1. Research and Discover 2003 ICESat Precise Elevation Data Over the Antarctic Megadunes Tom Daigle: NASA/University of New Hampshire Research and Discover Internship Christopher Shuman: NASA GSFC Mark Fahnestock: University of New Hampshire

2. Research and Discover 2003 Background Graduated Union College June 2003 BS in biology and geology Summer 2002 at UNH working with ice cores from NW Canada (R+D)

3. Research and Discover 2003 Summer 2003 Research and Discover Work at NASA Goddard Space Flight Center Evaluate and understand ICESat precise elevation data. Established topographic details of Antarctic Megadune study area. Ice, Cloud, and land Elevation Satellite (ICESat) carrying the Geoscience Laser Altimeter System (GLAS) Launched January 12, 2003

4. Research and Discover 2003 The Next Few Minutes… 1. Get familiar with the dunefield 2. Talk about the exact satellite tracks that cover the dunes 3. How to interpret and assess the quality of the data. 4. Name of the game is Quality Control…where is our good data and how good is it? Image courtesy of Radarsat/NSIDC

5. Research and Discover 2003 What Are These Megadunes Anyway? Broad, subdued features 2-5 km wavelength 2-5 m amplitude Rough, coarse-grained upwind surface, smooth downwind surface ~ 4 km Image courtesy of Mary Albert/CRREL

6. Research and Discover 2003 Satellite Tracks and Cycles ICESat operating on Laser 1 in CAL/VAL mode Each track is repeated in a cycle every 8 days. Each laser (3 total) has a 40 Hz pulse rate and produces an ~70 m footprint with ~170 m separation along track. Vertical resolution: 15 cm surface, 75 m atmosphere. 8-Day repeat ground tracks over Antarctica Graphic courtesy of Matt Beckley

7. Research and Discover 2003 Eight Tracks Over the Dunes Eight tracks cover the dunefield study area. We have 4 to 5 cycles for each track from Feb. 20 to March 29. Must assess the quality of each track/cycle. Terra/MODIS composite image showing the exact paths of the eight tracks of interest. MODIS basemap courtesy of Mark Fahnestock Dunefield basecamp December 2002

8. Research and Discover 2003 Summer Objectives…Quality Control To assess the quality of elevation data from all cycles of each track we will... 1. Check visually the elevation profiles and categorize them as good, okay, or poor. 2. Identify crossover points and their elevation differences. 3. Examined waveforms and gain settings at crossover points using ICESat Science Investigator-led Processing System (I-SIPS) and Visualizer software to better understand how the elevation measurement was made. okay 4. Categorize crossover points as good, okay, poor or bad based on waveforms and compare their elevation offsets.

9. Research and Discover 2003 Track 61 Profile Cycle 2 of Track 61 is good.

10. Research and Discover 2003 Cycle 3 of Track 61 is good in places Track 61 Profile

11. Research and Discover 2003 Cycle 4 of Track 61 is good. Track 61 Profile

12. Research and Discover 2003 Cycle 5 of Track 61 is poor. Track 61 Profile

13. Research and Discover 2003 Crossover Points Tracks 11 and 61 over the entire study area. Zoom in on the exact crossover to resolve each cycle. 6111

14. Research and Discover 2003 Crossover Points 20 unique crossover points for track 11 vs track 61. Assume the two crossover points describe a single, unique elevation despite 10 to 200 meter true horizontal offsets. Safe assumption since the Megadunes are extremely low-slope features (observe ~25 cm elevation offset over 500 m distance). ~2 km 170 m offset at 6_11 vs 5_61

15. Research and Discover 2003 Crossover Tables Crossover points for all tracks are summarized in a table with their corresponding elevation offset. Yellow cells are missing an elevation measure at the exact crossover (nearest point is used). Orange cells have no elevation measure at or near the crossover probably due to atmospheric phenomena. These cells are not included in the waveform analysis since they would introduce a bias.

16. Research and Discover 2003 Waveforms and Quality The laser return-pulse to the satellite is recorded and fit with a gaussian curve to describe elevation. The gain setting is a proxy for the strength of the return pulse, and how gain settings are selected is an issue being investigated by the ICESat science team. The goal is to resolve elevation data from pulses where the gain setting is too high or too low making waveforms saturated or undersaturated (13-25 is best). okayLooked at crossover point waveforms and gains and classified them as good, okay, poor, or bad based on shape and return gain From I-SIPS and Visualizer.

17. Research and Discover 2003 Refined Crossover Quality Index Table of all 306 crossover points and their quality based on waveform properties (color code), and early visual analysis (text). The technical waveform analysis of the preliminary visual categories reveals some tracks that look good are in fact impacted by saturation and consequent elevation uncertainty.

18. Research and Discover 2003 Crossover Quality Histograms Good crossover pairs have the most narrow range in offset, and the smallest average offsets. OkayOkay, poor, and bad crossover pairs have a wider range of offsets and higher average differences.

19. Research and Discover 2003 Average Offset Take absolute value of the offset and get the average deviation from zero. This gives the absolute magnitude of offset and negative offsets cannot be counteracted by positive offsets.

20. Research and Discover 2003 Conclusions Pointing of GLAS laser 1was only accurate to about plus/minus 800 m from theoretical track. Tracks subject to atmospheric phenomenon have a few meters of scatter around the probable surface. Clouds and/or blowing snow are contributing to bad gains and scatter around the surface. For saturated pulses, better processing of original data may provide more accurate and consistent results. Despite some of these errors we still have a better measure of the Antarctic Ice Sheet elevation than ever before.

21. Research and Discover 2003 Future Work on This Project Map actual laser shot-points on MODIS basemap to notice changes in upwind and downwind faces of the dunes. Document any correlations between slope and dune wavelength. Establish a grid of high quality elevation data at crossover and along track locations for a high resolution DEM.

22. Research and Discover 2003 Thank You Chris Shuman Vijay Suchedo Mark Fahnestock UNH/NASA Research and Discover