TITLE: Greenland Ice Sheet Surface Elevation Changes From ICESat and ATM AUTHOR(s): Beata Csatho 411 Cooke Hall Department of Geology University at Buffalo Buffalo, NY 14260 P: (716) 645 4325 F: (716) 645 3999 E: bcsatho@buffalo.edu Tony Schenk 411 Cooke Hall Department of Geology University at Buffalo Buffalo, NY 14260 P: (716) 645 4325 F: (716) 645 3999 E: afschenk@buffalo.edu FUNDING SOURCE AND GRANT NUMBER: NASA's Polar Program (Grants NNX07AB16G and NNX10AV13G) NSF (Grant OPP-0632310) DATA SET OVERVIEW: The integrated program surface elevation reconstruction and change detection (SERAC) [1] was specifically designed and developed for detecting surface elevation and elevation changes from ice cloud and land elevation satellite ICESat). ICESat carried geoscience laser altimeter system (GLAS) with the primary goal of measuring elevation changes of the polar ice sheets to sufficient accuracy to assess their impact on global sea level. GLAS had three lasers that operated sequentially, with two to three campaigns per year. The footprint size was about 70 m and the point-to-point spacing between neighboring laser points reached 170 m. SERAC copes with different scenarios. Originally developed for calculating surface elevation changes of crossover areas, it was extended to along-track areas and the inclusion of non-ICESat laser data, such as airborne topographic mapper (ATM), an airborne laser scanning system developed by NASA Wallops Flight Facility. The adjustment system of SERAC simultaneously computes the shape of surface patches containing laser points of the same time epoch, estimates surface elevation changes, and approximates the time series of elevation changes by a polynomial after removing the seasonal cycle. The gridded results in this data set demonstrate the potential of SERAC for calculating detailed ice sheet elevation and volume change histories. Greenland Ice Sheet volume changes, calculated from a combined ICESat/ATM data set from mission phases L1A (2/19/2003) to L2F (10/11/2009), show good agreement with previously published results and provide improved sampling in the rapidly thinning coastal regions of southern Greenland. Moreover, the polynomial approximation of the time series of surface elevation changes is taken to advantage in the last example of North West Greenland, illuminating the intricate thinning/thickening patterns that often vary considerably over short spatial scales. INSTRUMENT DESCRIPTION: NASA's ICESat was launched on January 13, 2003, into a near-circular orbit at 94 degrees inclination and approximately 600-km altitude, and decommissioned on August 14, 2010. ICESat carried the geoscience laser altimeter system (GLAS) instrument, designed, built, and tested at NASA Goddard Space Flight Center (GSFC) in Greenbelt, MD. GLAS consisted of three Nd:YAG lasers, operating at a wavelength of 1064 nm (infrared) for determining surface elevation [2]. To measure the vertical distribution of clouds and aerosols, a frequency doubler created a simultaneous beam of 532-nm wavelength. With a 40-Hz pulse frequency and 600-km mean altitude, the laser footprints (laser spots) had approximately 170-m center-to center spacing. The major axis of the elliptical footprint shape ranged from about 50 m to 90 m (laser beam divergence varied between 70 µrad to 120 µrad). For more details on the ICESat mission, the interested reader is referred to the metadata table Attributes for ICESat Laser Operations Periods (http://nsidc.org/data/icesat/laser_op_periods.html). Another notable feature of GLAS was the stellar reference system, comprising the star tracker, a set of gyros, and a CCD camera system, for the precise determination of the orientation of the optical bench, on which GLAS was mounted. This orientation, together with the measured range constitutes the measured range vector of GLAS. GPS data, provided by a JPL BlackJack receiver/antenna, were used to obtain accurate orbit determination [2]. The inclination angle of 94 degrees causes ascending and descending tracks to intersect at about 34 degrees in the center of Greenland. ICESat tracks are not precisely repeated, however. For example, errors in the satellite's attitude control system and cyclic solar array motion (e.g. [2]) caused repeat ground tracks to be up to a few hundred meters apart. The three lasers (L1, L2, L3) operated sequentially over the mission time until the failure of Laser 3 in late 2008 when Laser 2 was reengaged until it failed in October of 2009. The early failure of L1 after 38 days in an 8-day repeat calibration orbit prompted a special review board to adjust the mission scenario: instead of the planned 183-day repeat orbit, a 91-day repeat orbit with a 33 day subcycle was adopted [2]. This resulted in a reduction of the number of ground tracks and the switch from temporally continuous coverage to a discrete one with campaigns 2 to 3 times per year. The spatial separation of reference tracks is a function of latitude and is approximately 30 km in the southern part of Greenland. The airborne topographic mapping system (ATM), an airborne scanning laser altimetry system, was designed and developed by NASA Wallops Flight Facility (WFF). The system, operated by NASA WFF, became operational in the early 1990s and yearly missions in various parts of Greenland and elsewhere have been carried out ever since [3]. Each year, some of the missions are repeated, and some are over new terrain. The ATM system employs a conical scanner, with the mirror rotating about an axis that is at an angle to the mirror surface (nutating mirror), causing an elliptical scan pattern. With flying heights between 500 m and 1500 m, the typical range of the swath width is between 400 m and 1200 m and the footprint varies from 1 m to 3 m [4]. To reduce the raw scan data, planar surface patches are fitted to the original point cloud, typically 3-5 adjacent platelets covering the entire swath across track. The surface parameters, described by the center coordinates, NS and EW slopes, root-mean-square fit of the original ATM laser points, and auxiliary data, called ICESSN, are available from the National Snow and Ice Data Center (NSIDC). An elaborate discussion about accuracy and precision of ATM can be found in [3]. DATA COLLECTION and PROCESSING: We obtained the GLA12 data product (level-2 processed Antarctic and GrIS altimetry data) from the GLAS Science Computing Facility at NASA/GSFC. ATM data (ICESSN), are available from the National Snow and Ice Data Center (NSIDC). Reference 1 contains a full description of all data processing involved in the data set. We refer to a clustered intersections of ground tracks as a crossover area (several ascending and descending tracks), in contrast to a crossover (point) of two single tracks. The intersection angle and the separation of all tracks determine the size of a crossover area (approximately 1 km by 1 km). Each laser point of an individual track is of the same time period, and every new track is on a different surface. That is, for N ascending and descending tracks, we have N surfaces. These surfaces are not completely independent from each other, and we exploit their relationship by assuming that their shape remains constant during ICESat's lifetime; only their elevations may change. Another assumption we introduce is the shape of the surface. Experimental studies indicate that at higher altitudes, the ice sheet is smoother and for large regions, say 1 km by 1 km, can be well approximated by planar surfaces (e.g., [5], [6]). However, our analysis of ATM data indicated that quadratic or higher order surfaces are needed to model this km scale topography at lower and generally rougher elevations. Therefore, SERAC uses up to a third-order polynomial equation, but the degree can be specified by the user. Every laser point of the same track yields a third-order polynomial equation. Points of a new track, at a different time period, give a very similar equation, except for the absolute elevation. For a real crossover area, we have five to six points per track and a maximum of 2 * 19 = 38 ground tracks (19 laser campaigns) resulting in up to 200 equations for solving 38 absolute elevations and 9 shape parameters, thus providing a large redundancy for reconstructing the shape and the changing absolute (centroid) elevations of a surface patch. Surface patches typically cover regions centered at the intersection of reference ground tracks. The patches are large enough to cover tracks from all ICESat operational periods. They may or may not include data from the first two campaigns (L1A and L1B) as they were collected in a different, 8-day repeat calibration track pattern. We solve this redundant problem by least squares (e.g., [7]), using a Gauss-Markov model We use a very simple stochastic model, assuming that all laser points are uncorrelated and of the same precision. Of course, the error model of ICESat laser points is more complex. Apart from random errors, there are also systematic errors (biases) present, for example caused by atmospheric effects (clouds, ground fog, blowing snow). The proposed least-squares approach takes all laser points of all missions in a crossover area and estimates the parameters of a third-order polynomial surface. Because of the large redundancy, the estimated parameters are not significantly affected by the simplified error model, however. In this fitting process, the parameters (absolute elevations and shape parameters) are determined such that that the square of the distance between the observed laser points and the approximated surface becomes a minimum. The least-squares approach takes as an input laser points and transforms these observations into elevation change information. Note that all observations (laser points) of all missions are used. Outliers, or blunders, caused by ranging errors due to saturated signals or weak returns in the presence of clouds or ground fog, are detected and removed automatically. Biases between different ICESat missions are found and eliminated during the fitting of a polynomial through the surface elevations. All results in this data set are derived from actual ICESat observations, using GLA12 data product from the GLAS Science Computing Facility at NASA/GSFC. Elevations refer to the WGS-84 ellipsoid, and we also applied the saturation correction provided on the GLA12 product. The ICESat elevations are corrected to ocean tides and load tides using the global tide model GOT99.2 (GOT = Goddard/Grenoble Ocean Tide)[8]. Every crossover area begins and ends with a unique date. Therefore, we need to adjust the elevation to a common reference time. We have chosen August 31, 2006 as the reference time, because for most crossover data this date is within its observation period and no extrapolation is necessary. DATA FORMAT: All data sets are in a .zip archived format. All gridded data are in NetCDF format, indicated by .cdf file extensions. In addition to gridded data, we have included the elevation change rates at the ICESat ground track crossovers that were used for generating the grids (greenland_ice_sheet_dhdt_icesat_atm.csv). The point data are in csv (comma separated value) ascii format and have the following structure: Column 1 (xover): Unique numeric identifier for each crossover Column 2 (x in meter): X location of each crossover in meter, UTM 24N Projection, WGS-84 Ellipsoid Column 3 (y in meter): Y location of each crossover in meter, UTM 24N Projection, WGS-84 Ellipsoid Column 4 (z in meter): Z Elevation of the centroid of the crossover in meter, ellipsoidal height, WGS-84 ellipsoid Column 5 (l_cover): Landcover type classification based on [9]: 4 = Ice Sheet, 3 = Ice caps, 2 = Land, 1 = Ocean Column 6 (chg_rate1): Average elevation change rate in m/year during the full observation period at the crossover, i.e., between the first and last valid ICESat measurements. It is the slope of the straight line fitted to all dh(t) values Column 7 (chg_rate2): Same as Column 6 except that the straight line fitting to the dh(t) values was performed with blunder detection/removal Columns 8-26 (L1A to L2F): Elevation change rates in m/year during the 19 individual ICESat operational phases. These values are computed as the tangents to the polynomial fitted to all dh(t) values, at the midterm of each operational periods (see below). The polynomial is extrapolated by 8 months at both ends of the observation period. Mission dates outside that period have no valid elevation change rates. Midterm of ICESat operational periods (ddmmyy): L1A 30503 L1B 32503 L2A 100703 L2B 30504 L2C 60504 L3A 101904 L3B 30705 L3C 60605 L3D 110705 L3E 31106 L3F 60906 L3G 111006 L3H 32907 L3I 101907 L3J 30508 L2D 120608 L2E 31509 L2F 101509 DATA REMARKS: An Unarchiving tool will allow you access to the full datasets (WinZip, 7-Zip, etc...) A NetCDF viewer will allow you access to gridded data (Panoply, IDV, ncview, etc...) A text editor will allow you access to the csv point data file (Textpad, MS Excel, etc...) Despite of the various checks, there might still be crossovers containing blunders, i.e., anomalous elevation changes for certain ICESat operational periods. REFERENCES: 1 Schenk, T. and Csatho, B. "A New Methodology for Detecting Ice Sheet Surface Elevation Changes From Laser Altimetry Data," IEEE Trans. Geosci. Rem. Sens., In-Press. 10.1109/TGRS.2011.2182357 2 B. E. Schutz, H. J. Zwally, C. A. Schuman, D. Hancock, and J. P. Di Marzio, "Overview of the ICESat Mission," Geophys. Res. Lett., vol. 32, no. 21, p. L21S01, 2005. doi:10.1029/2005GL024009. 3 W. B. Krabill, W. Abdalati, E. B. Frederick, S. S. Manizade, C. F. Martin, J. G. Sonntag, R. N. Swift, R. H. Thomas, and J. G. Yungel, "Aircraft laser altimetry measurements of elevation changes of the Greenland ice sheet: Technique and accuracy assessment," J. Geodyn., vol. 34, no. 3/4, pp. 357-376, Oct./Nov. 2002. 4 R. Thomas, E. Frederick, W. Krabill, S. Manizade, and C. Martin, "Recent changes on Greenland outlet glaciers," J. Glaciol., vol. 55, no. 189, pp. 147-162, Feb. 2009. 5 C. J. van der Veen, Y. Ahn, B. Csatho, E. Mosley-Thompson, and W. B. Krabill, "Surface roughness over the northern half of the Greenland Ice Sheet from airborne laser altimetry," J. Geophys. Res., vol. 114, p. F01 001, Jan. 2009. doi:10.1029/2008JF001067. 6 C. J. van der Veen, W. B. Krabill, B. Csatho, and J. F. Bolzman, "Surface roughness on the Greenland ice sheet from airborne laser altimetry," 1121 Geophys. Res. Lett., vol. 25, no. 20, pp. 3887-3890, Oct. 1998. 7 K. R. Koch, Parameter Estimation and Hypothesis Testing in Linear Models. Berlin, Germany: Springer-Verlag, 1999 8 Goddard Space Flight Center, 58 R. D. Ray, "A global ocean tide model from Topex/Poseidon altimetry: GOT99.2," NASA Tech. Memo. 209478, p. 91999, 58. 9 Bamber, J.L., S. Ekholm, and W.B. Krabill, "A new, high-resolution digital elevation model of Greenland fully validated with airborne laser data", Journal of Geophysical Research, J. Geophys. Res. 106, B4, 6733-6745, 2000.