TH4.TO4.2.ppt
- 1. Multitemporal, Multisensor Fusion for Monitoring Ice Sheet Changes from Altimetry Bea Csatho, Toni Schenk, Sudhagar Nagarajan and Greg Babonis, University at Buffalo, SUNY, Buffalo, NY, USA
- 2. Motivation: Sea Level Rise and Ice Sheet Mass Balance IGARSS July 2011, Vancouver, Canada The red curve is based on tide gauge measurements . The black curve is the altimetry record (zoomed over the 1993–2009 time span) . (Nicholls and Cazenave, Science 2010) IPCC AR4, A1F1 model Sea level rise estimates, including the effect of ice dynamics
- 3. Ice sheet mass balance estimations from altimetry Surface Elevation Reconstruction And Change detection (SERAC), assumptions and solution Greenland Ice Sheet mass balance 2003-2009 from airborne and spaceborne altimetry Thickening/thinning (dh/dt) estimates of different ice sheet surface types Spatial and temporal evolution of ice sheet dh/dt Overview IGARSS July 2011, Vancouver, Canada
- 4. Ice loss sea level rise Ice sheets exhibit a complex behavior in space and time Monitoring should be synoptic and accurate Estimation of ice sheet mass balance: Mass budget: accumulation – ice loss Geodetic/altimetry method: surface change potentially high special resolution Measurement of gravity to infer mass changes high temporal resolution Temporal and Spatial Evolution of Ice Mass Locked in Polar Ice Sheets IGARSS July 2011, Vancouver, Canada
- 5. Geodetic Method/Altimetry IGARSS July 2011, Vancouver, Canada Mass balance estimated from change of ice sheet surface elevation Level up, mass balance increase Level down, mass balance decrease
- 6. Altimetry: Radar: ERS-1,2 radar altimetry, ESA, 1991- CryoSat-2, interferometric radar altimetry, ESA, 2010- ENVISat, Environmental Satellite, various instrument including radar altimeter, ESA, 2002- Laser: ATM; Airborne topographic Mapper, airborne laser altimetry, NASA, 1993- ICESat, Ice Cloud and land Elevation Satellite, satellite laser altimetry, NASA, 2003-2009 ICESat-2, NASA, 2016 - 2019 Operation IceBridge, ATM and LVIS airborne laser altimetry Airborne and spaceborne stereo imaging, e.g., ASTER, SPOT, aerial photography Ice Sheet Mass Balance Airborne and Satellite Missions IGARSS July 2011, Vancouver, Canada CryoSat
- 7. Goal: surface elevation change from multisensor multitemporal data set Basic principles: Surface shape and temporal behavior is constrained, e.g., by analytical functions Physics of different sensors, e.g., penetration depth, can be additional constrains Solution by least squares Surface Elevation Reconstruction And Change detection (SERAC) IGARSS July 2011, Vancouver, Canada
- 8. Fusion Framework Glaciological parameters and features and their evolution in time: Surface topography Velocity Strain rate Surface mass balance Dynamic thickening/thinning Crevasses, supraglacial lakes, moraines
- 9. Data: Ice, Cloud and Land Elevation Satellite (ICESat) Specifications IGARSS July 2011, Vancouver, Canada Impulse/analog laser altimetry, waveform digitizing Coverage: 86S-86N, 33 day subcycle of a 91 day repeat orbit Mean altitude: 600 km Frequency: 40 Hz Footprint size: 70-100 m Footprint spacing: 170 m Laser wavelength: 1,064 nm for topographic mapping and 532 nm for atmospheric measurements Mission duration: 2003 January – 2009 October
- 10. Altimeter, measures single profiles Footprint size: 70-100 m and spacing: 170 m Distance between groundtrack is typically several km ~30 day operation periods 2-3 times every year since 2003 - giving a total of 18 mission phases with the last phase concluded on Oct 11, 2009. Ice, Cloud and Land Elevation Satellite (ICESat) IGARSS July 2011, Vancouver, Canada
- 11. Given: m sets of n 3D points on bounded surface (surface patch) where each set is observed at different time, in a different location on a changing surface Problem: find changes of surface or reconstruct surfaces at different time epochs Change Detection from Profiler Data: Problem Statement IGARSS July 2011, Vancouver, Canada EGU 2009, Vienna, Austria L2B L2C L3A L3B L3C L3D L3E L3F L3G L3H L3I 500 m
- 12. Surface Elevation Reconstruction And Change detection (SERAC) Assumptions: Ice surface patches (e.g. 0.5 km to 2 km) can be well approximated by analytical functions In general, surface patches undergo no or minimal shape changes over time Surface patches lower or raise without shape deformation Properties Simultaneous adjustment of all points Rigorous error analysis Multi-sensor input data (ICESat, LVIS, ATM,…) 07/28/11 t 0 t 1 t 2 t 3 h 0 h 1 h 2 h 3
- 13. Ice sheet surfaces can be approximated by 2D polynomials on a n 100 m – km scale The fitting error (roughness) depends on surface roughness, size of surface patch, and elevation. Approximation of Surface Patches by Analytical Functions IGARSS July 2011, Vancouver, Canada Polynomial fitting error, 200 meter segments Surface topography from ATM data
- 14. Mathematical Model a 1 …a 5 : shape parameters a 0 : ‘absolute’ elevation
- 15. Adjustment Model
- 16. Result: Surface Elevation Change Curves IGARSS July 2011, Vancouver, Canada Temporal change is approximated by a polynomial function (red curve) Average change rate is estimated from fitted straight line (green line) Change rates at ICESat mission phases are determined from the fitted polynomial by computing it ’s tangent at the middle of each mission phase Data from different sensors are combined into a single solution (blue dots: ICESat; red dots: ATM)
- 17. Examples of h(t) Computation: Accumulation Zone, h=2566 m IGARSS July 2011, Vancouver, Canada average dh/dt= 0.12 m/yr 2003 2004 2005 2006 2007 2008 -0.7 0 0.4 dh(m)
- 18. Examples of dh/dt Computation: Ablation Zone: 1241 m average dh/dt= -2.5 m/yr 2003 2004 2005 2006 2007 2008 -5 5 0 dh(m)
- 19. Examples of dh/dt Computation: Petermann Glacier, Floating Tongue IGARSS July 2011, Vancouver, Canada
- 21. Examples of dh/dt Computation: Petermann Glacier, Floating Tongue 500 m
- 22. Average Thickening/Thinning Rate, North Greenland IGARSS July 2011, Vancouver, Canada Extensive thinning at the onset of Petermann Gl. Thickening bulge at the onset of NE Ice Stream Tributary Recovering surge of Storstrommen Gl. m/yr Thickening bulge on Humboldt Gl.
- 23. Fusing ICESat Satellite and Airborne Topographic Mapper Airborne Laser Altimetry DATA for Change Detection in Greenland IGARSS July 2011, Vancouver, Canada Principles of operation, ATM conical scanner 400 m altitude 250 m swath width ATM laser scanning trajectories 1993-2009
- 24. Locations of SERAC solutions, red: ICESat; blue: ICESat/ATM SERAC solution, h(t) curve blue: ICESat, red:: ATM elevations Point ID: 11002118 RMS surface fitting error all laser points: 0.171 m Elevation: 1197 m Surface change rates are computed as derivatives of the fitted polynomial (red curve) Note the good agreement between airborne and spaceborne altimetry!
- 25. Volume and Mass Change Rates from ICESat-ATM observations, Sept 2003-Oct 2009 Average thickening/thinning rates Sept 2003 – Oct 2009 (L2A-L2F) (m/yr) Average volume change rate, assuming constant surface change rate: 280 ±15 km 3 /yr Average volume change rate, assuming variable surface change rate: 257 ±15 km 3 /yr Mass loss rate from (2): 232 ±20 Gt/yr This study (2)
- 26. IGARSS July 2011, Vancouver, Canada L2A 2003 Fall L3K 2008 Fall m/yr Greenland Ice Sheet thickening/thinning rates m/ m
- 27. Volume Change Rates from ICESat-ATM Observations, Sept 2003-Oct 2009 IGARSS July 2011, Vancouver, Canada (m/yr) Ice sheet mass balance from mass budget method (black) and GRACE (red), Rignot et al., 2011, GRL Average thickening/thinning rates Sept 2003 – Oct 2009 (L2A-L2F)
- 28. IGARSS July 2011, Vancouver, Canada N NW Jak SW NE E SE Volume change rates of major drainage basins
- 29. IGARSS July 2011, Vancouver, Canada L2A 2003 Fall L3K 2008 Fall m/yr Greenland Ice Sheet thickening/thinning rates m/ m
- 30. SERAC, Surface Elevation Reconstruction And Change detection, enables the reconstruction of spatial and temporal evolution of ice sheet dh/dt and volume balance from multisensor, multitemporal elevation data Results from airborne and spaceborne laser altimetry show complex spatial and temporal patterns of ice sheet changes during the last decade at all spatial scales - from crossover points through drainage basin to whole ice sheet volume changes. Concluding Remarks IGARSS July 2011, Vancouver, Canada
- 31. . This work is supported by NASA ’s IceBridge and ICESat-2 Programs and Earth Science Fellowship and NSF’s Office of Polar Programs. We thanks William Krabill and Serdar Manisade for providing ATM data used in this study.
- 32. NW sector, a closer look: Increasing thinning rates, propagating inland on Upernavik Glacier IGARSS July 2011, Vancouver, Canada (from Khan et al, submitted) Crossover C, 1197 m
- 33. dh/dt= - 0.99 m/yr Examples of dh/dt Computation: Petermann Glacier, Floating Tongue 2003 2004 2005 2006 2007 2008 2009 5 0 -5 (m)
- 35. dh/dt= - 0.99 m/yr Examples of dh/dt Computation: Petermann Glacier, Floating Tongue 2003 2004 2005 2006 2007 2008 2009 5 0 -5 (m) dh/dt= + 1.33 m/yr
- 36. IGARSS July 2011, Vancouver, Canada
Editor's Notes
- #3: There is a large uncertainty in the prediction of sea level rise as temperature increases – and probably the largest uncertainty is due to the uncertainty in predicting losses of ice sheets/glaciers due to ice dynamics, for example due to increase in sliding temperature and resulting calving…. Here are some predictions of sea level rise – note the uncertainties from 50 cm to almost 2 meters in the next 100 years.
- #24: A few words about he airborne altimetry. Sketch with aircraft shows the geometry of data acquisition, while the map of Greenland shows the location of point elevation measurements in Greenland.