C31A-0452
Relationship Between ELA and SAR Backscatter Parameters for Selected Transects of the Greenland Ice Sheet
Changes in the position of the Equilibrium Line Altitude (ELA) provide important information on the state of mass balance of the Greenland Ice Sheet. Direct measurement of the ELA requires a significant amount of meteorological information, and is not practical on a large-scale basis, so it is desirable to develop reliable remote sensing methods for this purpose. The sensitivity of the backscatter of Synthetic Aperture Radar (SAR) to the snow characteristics suggests a possible relationship between the ELA and some backscatter features, and as a result, there have been several attempts to derive a SAR-based ELA retrieval technique. These attempts have focused on the detection of snow line, which is usually located close to the ELA; however no significant correlation was found between the ELA and the SAR-derived snow line. A new technique to monitor the ELA change by SAR is suggested and presented here. Based on an enhanced semi-empirical model of radar backscatter from snow pack, we developed a method to retrieve the snow depth normalized by the penetration depth (snow optical thickness). We exploit the fact that for areas where the snow optical thickness (SOT) is less than unit, the backscatter is sensitive to the snow depth changes. The SOT profiles for the end of the ablation season have been retrieved for period 1997-2003 from RADARSAT-1 C-band SAR data for several transects within an area around the Swiss Camp. These were compared to field data of the ELA (courtesy of Dr. K. Steffen). A high correlation (R2=0.845) has been found between the ELA and the position of the mathematical center of gravity (COG) of a section of the SOT profile that lies between 1000 m and 2000 m altitude. In accordance with the known field data, these altitudes are close to the borders of firn area impacted by melt-freeze cycles. Thus, we assume that the COG altitude change is governed by the firn warming zone extension and transformation and it can successfully be used to determine changes in the ELA by SAR for large-scale areas as well.
C31A-0453
Observations of atmosphere and ocean/land surfaces using UAVs in Ny-Alesund
Using a Norwegian small robotic aircraft called the Cryowing, atmospheric and ocean/land surface observations were made over Ny-Alesund, Svalbard, Norway from 31 July to 17 August 2008. The aircraft has meteorological sensors to observe dynamic and static air pressures, air temperature, and relative humidity, infrared thermometer to observe surface temperature, and digital camera to record the surface characteristics. From the results of surface mapping, the atmosphere responded differently with different types of surface (i.e., grass, crust, glacier, and ocean). The continuous observation at the same region showed that the time change in incoming solar radiation affects the variation of surface temperature due to difference in surface albedo, resulting in spacial distribution of air temperature in the boundary layer. While in the long ocean flight, the strong SST gradient was observed between the warm Atlantic water and melted water of sea ice near the Fram Strait. The air temperature above 150-m level observed by the UAV was clearly high over the warm current. These results demonstrate the utility of recent advances in UAV technology for monitoring and interpreting the spatial variations in cryosphere.
C31A-0454
Detecting Surface Melt on the Southern Patagonian Icefield using AMSR-E
The Southern Patagonian Icefield (SPI) straddles the Andes Mountains between Argentina and Chile, and covers 13,000 km2 in one of the stormiest regions in the world. Passive microwave observations from the Advanced Microwave Scanning Radiometer for Earth Observing Systems (AMSR-E) make it possible to monitor the timing of melting and refreezing on the remote and inaccessible SPI. AMSR-E captures both frozen and wet snow surface features because the presence of liquid water dramatically changes the emissivity of snow. Seasonality has been successfully characterized on glaciers in the northern hemisphere by retrieving surface melt, freeze, and transition duration via the Special Sensor Microwave Imager and AMSR-E. Contrasting mass balance controls on glaciers on the eastern and western sides of the SPI motivates this exploration into the spatial variability of melt condition timing. Temporal melt-refreeze patterns are extracted using brightness temperatures (Tb) and diurnal amplitude variations (DAV). Short-term synoptic variability is superimposed on a distinct annual cycle. The most dramatic melt-refreeze intervals occur in the austral fall. Full resolution (12.5 km) AMSR-E makes it possible to distinguish melt regimes throughout the icefield. Here we use histograms of Tb at 19 and 37 GHz and time series data to characterize melt signatures and contrast them with other large icefields. Histograms of Tb exhibit a normal distribution for low elevation pixels (0-1000m) and show that the surface is often above the melt thresholds, resembling low maritime icefields in Alaska. Similar to high elevation Alaskan icefields, medium to high elevation pixels (above 1000m) have a bimodal distribution with significant periods of high DAV due to daytime melt and nighttime refreeze. On transects from west to east over the SPI, diurnal fluctuation signatures are higher in all seasons except austral winter. Detailed investigation of spatiotemporal patterns in surficial melt contributes key observations in support of mass balance calculations for the whole icefield.
C31A-0455
Time Correlation between Runoff and Different Snow Products in the Sierra Nevada
In order to improve runoff forecast for snowmelt-fed streams, it is important to study the effects that changes
in the snowpack have on the streamflow. This work investigates the temporal cross correlation of snow
covered area (SCA), blended snow water equivalent (SWE) and temperature with the full natural flow (FNF)
for three watersheds in the Sierra Nevada during 2006-2008. Fractional SCA is derived through spectral
mixture analysis from daily MODIS data at 500m spatial resolution. SWE is measured daily by snow pillows
that are distributed throughout the basins. These point measurements are spatially interpolated, taking
elevation into account, and multiplied by SCA to create the blended SWE product, which indicates the total
volume of snow per basin. We calculate the cross correlation of SCA, blended SWE and temperature with the
FNF for the American, San Joaquin and Kern River basins. To remove trends from the non-stationary time
series, the data are normalized using moving averages and standard deviations that are adjusted to the time
shifts. The correlations of SCA with FNF and temperature with FNF both show very similar results for all three
basins and they are exclusively positive (i.e. increase in snowpack is followed by increase in flow after a time
interval Δt). While the temperature-FNF correlation shows maxima at a Δt of 4, 24 and 40 days,
the SCA-FNF correlations have their maxima at 0 and 40 days, a small maximum at Δt=20d is
apparent but does not exceed the confidence interval. The cross correlation of the blended SWE with FNF
lacks the 20 day maximum completely, but also exhibits positive correlations at 0 and 40 days.
http://www.snow.ucsb.edu
C31A-0456
Changes in the age structure of Arctic sea ice
Sea ice age is estimated and mapped using ice motion data to follow pixels of ice forward through time while maintaining an age-class distribution function (ADF) within each pixel. During January 1979 – December 2006, ice motion vectors are applied to move the entire area of ice-covered pixels at monthly time-steps for up to 121 months. After each time-step, new ADFs are rebuilt from a superposition of ADFs in cases of ice convergence, or a new age class is introduced to the ADF in cases of ice divergence and new ice formation. Coincident pixel area are used to convert ADFs to estimates of ice extent for each age class. Maintaining internal ADFs preserves methodological precision by circumventing the need to assign a single age class to each pixel after each monthly iteration. Consistent with other studies of ice age based on ice motion data, our results show a dramatic loss of old ice in the Arctic. During January 1989 to December 2006, extent of ice >10 years old decreased from 1.65×106 km2 to 0.063×106 km2. Older ice can only be restored by survival of younger age classes. Ice formed in autumn 1993 and autumn 1995 experienced higher than average long-term survival, causing ice >4 years old to increase in extent during the late 1990s, and ice >10 years old to slow its rate of decline in the mid-2000s.
C31A-0457
Comparing velocities of Juneau Icefield glaciers (1992-2008) from various methods: Differential InSAR, ASTER and SAR pixel tracking, and GPS surveying
We compare different remote sensing methods for measuring the velocities of Juneau Icefield (Alaska) glaciers with ground-based GPS-surveyed point measurements collected by the Juneau Icefield Research program (JIRP). The goals of our study are to: 1) assess how well the different inferences of surface ice velocity compare to each other; and 2) to search for temporal variations in velocity over the entire time period for which data is available (1992-2008) as well as sub-annual changes in velocity. Based on the JIRP GPS surveys of the Taku glacier system, there is no evidence for velocity variations between the seasons or between 1993-2006, although a velocity increase was observed for some stakes in 2007. We seek to use the remote sensing data to determine if this stable velocity pattern at Taku is applicable to the other glaciers which span different physical and climatic conditions in the icefield. Our primary satellite remote sensing tools are differential Interferometric Synthetic Aperture Radar (InSAR) and ASTER (an optical instrument on NASA's Terra satellite) and SAR pixel tracking. We attempt InSAR measurements from different satellites (including Japan's ALOS and JERS, Canada's Radarsat, and Europe's ERS-1/2) including a range of repeat intervals between observations (1-46 days) and both L and C band wavelengths (about 24 and 6 cm, respectively). We combine two SAR images with the known digital elevation model (DEM), using the so-called 2-pass method. We use the Shuttle Radar Topography Mission DEM (generated in 2000), since our tests with ASTER stereo DEM's (2001-2008) and a DEM that we generate from two interferograms from 1995 (the double-difference method) indicate that the changing topography of the Icefield between 1992-2008 has little impact on the derived InSAR velocities over most of the icefield. We test different techniques for doing pixel tracking of ASTER images (phase correlation and cross-correlation) and use the cross-correlation method to do SAR pixel tracking with data from the various satellites. By combining ascending and descending 1-day ERS interferograms and SAR pixel tracking from October 28-29, 1995 and October 29-30, 1995, we invert for the three-dimensional displacement field over the entire icefield. This allows us to directly compare the horizontal velocities measured by the pixel tracking with the inverted InSAR horizontal velocities.
C31A-0458
SNOWCOVER: An approach for continuous daily snow cover Estimation from AVHRR and MODIS Imagery.
The Global Climate Observing System calls for historical and ongoing daily snow cover estimates at 1km resolution. Polar orbiting VNIR sensors such as NOAA AVHRR and MODIS provide sufficient historical coverage and spatial resolution to meet these requirements. However, clouds and sub-pixel variability in land surface properties remain obstacles for continuous daily snow cover estimation. We present a new approach, SNOWCOVER, capable of continuous daily estimates of snow cover using top of atmosphere thermal and visible band measurements from sensors such as NOAA AVHRR and MODIS. The approach includes a new temporal filter to identify clear sky pixels, a pixel based normalization of aquisition geometry and a robust approach, based on radiative transfer modelling, for identifying thresholds for snow discrimination. The algorithm is applied to mapping northern hemisphere snow cover and validated over in-situ snow courses in Canada and the former Soviet Union. Results indicate performance similar to Version 005 MODIS snow cover product but with 95 percent retrieval rates.
C31A-0459
Integrated pan-Arctic melt onset detection from satellite scatterometer and radiometer measurements
Satellite microwave and scatterometer measurements are very sensitive to the presence of liquid water in snow cover, and have been utilized to develop melt onset detection algorithms for various elements of the cryosphere. Given observed and predicted perturbations in the high latitude climate system as a result of global warming, an integrated pan-Arctic and pan-cryosphere melt onset and duration dataset is required to determine intra- and inter-seasonal variability in melt timing. In this study, we combine satellite derived melt onset estimates from unique, previously published, algorithms developed for the land surface, ice caps, large lakes, and sea ice. Backscatter threshold techniques were applied to scatterometer image reconstruction (SIR) QuikSCAT Ku-band backscatter measurements to retrieve melt parameters for terrestrial snow cover, lake ice, ice caps, and Canadian Arctic Archipelago sea ice. Melt onset outside the Archipelago was estimated with Special Sensor Microwave/Imager (SSM/I) passive microwave brightness temperatures. Evaluation of the integrated melt onset product shows the various algorithms perform consistently through the spring period. Some regional issues are apparent, such as conflicts in terrestrial versus ice cap melt onset (early interior versus late marginal onset) on some Eurasian islands because of differences in the melt detection algorithms for land snow and ice caps. Still, the integrated dataset illustrates regional variability in melt anomalies within and between seasons, and between cryospheric elements. Challenges remain in integrating melt duration due to dynamic processes associated with lake and sea ice that are absent on land, and distinguishing seasonal (snow, lake ice, first-year sea ice) from perennial components of the cryosphere (ice caps, multi-year sea ice).
C31A-0460
A MISR-based Surface Roughness Map of the Greenland Ice Sheet
A surface roughness map of the Greenland ice sheet has been constructed from over 400 MISR (Multi-angle Imaging SpectroRadiometer) 670 nm (red band) Cf and Ca ((60.0° forward- and aft-looking) camera images. The use of the red band allows for the maximum spatial resolution (275 m) for MISR, whereas the Ca and Cf camera angles are highly sensitive to surface roughness properties of the ice sheet. The map is a mosaic of normalized difference angular index (NDAI) images acquired during the 2007 sunlit season (April- September). NDAI images have been shown to be a proxy for ice surface roughness in previous work; it is derived by computing a band ratio of MISR's (60° bands. For the northern hemisphere, NDAI is computed by (Ca - Cf)/(Ca + Cf) using red-band radiance or reflectance images. Low values (negative) of NDAI correspond to smoother surface conditions and higher (positive) values correspond to rougher surface conditions. Multiple atmospherically-corrected NDAI images are composited in order to reduce cloud affects over regions where no cloud-free images existed. Images were georeferenced to Equal-Area Scalable Earth Grid (EASE-Grid) projection, and processed and assembled using the ITT ENVI and ESRI ArcGIS software packages. The roughness mosaic reveals detailed features on the ice sheet and associated outlet glaciers. In particular, ice flowlines on outlet glaciers, melt ponds, crevasse regions, and upper glacial regions are clearly distinguishable as distinct roughness zones, suggesting the locations of wet snow, percolation, and dry snow zones. These differentiations allow possible location of the ablation zone versus the accumulation zone, as well as superimposed ice, an important delineation when examining the mass balance status of the ice sheet and a significant factor to consider under current Arctic climate change.
C31A-0461
New Enhancements of an ERS1-2 + ICESat Digital Elevation Model of West Antarctica Using MODIS Imagery, Shapelets, and Kriging
An image enhancement approach is used to develop a new digital elevation
map of West Antarctica, combining multiple MODIS images and both radar
altimetry and ICESat laser altimetry Digital Elevation Model (DEM)
data. The method combines the wide image coverage of MODIS, and its high
radiometric sensitivity (which equates to high sunward slope sensitivity),
with the high precision and accuracy of ICESat and combined ICESat and
radar altimetry DEMs.
ICESat (as of September 2008) has acquired a series of fourteen
near-repeat observational campaigns over the Antarctic during the period
September 2003 to March 2008, covering the continent to 86 deg S. ICESat
data are acquired as a series of spot elevations, averaging a ~60m
diameter surface region every ~172m. However, ICESat track paths have
spacings wide enough (2 km at 85 deg; 20 - 50 km at 75 deg) that some
surface ice dynamical features (e.g. flowlines, undulations, ice rises)
are missed by the track data used to construct the ICESat DEM. Radar
altimetry can provide some of the missing data north of 81.5 deg, but only
to a maximum resolution of about 5 km.
A set of cloud-cleared MODIS band 1 data from both the Aqua and Terra
platforms acquired during the 2003-2004 austral summer, used in generating
the Mosaic of Antarctica (MOA) surface morphology image map, were used for
the image enhancement. Past analyses of the slope-brightness relationship
for MODIS have shown ice surface slope precisions of +/- 0.00015. ICESat
spot elevations have nominal precisions of ~5 cm under ideal conditions,
although thin-cloud effects and mislocation errors can magnify these. Only
cloud-free areas of MODIS scenes that also meet specific slope,
grain-size, solar zenith, brightness, latitude, and elevation criteria are
used for image enhancement.
We calibrate brightness-to-slope relationships for several MODIS images of
the central West Antarctic using a smoothed version of the combined
DEM. Using the calibrations, we then create a slope map of the ice sheet
surface from the image data by regressing slope information from many
images. We then construct an elevation surface from this slope map using a
shaplets-based technique, and compute differences between the constructed
surface and a set of ICESat-derived elevations at specific locations.
Finally, we distibute these elevation differences over the entire surface
using a kriging techinique, and apply this difference surface as a
correction to the computed surface to yield a complete DEM for the region.
Validation of the DEM using airborne laser altimetry datasets is also
presented.
Special thanks to Jonathan Bamber U. Bristol for ERS1-2 + ICESat DEM data
contribution.
http://nsidc.org/
C31A-0462
Thin Sea-ice Region Inferred from Passive Microwave Radiometry
In a previous study we reported the results from concurrent measurements of sea-ice thickness from a ship and passive microwave radiometry from an over-flying aircraft in the Sea of Okhotsk (Naoki et al., JGR, 2008). The results indicate that both the brightness temperature and emissivity increase with thickness within thin ice for a frequency range of 10-37 GHz. The relationship is found to be more pronounced at lower frequencies and at the horizontal polarization. In extending the previous results, here we attempt to estimate sea-ice thickness during the freezing season for the Sea of Okhotsk. The estimation method utilizes brightness temperature of 19 GHz H from AMSR-E along with physical temperature of the sea ice surface estimated from brightness temperature of 6 GHz and ice concentration estimated using both NASA team and bootstrap algorithms. The estimated emissivity of sea-ice was compared with grand-truth data taken from the ship. The difficulty in comparing with ship's data for very thin ice (<10cm) is recognized, which was then remedied by a comparison with visible images from MODIS. These comparisons provide a basis for a classification scheme with three ice thickness categories representing sea ice of approximately <10cm, 10-20cm, and >20cm thickness. Applying this classification to AMSR-E data, daily maps of sea ice thickness for six winter seasons (2002- 2008) were made for the Sea of Okhotsk. The results indicate that the area fraction of the thinnest ice category is about 20 to 40 percent, which tends to increase with time during the early winter. Spatially it occupies the coastal region, in which polynyas are known to exist, and the areas along the ice edge. Time series of the thin ice fraction also shows a significant amount of variability with some jumps, which point to strong influences from the passage of low-pressure systems. Overall, the method seems to work for the SIZ when restricted to the freezing season.
C31A-0463
Characterization of Deep Internal Layers and Basal Conditions Around the WAIS Divide Drill Site by Surface-Based Radar
We used an ultra-wideband, very high frequency (120 to 300 MHz) surface-based radar to simultaneously
map ice thickness, deep internal layers and the ice-bed interface around the West Antarctic Ice Sheet (WAIS)
Divide deep drill site at a fine resolution. The radar was built by the Center for Remote Sensing of Ice
Sheets (CReSIS) as part of the Polar Radar for Ice Sheet Measurements (PRISM) project with the main goal
of developing and testing surface-operated radars to characterize ice thickness and bedrock conditions in
Antarctica and Greenland. The system was fine-tuned in the field to a center frequency of 150 MHz with a
bandwidth of 20 MHz to produce greater sensitivity. The survey covered a 30 km by 8 km area with 1-km line
spacing along a polar stereographic grid that overlapped both the drill site and the WAIS Divide.
The data have been processed for general use and are available on the CReSIS website
(www.cresis.ku.edu). Echograms and digital ice thickness, bed elevation and bed reflectivity maps have been
produced while analysis continues. Our major findings to date include: 1) internal layers are observed
nearly continuously to 2800 m depth, as much as 500 m below the deepest previously mapped layers in this
region, 2) internal layers have been detected to within 350 m of the bed, covering about 90% of the ice
thickness, 3) ice thickness varies between approximately 3100 m and 3550 m over the grid and is about 3500
m at the drill site, 4) basal returns were mapped nearly continuously along grid lines and vary by more than
30 dB, indicating a wet bed at the drill site and frozen conditions elsewhere.
The data will aid rigorous interpretations of the WAIS ice cores (including impurity records and the depth/age
scale) and the morphology and evolution of the WAIS (mean annual accumulation rates, spatial extent, divide
migration and volcanism). Fine-resolution information on deep internal layers, basal conditions and ice
thickness/bed elevation will help refine glacial, sea-level and climate-change histories by constraining both
forward and inverse dynamic ice flow models.
http://www.cresis.ku.edu/research/data
C31A-0464
A new 1 km Digital Elevation Model of the Antarctic Derived From Combined Satellite Radar and Laser Data
Digital elevation models (DEMs) of Antarctica have been derived, previously, from satellite radar altimetry (SRA) and limited terrestrial data. Near the ice sheet margins and in areas of steep relief the SRA data tend to have relatively poor coverage and accuracy. To remedy this and to extend the coverage beyond the latitudinal limit of the SRA missions (81.5° S) we have combined laser altimeter measurements from the Geosciences Laser Altimeter System onboard ICESat with SRA data from the geodetic phase of ERS-1. The former provide decimetre vertical accuracy but poor spatial coverage. The latter have excellent spatial coverage but poorer vertical accuracy. By combining the radar and laser data using an optimal approach we have maximised the vertical accuracy and spatial resolution of the DEM and minimised the number of grid cells with an interpolated elevation estimate. We assessed the optimum resolution for producing a DEM which was found to be 1 km. This resulted in just under 35% of grid cells having an interpolated value. The accuracy of the final DEM was assessed using a suite of independent airborne altimeter data. The RMS error in the new DEM was found to be roughly half that of the best previous 5 km resolution, SRA-derived DEM, with marked improvements in the steeper marginal and mountainous areas and between 81.5 and 86° S. RMS differences varied from 4.84 m over the Siple Coast region of West Antarctica to 29.28 m when compared to a more limited dataset over the Antarctic Peninsula. The airborne data sets were used to produce an error map for the DEM by developing a multiple linear regression model based on the variables known to influence errors in the DEM. Errors were found to correlate highly with surface slope, roughness and density of satellite data points. Errors ranged from typically sim1 m over the ice shelves to about 4-10 m for the majority of the grounded ice sheet. In the steeply sloping margins and mountain ranges the estimated error is several 10's m. Slightly less than 7% of the area covered by the satellite data had an estimated random error greater than 20 m.
C31A-0465
Assessment of EOS Aqua AMSR-E Sea Ice Concentration Using MODIS
Aqua Advanced Microwave Scanning Radiometer Earth Observing System for EOS (AMSR-E) Antarctic sea ice concentrations under winter conditions in 2006 are assessed using ice concentration data derived from Aqua Moderate Resolution Imaging Spectroradiometer (MODIS) radiances. The AMSR-E sea ice concentrations are produced using the NASA Team 2 algorithm at a spatial resolution of 12.5 km, whereas the MODIS ice concentrations are derived from a broad-band albedo using a threshold technique at a spatial resolution of 500 m. There was generally good agreement between AMSR-E/MODIS ice concentrations for areas of first year ice, with the bias staying between -0.3% and -4.5%. For four of the five data sets studied, the vast majority of the AMSR-E pixels was greater than or equal to 80% first year ice, and resulted in a root mean square error (rms) ranging from 3.2% to 9.2% relative to MODIS. The fifth set of sea ice concentration data had about a -25% bias with a 30% rms error. Possible sources of error include unresolved cloud cover, the presence of new ice types, and flooded ice. Further comparative studies are needed to validate AMSR-E sea ice concentrations for other regions and seasons.
C31A-0466
Comparison of snow mapping methods over the Weber River Basin, Utah using MODIS observations
The MODIS instruments on the NASA Terra and Aqua spacecraft provide snow cover observations on a daily basis when cloud cover permits over the globe as well as for regional and local areas. This study reports progress on some work using Terra MODIS snow cover observations centered on the Weber River Basin. The Weber Basin covers 2500 square miles/6400 square kilometers within the Great Salt Lake Basin in Utah. The study is presently focused on comparing various methods for mapping snow cover in this basin using primarily the first six months (January to June) for the 2001, 2002, and 2003 years. The snow-mapping methods compared are the standard SNOWMAP (daily and 8-day) products and the fractional snow cover product provided operationally by MODIS (designated MOD10 A1 and MOD10 A2) on lists of MODIS products, and a spectral end-member approach (ENVI ) employing atmospherically-corrected surface reflectance (the MOD09GHK product). The estimates of snow cover obtained by each of products noted above are compared for each of the 3 years studied. ASTER data are analyzed and used for ground- truth/comparison purposes. Most results are compiled using ENVI/IDL and ArcMAP. The results indicate that the spectral end-member approach consistently obtains lower values of snow cover than the other methods and the results seem to compare more favorably with the higher resolution ASTER snow cover observations. In addition the results show differences in the rate of snow cover depletion in each of the years with 2003 distinctly showing less snow cover than the other two years. The snow cover results as a function of land cover are also being examined. Future efforts are going to be placed on relating temporal snow cover observations to observed runoff statistics to with the ultimate objective of assessing the utility of these satellite observations in runoff prediction.
C31A-0467
Insights Into GLAS Waveforms Using Google Earth
The Geoscience Laser Altimeter System (GLAS) instrument aboard the Ice, Cloud, and land Elevation
(ICESat) satellite was launched on 12 January 2003. The primary objective of the ICESat mission is to
provide global measurements of polar ice sheet elevation to discern changes in ice volume and ice sheet
mass balance over time. Secondary objectives of the mission are to measure sea ice thickness, cloud and
atmospheric properties, land topography, vegetation canopy heights, ocean surface topography, and surface
reflectivity. The GLAS instrument has three lasers, each of which has a 1064 nm laser channel for surface
altimetry and dense cloud heights, and a 532 nm lidar channel for the vertical distribution of clouds and
aerosols. The laser emits a pulse every 0.02 seconds, and receives a return signal. Laser footprints are
roughly 70 meters in diameter and spaced 170 meters apart and are assigned terrestrial positions at the 10s
of meters level of accuracy.
As an aid to data selection we demonstrate how GLAS footprints, waveforms and quality information can be
displayed in Google Earth. We represent the approximate spatial coverage of each laser shot on the
Earth's surface, allowing users to assess the shot in the context of the surface
characteristics gleaned from the underlying image and topography in Google Earth. Each footprint can be
expanded to show the associated waveform, summarizing the detected return signal, along with numerical
values for latitude and longitude, elevation, and date/time. Surface characteristics such as tree canopy, low-
level dust or clouds, snow or ice cover, extreme surface roughness, have significant, easily-visible effects on
the waveform. This application will provide extremely useful information, and will facilitate a detailed data
preview before ordering or processing.
http://nsidc.org/data/icesat/index.html
C31A-0468
Lobate Debris Flows as Permafrost Condition Indicators
Degradation of permafrost is becoming more evident as the climate warms. Observation of critical regions such as infrastructure and buildings on permafrost becomes more important. The Permafrost Laboratory at the Geophysical Institute of the University of Alaksa Fairbanks started a new initiative to actively observe permafrost in critical regions it is called "Permafrost Watch". This proposal is aimed at researching the possibility of using lobate debris flows as a permafrost observation tool. Lobate debris flows are sediment lobes generated in areas with permafrost in eroding mountains. In these areas the sediments are usually part of the permafrost where ground ice cements the sediment together. During periods of warming the ground ice thaws and the sediments start moving down the mountain. The size of these features allows us to observe these features remotely through satellite images. This creates a connection between permafrost conditions and remote sensing that was previously difficult. We propose to research these lobate debris flows in more detail to study their behavior and their potential for describing permafrost conditions through remote sensing. We will be using field based techniques of surveying and thermal monitoring, but also thermal modeling and mechanical modeling to relate movement to permafrost conditions. We will use satellite imagery to measure movement and compare this to ground based measurements. Preliminary data from the Brooks Range of Alaska shows that a particular lobate debris flow there is currently moving at a rate of 1 cm per day, this number is comparable to longer term observations through photogrammetry. Tree ring analysis of spruce trees growing on these features suggests that there is movement in episodes rather than continuous down slope movement. With this study we will be able to expand our valuable borehole temperature data to a much wider area of interest. This study will also contribute to help decide on new borehole locations in the future.
C31A-0469
Movement of Glaciers, Sea Ice, and Ice Shelves in Canisteo Peninsula, West Antarctica Observed by 4-Pass SAR Interferometry
We have extracted a surface deformation map of a part of Canisteo Peninsula on Amundsen Sea in West Antarctica by applying 4-pass differential interferometric SAR (DInSAR) technique to two ERS-1/2 tandem pairs, and analyzed changes of glaciers, sea ice, ice shelves, and their interactions. As there is no digital elevation model of Antarctica with details enough for 2-pass DInSAR, we used 4-pass DInSAR method in which one tandem interferogram, obtained on October 20-21, 1995, was used to estimate the topographic phase (topo-pair) to be subtracted from the other tandem interferogram, obtained on March 9-10, 1996 that contains phases by surface deformation (diff-pair). October is spring in Antarctica where ice begins to melt. The motion of glaciers, sea ice, ice shelves, and their kinematic interactions were imprinted in the diff-pair. Sea ice disappeared completely by the acquisition time of topo-pair in March. We observed fast motion of glaciers pushing the adjoining sea ice. Some interferometric phases indicate the up rise of sea ice of which type is thought to be land-fast ice to exert repulsive force against the pushing glacier. There were other glaciers and sea ice that moved to the same direction, suggesting that the sea ice in these regions was land-fast ice weakly harnessed to sea bottom or pack ice not harnessed at all. Sea ice patches showing no interferometric phases were drift ice that was moving fast by wind or tide. Several small circular fringes on ice shelves suggested that islands or seamounts on the bottom of ice shelves deterred the movement of ice shelves, resulting in the rise of ice surface. So far, we could analyze the mechanism of moving glaciers and the reaction of sea ice, classify sea ice type and the interaction of bottom topography underneath ice shelves. More detailed numerical analysis on the dynamic relationships between glaciers, sea ice, and ice shelves remains as an ongoing research.
C31A-0470
Detection of mesoscale sea ice vortices and their use in sea ice model validation
From 30 March to 5 April 2007 during the Sea-ice Experiment: Dynamic Nature of the Arctic (SEDNA) ice
camp project in the Beaufort Sea, a high spatial resolution (400m) sea ice motion tracking system was used
to detect and track a rotational feature with a ~60 km radius in the sea ice. The ice field at the time of
detection contained only weak discontinuities thereby providing ideal conditions for observing this type of
isotropic rotational motion. The feature propagated NNE at a rate of 10 km/day relative to the drifting ice
which was also moving ~3.5 km/day direction WNW. Around 5 April, changes in wind direction and speed
altered the ice motion field and the break up this rotational signal. We conduct a three-way comparison of
this data with GPS drifters in the field area and with output from an ice-ocean model to ground-truth our
velocity estimates against GPS buoys and test for the existence of similar gyre features in simulations,
respectively. We find relatively good agreement between the buoys and remotely estimated sea ice velocity.
Due to the time span of our velocity estimates based on 3-day repeat SAR imagery, we continue to decipher
whether or not the observed vortices are in fact wind driven effects on the ice-ocean boundary layer, or
whether we have observed the imprint of an ocean eddy on the Beaufort Sea ice cover.
http://research.iarc.uaf.edu/SEDNA/
C31A-0471
Examples of C- and L-Band InSAR to Monitor Snow Water Equivalent (SWE) of Dry Snowpacks
SAR backscatter data from C- and L-band (i.e. longer wavelengths) experience insensitivities in dry snow conditions, whereas shorter wavelengths become increasingly more sensitive to snow metamorphic states. Additionally, the differences in the radar propagation through dry snow and the atmosphere has been related to the interferometric phase shift and, in turn, has been effectively related to changes in SWE. Our previous research demonstrated the use of 3-day repeat C-Band interferometry from crossing orbits of ERS-1 to monitor the dry snow pack conditions over the Kuparuk Watershed, North Slope, Alaska during March 1994. Despite promising results with the 3-day repeat C-Band interferometry, phase ambiguities between satellite acquisitions caused by large snow events may only be overcome on current satellite platforms by using the InSAR technique at longer wavelengths (i.e. L-Band versus C-Band). Therefore, these results have shown additional motivation to explore L-Band PALSAR data from the Japanese ALOS satellite. L-Band InSAR data will be identified and processed in locations (e.g. Kuparuk Watershed) where appropriate satellite data, ground-based measurements, and snow precipitation events overlap. The C- and L-Band interferometric results will be compared highlighting advantages and disadvantages of historic, current, and future satellite platforms.
C31A-0472
Sensitivity Analysis of Repeat Track Estimation Techniques for Detection of Elevation Change in Polar Ice Sheets
Interest in elevation change detection in the polar regions has increased recently due to concern over the potential sea level rise from the melting of the polar ice caps. Repeat track analysis can be used to estimate elevation change rate by fitting elevation data to model parameters. Several aspects of this method have been tested to improve the recovery of the model parameters. Elevation data from ICESat over Antarctica and Greenland from 2003-2007 are used to test several grid sizes and types, such as grids based on latitude and longitude and grids centered on the ICESat reference groundtrack. Different sets of parameters are estimated, some of which include seasonal terms or alternate types of slopes (linear, quadratic, etc.). In addition, the effects of including crossovers and other solution constraints are evaluated. Simulated data are used to infer potential errors due to unmodeled parameters.
C31A-0473
Hydroacoustic monitoring of seafloor earthquake and cryogenic sounds in the Bransfield Strait, Antarctica
To record signals from submarine tectonic activity and ice-generated sound around the Antarctic Peninsula, we have operated an Autonomous Underwater Hydrophone (AUH) array from 2005 to 2007. The objectives of this experiment are to improve detection capability in the study area which is poorly covered by global seismic networks and to reveal characteristics of cryogenic sound which is hard to detect using low-latitude hydrophone array. NEIC has reported ~10-20 earthquakes per year in this region, while the efficiency of sound propagation in the ocean allows detection of greater than two orders of magnitude more earthquakes. A total of 5,160 earthquakes including 12 earthquake swarms are located during the deployment period. A total of 6 earthquake swarms (3,008) occurred in the western part of the Bransfield Strait (WBS), show an epicenter migration of 1-2 km/hr, exhibit a deficiency in high-frequency energy, and occurred near submarine volcanic centers along the back-arc rift axis. Cross-correlation analysis with ocean and solid earth tides indicates the WBS seismicity is modulated by tidal stress, where volcanic earthquake activity reflects variations in tidal forcing than do tectonic earthquakes. On-the-other hand, earthquake swarms from the eastern part of the BS (EBS) show features typical of tectonic earthquakes such as widely distributed epicenters with no clear spatio-temporal pattern and full-spectrum (broadband) signals. These results are consistent with previous crustal models indicating the WBS is undergoing volcanically dominated rifting, whereas rifting in the EBS is tectonically driven. A total of 5,929 ice-generated signals were also derived from the data and are the first detailed observation of various cryogenic phenomena in the region. These cryogenic signals exhibit unusual, tremor-like signals with a high-frequency fundamental (~40 Hz) and 5-6 overtones caused by iceberg resonance, as well as impulsive, short-duration "icequakes" caused by ice break-up and iceberg flow directed along seafloor canyons.
C31A-0474
Towards an Improved Algorithm for Estimating Freeze-Thaw Dates of a High Latitude Lake Using Texture Analysis of SAR Images
Analyzing the freeze-thaw dates of high latitude lakes is an important part of climate change studies. Due to the various advantages provided by the use of SAR images, with respect to remote monitoring of small lakes, SAR image analysis is an obvious choice to estimate lake freeze-thaw dates. An important property of SAR images is its texture. The problem of estimating freeze-thaw dates can be restated as a problem of classifying an annual time series of SAR images based on the presence or absence of ice. We analyzed a few algorithms based on texture to improve the estimation of freeze-thaw dates for small lakes using SAR images. We computed the Gray Level Co-occurrence Matrix (GLCM) for each image and extracted ten different texture features from the GLCM. We used these texture features (namely, Energy, Contrast, Correlation, Homogeneity, Entropy, Autocorrelation, Dissimilarity, Cluster Shade, Cluster Prominence and Maximum Probability as previously used in studies related to the texture analysis of SAR sea ice imagery) as input to a group of classification algorithms to find the most accurate classifier and set of texture features that can help to decide the presence or absence of ice on the lake. The accuracy of the estimated freeze-thaw dates is dependent on the accuracy of the classifier. It is considered highly difficult to differentiate between open water (without wind) and the first day of ice formed on the lake (due to the similar mean backscatter values) causing inaccuracy in calculating the freeze date. Similar inaccuracy in calculating the thaw date arise due to the close backscatter values of water (with wind) and later stages of ice on the lake. Our method is promising but requires further research in improving the accuracy of the classifiers and selecting the input features.
C31A-0475
Evaluating Spatio-Temporal Characteristics of Supra-Glacial Lakes in Western Greenland during the 2007 Melt Season using SPOT and LANDSAT Data
Accelerated ice flow near the equilibrium zone of west-central Greenland Ice Sheet has been attributed to an increase in infiltrated surface melt water as a response to climate warming. Warmer and longer melt seasons, are associated with increase melt duration. Documented short-term drainage of supra-glacial lakes indicates an important dynamic in the infiltration process. A more comprehensive assessment of melt lake density, area, and fill length can shed a light on the sub-surface ice melt transport system. High spatial resolution SPOT imagery (4m2) acquired at the end of the 2007 melt season over the Jakobshavn drainage basin were used to assess late-season melt lake characteristics , while time-lapsed LANDSAT ETM+ (25m2) from June through August, 2007 over the same region were used to track temporal changes in melt lake attributes and drainage frequency.
C31A-0476
Validation of Binary, Fractional and Interpolated Snow Maps at Multiple Resolutions
Mapping snow cover from multispectral sensors began with a simple normalized index using visible and near
infrared wavelengths to classify pixels as either snow covered or snow free, a "binary" classification. Using a
canopy reflectance model and incorporating a vegetation index improved the binary algorithm. Although the
binary snow mapping methods are computationally simple, they are in practice flawed because sensors with
fine spatial resolution usually have a coarse temporal resolution, and vice versa. For sensors with fine
enough temporal resolution to track the dynamic seasonal snow environment, few pixels are either completely
snow covered or completely snow free. Methods to estimate snow cover enable us to determine the fraction
of the pixel covered with snow. Fractional methods include: decision tree classifiers, relationships of snow
cover to snow index developed using regressions with finer-resolution data, and spectral un-mixing. Finally,
daily data can be interpolated to produce a best estimate of snow cover. Here, we compare snow cover
retrievals from binary and fractional snow cover algorithms using various satellites at fine and moderate
resolution: AVHRR (1km), MODIS (500m), Landsat (30m) and ASTER (15m), AVIRIS (2m), and 1m data from
degraded classified imagery. For binary snow cover we use both NDSI and NDSI with vegetation correction.
For fractional snow cover we use a currently implemented operation algorithm MOD10A1 and our own
estimates from MODSCAG spectral un-mixing. For smoothed estimates of snow cover we use another
operational algorithm, MOD10A2 and our own reanalysis of MODSCAG fractional snow cover. The main
study area is the Sierra Nevada of California, along with scenes in the Upper Rio Grande, Colorado Rocky
Mountains and the Annapurna and Khumbal Himal. We find that fractional methods are superior to binary
methods. Moreover, we find that linear spectral un-mixing gives the best estimates of snow cover at moderate
resolution over other fractional products. Finally, space-time interpolation and MODSCAG perform similarly
when MODIS is near nadir, but for off nadir geometry space-time interpolations give the best estimates of
snow cover.
http://www.snow.ucsb.edu
C31A-0477
Designing ICESat-II: Laser waveform simulations, and effects of different multi-beam configurations on change detection.
The ICESat-I laser altimeter has been measuring changes in the polar regions since its launch in 2003. It is expected that the follow-on mission, ICESat-II, will build on its predecessor's legacy, and further the understanding of ice sheet dynamics and mass balance. Laser altimetry waveforms are modified by within- footprint relief due to ground slope and roughness, and land cover type. Laser waveform simulations, including an instrument model and high resolution topography, can be used to determine the impact of a suite of instrument and sampling configurations on the overall measurement capability and recovered parameters, and help answer questions regarding the influence of various designs on the characterization of the ice sheets. The goal is to examine ice sheet and land topography aliasing issues using an instrument model applied to high-res DEMs. Waveform simulations examine the effects of different footprint sizes and multi-footprint arrangements on extracting a surface signal. The benefit of cross-track measurements is investigated, to evaluate their ability to de-couple changes in elevation from slope effects and pointing errors. Single beam configurations with different along track spacing, and other proposed configurations will be examined. We will present a comprehensive evaluation of the influence of the identified key elements on the measurements, which can help determine mission sampling schemes that would allow achieving ICESat-II science goals, in accordance with the guidance provided in the Decadal Survey.
C31A-0478
Changes to North American Snowpacks from 1979-2004 based on Snow Water Equivalent data of SMMR and SSM/I Passive Microwave and related Climatic factors
Changes to the North American (NA) Snowpacks from 1979-2004 based on the Snow Water Equivalent (SWE) values retrieved from the SMMR and SSM/I Passive Microwave data were analyzed using the non- parametric Kendall's test, given that significant decrease in snow cover has been observed in the Northern Hemisphere since the 1980s and drastic increase in the surface temperature of NA has been observed since the 1970s. About 30% of detected decreasing trends in SWE for 1979-2004 are statistically significant at á/2 = 0.05, which is about 3 or more times more frequent than detected increasing trends in SWE. Significant decreasing trends in SWE are more extensive in Canada than in the US, where such decreasing trends are mainly found along the American Rockies. The overall mean trend magnitudes are about -0.4 to - 0.5 mm/year which means an overall reduction of snow depth of about 10 to 13 cm (assuming a snowpack density of 0.1) which can have significant impact to regions relying on spring snowmelt for water supply. The PC1 of NA's SWE are found to be significantly correlated to the Pacific Decadal Oscillation (PDO) index, marginally correlated to the Pacific North American (PNA) pattern, but not much related to El Niño-- Southern Oscillation (ENSO). To assess the possible impact of climatic change to the snowpack of NA, the SWE-air temperature relationships are also analyzed.
C31A-0479
Ice Penetrating Radar Sounding Over Glaciers in Alaska and Greenland
Using low-frequency (1-5 MHz) ice penetrating radar, we have measured the thickness of "warm" ice over outlet glaciers in Alaska and Greenland. The radar mainly consists of control software for National Instrument (NI) boards, a custom-made transmitter, a receiver, and an antenna towed at the back of an airplane. The radar can operate with either a short or a chirp exciter pulse. The same antenna receives echoes bounced from the surface and sub-surface ice layers. The echoes are digitized after being passed through a band-pass filter. We have run the radar in burst mode so that there is no pulse in air while receiving the echoes. To make a radar sounder image, multiple bursts are vertically stacked together in a 2-dimensional format named as echogram. The horizontal axis corresponds to aircraft motion, while the vertical axis corresponds to the arrival time inside a burst. Because the transmitted signal is reflected from various interfaces at different distances, the received echo has multiple peaks. The earliest and strongest peak is caused by the interface between the atmosphere and ice surface. It is very sharp for a flat surface, while becoming diffusive and relatively weak for a rough or sloped surface. After the initial rise, more complex and weak echoes follow. These are caused by various sources such as subsurface deposits, discontinuities in dielectric layers, and, most often, off-nadir surface reflections called surface clutter. We have applied an omega-k method to reduce the along-track surface clutter and thereby enhance the sub-surface features. In this way, we have been able to see 1.5 km deep ice bed at Jakobshavn glacier in Greenland and about 1 km deep glacier bed at Bering Glacier in Alaska. This radar is still in the development and improvement stage, and is expected to continue providing complementary data to existing airborne radar sounders.
C31A-0480
Hemispheric Contrasts in the Trait of Snowmelt on Sea Ice as Indicated Through Microwave Satellite Data
The succession of the different phenomenological stages of snow on sea ice during the melt period and the contribution of each stage to total surface melt is of great interest for sea-ice monitoring studies. In the Arctic, snow disappears quickly once the melt season has started. An early melt stage is characterized by freeze- thaw cycles and marks a short transition between the pre-melt and melt phase. In contrast, the early melt stage with pronounced freeze-thaw cycles is prevailing the surface melt period throughout the whole summer in the Antarctic and persistent snowmelt is very rare. This hemispheric difference impedes a transfer of melt detection methods that were successfully applied in the Arctic. To identify snowmelt on sea ice in the Southern Ocean, we introduce a diurnal-difference melt indicator based on microwave brightness temperatures. This accounts for the characteristic phenomenological stages of snowmelt on Antarctic sea ice and detects the onset of diurnal snow wetting, the early melt stage respectively. Applied to the period from 1988 to 2007, we obtain results that indicate a high regional and interannual variability in distinctiveness and onset of this stage. Applied also to sea ice in the Arctic, where freeze-thaw cycles just played a minor role in previous melt studies, our method provides insight into the hemispheric contrast of the evolution and strength of surface melt during spring and early summer. The investigation shows the intensity of diurnal freeze-thaw cycles in the Antarctic to be one order of magnitude higher than in the Arctic. Moreover, the presented results clearly display that the occurrence of refreezing-events is subject to strong interannual variability in both hemispheres.
C31A-0481
Estimating First Year Sea Ice Surface Roughness from SHEBA RADARSAT Images
The temporal evolution of micro scale surface roughness of first year sea ice in the Beaufort Sea was studied using an empirical model and forward simulation. 15 ice parcels were tracked using RADARSAT images taken during the SHEBA experiment (December 1997 through June 1998). The 40 x 40 km RADARSAT images have a resolution of 50 m. Average backscatter of each tracked parcel was sampled using a seven by seven pixel area. Incidence angle was corrected as described by Stern and Moritz (2000). A backscatter model adapted from Carlstrom and Ulander (1995) was used in forward simulated mode to derive RMS heights. Estimated RMS heights during the winter months were on the order of published values (Paterson et al. 1990; Onstott 1992); however, some values fall below physically reasonable levels. A common feature of the backscatter evolution of the parcels is the presence of a rapid increase in RMS height, typically two to five cm during newly opening leads. This rapid increase is hypothesized to be caused by the formation and decay of frost flowers on newly formed lead ice.
C31A-0482
Lidar Altimeter Mapping of a Supraglacial Lake Region via an Unmanned Aircraft System (UAS) Near Ilulissat, Greenland
The Lidar Altimeter elevation profiles from the Arctic MUSCOX (MUltiSensor Cryospheric Observation eXperiment) project are presented and compared against other techniques for digital elevation mapping of the supraglacial melt lake region in the vicinity of the Jacobshavn Glacier. The hypsography of the sampled region will be discussed, and an estimated volume of a recently drained supraglacial lake will be made. Additionally, a first look at some of the hyperspectral imagery collected over a supraglacial lake will be shown.
C31A-0483
Advanced new daily products of cloud-free snow cover area and snow water equivalent from MODIS/Terra-Aqua & AMSR-E measurements
Taking advantage of high spatial resolution of optical sensor and cloud penetration of passive microwave sensor, an advanced method is presented to combine daily MODIS/Terra-Aqua and AMSR-E (MTAA) products and to generate new daily cloud-free snow cover area (MTAA-SCA) and snow water equivalent (MTAA-SWE) products both in 500 m spatial resolution. This method consists of three major processes: unifying codes, compositing products, and redistributing snow water equivalent. The method was tested and applied to Fairbanks and Upper Susitna Valley, Alaska region for the period of October 2006 to September 2007. The result confirms again that MODIS has high classification accuracy in clear-sky condition and that the daily Terra-Aqua combination reduces cloud blockage by 12.2% and 7.3% respectively for MYD10A1 and MOD10A1 alone. As compared with in situ observations, the snow agreement of the new MTAA-SCA product is 85.6%, which is much higher than the 48.6% of the MODIS Terra-Aqua combined SCA product, and the 75.4% of the MODIS/Terra-AMSR-E combined SCA product. The result shows that the agreement between the AMSR-E SWE and in situ SWE is 54.4% and that the AMSR-E intends to overestimate SWE. Redistribution of SWE, based on sub-pixel analysis of AMSR-E pixels, not only generates the new MTAA- SWE product of 500 m spatial resolution, suitable for basin and regional monitoring and modeling, but also increases the agreement with in situ measurement and reduces the overestimation of SWE. Therefore, the new MTAA products promise a significant addition to the standard MODIS and AMSR-E product series and future NPP and NPOESS products.
C31A-0484
L-Band Polarimetric InSAR Observations of Greenland Ice Sheets using ALOS
The ALOS PALSAR instrument has acquired L-band (23.6 cm wavelength) fully polarimetric synthetic aperture radar (SAR) observations of Greenland with 10 meter single-look resolution. We examine images from a strip in northern Greenland extending from latitudes of 75 degrees N to 80 degrees N, which covers the dry snow, percolation, and wet snow zones of the Greenland ice sheet, as well as the rocky coastal area. Images for repeat-pass interferometry with a 350 meter baseline were acquired at a 46 day interval in March and April 2007. The images from the two dates are coregistered by cross-correlating the HH observations, and we observe fringes in all polarizations in the dry snow, percolation, and wet snow zones, and also in the stable parts of the rocky coastal area. In the dry snow zone of inner Greenland, we observe significantly higher coherence in the HH-HH interferograms (around 0.7) compared to the HV-HV interferograms (around 0.4), and similarly higher coherence in the VV-VV interferogram compared to the VH-VH interferogram. These differences between co-polarized and cross-polarized signals result from volume scattering and lower SNR in the cross-polarized channels. They indicate that scalar models do not fully describe L-band microwave scattering from firn. On each observation date, the phase difference between the HH and the HV returns is almost constant over the dry snow zone of the interior of Greenland. However, there is significant variability in the phase difference between HH and HV returns closer to the coast. The phase difference between the VV and VH returns shows similar behavior, again indicating a difference between co-polarized and cross-polarized scattering mechanisms. We derive polarization signatures for the various scattering regions in the Greenland ice sheets to better understand the scattering mechanisms involved. We model the firn in the dry snow zone as a layered medium with rough interfaces between the layers, and we use the polarimetric InSAR observations to constrain the parameters of this model. Firn structure is directly influenced by accumulation rate in the dry snow zone, and is thus important for understanding the mass balance of the ice sheet.
C31A-0485
Greenland Surface Melt Trends From SSM/I And QuikSCAT Data
We estimated surface melt on the Greenland Ice Sheet using Special Sensor Microwave/Imager (SSM/I) data during 1989-2007 and Quick Scatterometer (QuikSCAT) data during 2000-2007. We applied a multi-scale wavelet transform based edge detection technique to the SSM/I data for melt detection. For QuikSCAT data, we used a threshold based method to compute surface melt. The correlation coefficient between the surface melt areas calculated from these two different sensors and two different algorithms reaches 0.98. Our analysis shows that the temporal variation in the surface melt in Greenland can divided into three sub- periods: 1979-1989, 1990-2000, and 2000-2007. Each sub-period has a distinctive change rate in melt extent, which implies a decadal scale variability in surface melt on the Greenland Ice Sheet.