural terrain and infrastructure.

Lakes that form a seasonal ice cover are a major component of the terrestrial landscape. They cover approximately 2% of the Earth’s land surface, with the majority of them located in the Northern Hemisphere. The presence (or absence) of ice cover on lakes during the winter months affects both regional climate and weather events, such as lake-effect snowfall. Monitoring of lake ice is critical to our skill at forecasting high-latitude weather, climate, and river runoff. Lake ice is also a sensitive indicator of climate variability and change. Despite the wide-ranging influences of lake ice and its robustness as an indicator of climate change, few satellite-derived lake ice products have been developed to date.

In this talk, we will present results from: 1) assessment of the existing NASA Aqua/Terra Moderate Resolution Imaging Spectroradiometer (MODIS) Snow Products (MYD10_L2 and MOD10_L2, Collection 5) for lake ice; 2) development and assessment of a new lake ice cover algorithm using reflectance from Bands 2,3 and 4 of MODIS Calibrated Radiances 5-Min L1B Swath 500m V005 (MYD02HKM and MOD02HKM) products; and 3) development and assessment of a new lake ice cover algorithm based on reflectance from Suomi NPP Visible Infrared Imaging Radiometer Suite (VIIRS) I-Bands 1 and 2. The NASA and new MODIS/VIIRS-derived lake ice products have been evaluated against lake ice cover extent extracted manually from MODIS/VIIRS color composite images as well as Canadian Ice Service weekly ice fractions from three lakes (Lake Winnipeg, Great Slave Lake and Great Bear Lake) over two ice seasons (2012-2013 and 2013-2014), with a focus on the break-up period. The new MODIS lake ice algorithm is shown to outperform the existing NASA algorithm by 15-60 %.

Information on soil freezing and thawing is important for hydrological, biogeochemical, and climatological applications. Soil freezing affects  surface energy balance, surface and subsurface water flow, and exchange rates of carbon with the atmosphere. Soil freezing also has an effect on important biogeochemical processes (photosynthesis and microbial activity).

Together with Gamma Remote Sensing and Consulting AG and Environment Canada, the Finnish Meteorological Institute (FMI) has developed a method for detecting soil freezing and thawing on a global scale using L-band satellite microwave radiometer observations. The work has been carried out within the frame of several ESA studies: STSE SMOS+ Innovation and STSE Pathfinder programs and under nationally funded SMOS Calibration and Validation Activities in Northern Areas (NORA) program. The soil freeze/thaw detection methods are based on using L-band observations from the SMOS (Soil Moisture Ocean Salinity)  mission. In the most simplistic form, only SMOS brightness temperature (T_B) data is used to retrieve global soil state maps for the Northern Hemisphere. Soil permittivity decreases at L-band due to soil freezing, which results in higher T_B values for frozen soil than for thawed soils. This forms the basis of the soil state monitoring algorithm. Threshold values for frozen and thawed states are first defined for SMOS T_B observations; the soil is then categorized to “frozen”, “partially frozen” or “thawed” states by comparing the current observation to these threshold values. Naturally, the threshold values are highly dependent on the composition of: land class, soil type, open water fraction, annual physical temperature variation, etc within each pixel. To standardize the method and to enable automatic threshold detection, SMOS T_B values are normalized to relative values, which reach a maximum value of 100 in winter and minimum value of 0 in summer. The summer minimum and winter maximum values are experimentally determined from SMOS data for each pixel. Auxiliary data are used to define the seasonal variation in each pixel. The end of the summer period is estimated using WMO synoptic air temperature data interpolated over the Northern Hemisphere.  The end of winter period is estimated using snow masks derived from the ESA DUE GlobSnow product. With auxiliary data, clearly erroneous soil state detections can be automatically corrected. Currently the method is being validated against in situ data acquired from different environmental conditions and geographic regions. The objective is to provide a global operational soil freeze/thaw product for SMOS, which is to be updated on a daily basis.

Further work is conducted to enhance the algorithm to take into account the dry snow pack on top of the soil. Snowpack affects the observed signal through refraction and impedance matching, affecting also soil freeze/thaw estimates in areas with seasonal snow cover. With the enhanced algorithm the soil permittivity is retrieved by means of inverting a simplified emission model for L-band, taking into account the impedance matching and refraction angle caused by a dry snowpack. Instead of using threshold values for different soil state based on SMOS T_B data, these values are defined directly based on the retrieved soil permittivity. The soil permittivity data obtained from SMOS can be further used to parameterize other satellite retrievals using both passive and active microwave observations. Using a Bayesian assimilation scheme, we have used SMOS soil permittivity information to regulate  SAR (synthetic aperture radar) –based retreivals of soil permittivity. This enables the regional downscaling of coarse resolution SMOS retrievals  with high resolution SAR data. Results using SMOS with historical datasets from  the ENVISAT ASAR instrument for autumn freezing periods in  2010 and 2011 are presented, as well as the first results using SMOS and Sentinel-1 SAR for the autumn of 2015. In principle, the presented methodology is applicable also to to NASA SMAP (Soil Moisture Active Passive) observations, as well as other space-borne radiometers and SAR sensors.

Knowledge of water-surface velocities in rivers is useful for understanding a wide range of processes related to river hydraulics and hydrology, geomorphology and ecology. In cold regions, river-ice break up and the associated downstream transport of ice debris is often the most important hydrological event of the year, producing flood levels that commonly exceed those for the open-water period and dramatic consequences. The economic costs from break-up ice jams are estimated to average almost 250 million USD per year in North America and to have been over 100 million USD for a single 2001-event in Siberia. Although typically less dynamic than break-up, freeze-up can create a similar set of problems on many cold-regions rivers. River surface velocities also reflect water fluxes and with that fresh water input to oceans, such as the Arctic ocean, that are notoriously difficult to measure during ice break-up.

In a novel approach, we exploit the typical time lag between air- or spaceborne stereo images on the order of seconds to minutes to track river ice debris floating with the water during freeze-up and break-up. We demonstrate the method for the Lena River, Siberia, and Mackenzie River and Yukon, North America, using stereo data from ASTER, ALOS PRISM, satellites from the DigitalGlobe constellation, and airplanes. For reaches of several hundred kilometres length river velocities are obtained for Lena and Yukon rivers exhibiting a range of new insights into river morphology. For Mackenzie River we estimate water fluxes from surface velocities and compare to measured discharge, confirming that such measurements can give reliable discharge estimates at flux gates with known bathymetry. Over a reach of Tanana River, a tributary to the Yukon, we compare space-derived water velocity fields during freeze-up and break-up, and over several years. Fine velocity details are investigated from airborne stereo imagery acquired over the Yukon.

Requirements  for  space  based  monitoring  of  permafrost  features  had  been  already  defined  within  the  IGOS Cryosphere Theme Report at the start of the IPY in 2007 (IGOS, 2007). The  WMO Polar  Space  Task  Group  (PSTG,  http://www.wmo.int/pages/prog/sat/pstg_en.php)  identified the need to review the requirements for permafrost monitoring and to update these requirements in 2013.

Relevant surveys with focus on satellite data are already available from the ESA DUE Permafrost User requirements survey (2009), the United States National Research Council (2014) and the ESA – CliC – IPA – GTN –P workshop in February 2014. These reports have been reviewed and specific needs discussed within the community and a white paper submitted to the WMO PSTG. Acquisition requirements for monitoring of especially terrain changes (incl. rock glaciers and coastal erosion) and lakes (extent, ice properties etc.) with respect to current satellite missions have been specified.

About 50 locations ('cold spots') where permafrost (Arctic and Antarctic) in situ monitoring has been taking place for many years or where field stations are currently established have been identified. These sites have been proposed to the WMO Polar Space Task Group as focus areas for future monitoring by high resolution satellite data. The specifications of these sites including meta-data on site instrumentation have been published as supplement to the white paper (Bartsch et al. 2014).

The representativity of the ‚cold spots‘ around the arctic has been in the following assessed based on a landscape units product which has been developed as part of the FP7 project PAGE21. 

The ESA DUE Permafrost service has been utilized to produce a pan-arctic database (25km, 2000-2014) comprising Mean Annual Surface Temperature, Annual and summer Amplitude of Surface Temperature, Mean Summer (July-August) Surface Temperature. Surface status (frozen/unfrozen) related products have been also derived from the ESA DUE Permafrost service. This includes the length of unfrozen period, first unfrozen day and first frozen day. In addition, SAR (ENVISAT ASAR GM) statistics as well as topographic parameters have been considered. The circumpolar datasets have been assessed for their redundancy in information content. 12 distinct units could be derived. The landscape units reveal similarities between North Slope Alaska and the region from the Yamal Peninsula to the Yenisei estuary. Northern Canada is characterized by the same landscape units like western Siberia. North-eastern Canada shows similarities to the Laptev coast region. 

This paper presents the result of this assessment and formulates recommendations for extensions of the in situ monitoring networks and categorizes the sites by satellite data requirements (specifically Sentinels) with respect to the landscape type and related processes.

Bartsch, A et al. (2014): Permafrost longterm monitoring sites (Arctic and Antarctic). doi:10.1594/PANGAEA.847003, Supplement to: Bartsch, A. et al. (2014): Requirements for monitoring of permafrost in polar regions - A community white paper in response to the WMO Polar Space Task Group (PSTG), Version 4, 2014-10-09. Austrian Polar Research Institute, Vienna, Austria, 20 pp, hdl:10013/epic.45648.d001