|Glacier||Glacial lake/pond||Ice cliff|
Manual survey: Glacier velocity can be determined manually, using a total station, theodolite, or a dGPS for example, to survey stakes inserted into the glacier surface. This produces accurate measurements that also quantify vertical as well as horizontal movement, which is important for debris-covered glaciers which can be responding to climatic change through surface lowering, rather than terminus retreat.
Manual surveys require repeat travel and access to the galcier surface, which is time consuming, likely to be expensive, and may not be possible in certain areas of the glacier where crevasses are numerous. The spatial and temporal coverage of the survey therefore requires consideration.
Remotely sensed: Semi-automatic workflows using optical and radar imagery can be used to quantify glacier velocity by tracking the displacement between two images of a known time separation (see diagram). This requires that surface features are preserved between the images so the technique performs well on slow moving debris-covered glaciers. The spatial coverage of measurements is also likely to be increased compared to that achievable in a manual survey, including hazardous crevassed areas.
Typical feature tracking workflow for optical imagery.
Velocity field derived for the Batura Glacier using Landsat 7 ETM+ panchromatic scenes and CIAS software, available from: http://www.mn.uio.no/geo/english/research/projects/icemass/cias
Back to top
Glacier mass balance studies derive and compare net mass loss and gain over an entire glaciers surface during a given time period. Field-based mass balance surveys require substantial amounts of time and logistical effort to complete, so the geodetic approach is now more commonly used and applied to wide areas. The geodetic method calculates glacier mass balance through glacier wide surface elevation comparisons (Nuth et al., 2007), typically over the period of a few years to decades, using digital elevation models (DEMs). DEMs of very high resolution can be generated using close-range remote sensing techniques, such as laser scanning, or of moderate resolution using satellite imagery. DEMs generated using satellite imagery cover can cover thousands of km2, and so are excellent data sources for assessing glacier mass loss at the catchment scale. Stereoscopic DEMs are generated considering photogrammetric principles (Nuth and Kaab, 2011) using overlapping imagery from sensors such as ASTER, ALOS PRISM, the SPOT satellites and, most recently, the Worldview and Pleiades sensors.
DEM differencing to show surface elevation change over glaciers in the Everest region of the Himalaya, between 2000 and 2015. DEMs of the 2015 land surface were generated from ASTER and Worldview imagery and differenced from the Shuttle Radar Topographic Mission (SRTM) DEM generated in 2000. Elevation difference data have been converted to annual rates of surface elevation change.
When glacier surface DEMs are available for different time periods, the analysis of the difference between them can yield ice-surface elevation change data. The differencing of glacier surface DEMs on a pixel by pixel basis and the subsequent multiplication of elevation differences by pixel area can also give estimates of volumetric changes over the study period. This volume change can then be converted to mass change, a more relevant quantity for climate impact assessments of sea level rise contributions and mountain hydrology (Huss, 2013), through the multiplication with the density of glacier/ firn ice (Racoviteanu et al., 2008; Huss, 2013).
The accuracy of mass-change estimates from DEM-differencing work can be estimated through the analysis of elevation change data over ground thought to be stable over the study period- i.e. off-glacier. When the difference between DEMs used over the study period is minimised in these stable areas, we can be more certain in any more substantial changes over glacier surfaces.
HUSS, M. 2013. Density assumptions for converting geodetic glacier volume change to mass change. Cryosphere, 7, 877-887.
NUTH, C., KOHLER, J., AAS, H. F., BRANDT, O. & HAGEN, J. O. 2007. Glacier geometry and elevation changes on Svalbard (1936-90): a baseline dataset. In: SHARP, M. (ed.) Annals of Glaciology, Vol 46, 2007. Cambridge: Int Glaciological Soc.
NUTH, C. & KAAB, A. 2011. Co-registration and bias corrections of satellite elevation data sets for quantifying glacier thickness change. Cryosphere, 5, 271-290.
RACOVITEANU, A. E., WILLIAMS, M. W. & BARRY, R. G. 2008. Optical remote sensing of glacier characteristics: A review with focus on the Himalaya. Sensors, 8, 3355-3383.
Back to top
Glacial lake/ pond
Glacial lakes can be mapped manually using remotely sensed imagery if sufficient contrast is observable between the surrounding terrain. However, to allow a semi-standardised comparison between imagery of different times, the Normalised Difference Water Index (NDWI) is commonly applied to multi-spectral imagery (e.g. Landsat). This uses the difference between high and low water reflectance at different wavelengths to distinguish water from the surrounding terrain. The accuracy is dependent upon this contrast and the threshold selected to separate water from land, and also the resolution of the imagery, which determines the size of lake that can be mapped and the area of mixed pixels that are included. Mixed pixels cross the water-shore boundary so have a spectral signature incorporating both land covers. Coarser resolution data increases this overlap, which increases the uncertainty when delineating the lake-shore boundary.
Landsat imagery is commonly used to delineate large proglacial lakes and has also been used to map smaller supraglacial lakes. The variable size of supraglacial water bodies means the 30 m pixel size of the Landsat imagery leads to increased uncertainty for smaller lakes, whereas finer resolution ASTER (15 m) or ALOS ANVIR (10 m) imagery can improve this classification accuracy.
Application of the Normalised Difference Water Index (NDWI) to Landsat imagery of the Everest region of Nepal.
Quantifying the expansion of Chubda glacial lake using multi-temporal Landsat imagery.
Back to top
Knowledge of the bathymetry of glacial lakes is important when estimating the volume of water contained in the lake. This can be used in conjunction with dam-breach models to estimate downstream inundation characteristics if an unstable moraine dam was to breach (see GLOFs). Deepening bathymetry over time indicates the depth of the lake is increasing, which suggests the bottom is underlain by dead ice. This ice melts due to the thermal energy transmitted by the water above which means the volume of the lake can expand over time. A thick debris cover on the lake bed can restrict this melting by insulating the ice beneath.
The simplest means of measuring lake depth is to use a weighted line (plumb line) deployed from a boat. These can be conducted in a regular grid across the lake so that the spatial variation of depth can be interpolated between measurements. Each measurement of depth therefore requires a GPS position fix. Measurements are limited by the length of the line: proglacial lakes are typically < 200 m in depth and supraglacial lakes may be ~ 10 m. The technique is therefore time-consuming over a large lake area. If a lake is frozen sufficiently to walk on, holes can be drilled through the ice to insert the plumb line.
Echosounding with a sonar device uses sound pulses transmitted vertically into water and uses the time between the emission and return of the pulse to calculate lake depth. These devices can produce an accuracy on the order of centimetres to metres and may also have an integrated GPS device. ‘Fish-finders’ have successfully been used to map lake depth and can even work through ice in some circumstances.
Bathymetry mapping using echo sounding. U.S. Navy Graphic Illustration.
A shore-based method of deriving lake bathymetry is to use an underwater exploration robot fitted with a depth sensor. An OpenROV fitted with a 100 m tether is one example, which can be operated using a standard laptop for 2-3 hours on one charge. A technique to precisely locating the ROV underwater is under development, but using knowledge of the ROV’s speed and heading it can be estimated. If conducting transects across the lake, the position can also be estimated using the known length of tether paid out and the ROV’s depth.
OpenROV underwater exploration robot with tether attached.
Back to top