Close

Service Interruption

NASA Distributed Active Archive Center (DAAC) at NSIDC

ICESat / GLAS Data

Ice, Cloud, and land Elevation / Geoscience Laser Altimeter System

DAAC Home

Frequently Asked Questions


This page contains recent and past questions from users about NASA ICESat/GLAS data products.

All ICESat/GLAS Products

  1. What GLAS data products does NSIDC distribute?
     
  2. What are the different release numbers of GLAS data?
     
  3. In what format is the GLAS data?
     
  4. How often is data collected for a given geographic area?
     
  5. Why is the center point 0.00° latitude and longitude for all data granules?
     
  6. How does GLAS geolocate data and provide off-nadir pointing?
     
  7. How does the Reverb tool select which granules contain data in a specific geographic region?

  8. What are the limits of spatial coverage for GLAS data?
     
  9. How does the GLAS data improve upon past ice sheet altimetry data?
     
  10. What is GRACE and how does it contribute to GLAS?

Atmospheric products

  1. All of the "i_atmQF" values in Release-12 data indicate that the "Flag has not been set -- DO NOT USE." I cannot tell where the clouds are in the data. Is this flag computed in the laser #2 data?
     
  2. What do values of 14 mean in "i_LayHgt_Flag" (GLA08)?
     
  3. Since two different algorithms are used in GLA08 and GLA09, are the aerosol layers different between the two products?
     
  4. GLA08 and GLA09 were originally designed to use only the 532 nm profiles from GLA07; however, these two products use the 1064 nm channel data in addition to 532 nm. Where do the 1064 nm layers come from, and do the 532 nm and 1064 nm algorithms differ?
     
  5. How different are the cloud layer heights in GLA09 identified by the 532 nm and 1064 nm channels?
     
  6. What do values of 14 mean in "i_LRCL_Flag" (GLA09)?
     
  7. In GLA09, is there a cloud bottom for every cloud top? How do you match them?
     
  8. What do the values of i_SolAng represent in GLA07?

Altimetry products

  1. What is the format of the regional mask ancillary files, and what values do they contain?
     
  2. Does GLA14 cover the ice sheets, and if so, does it use a different algorithm over ice sheets?
     
  3. Why is spatial searching not enabled for GLA01 to GLA04?
     
  4. For GLA13, all of the "i_RufSeaIce" and "i_AvgRuf" values are 32767 (invalid). Was the sea ice surface roughness not computed in the first release of data?
     
  5. In preliminary data from laser #1, why does the "i_surfType" record indicate surface classifications that are simultaneously land, sea ice, ocean, and ice sheet?
     
  6. How do you compute sea ice freeboard using GLA13 data?
     
  7. How does the GLAS ellipsoid compare with WGS 84?
     
  8. In GLA13, values of "reflctUncorr" are greater than 1. Is it possible for uncorrected reflectivity to be greater than 1?
     
  9. In GLA13, values of "surfType" all indicate combinations of surface types, with no values for pure sea ice or pure land. Is this correct?
     
  10. In GLA01 waveform plots, what does the "relative time" axis mean?
     
  11. What is the basic difference between a standard and alternate fit and when would you normally choose one over the other?


Answers

All ICESat/GLAS products

  1. What GLAS data products does NSIDC distribute?

    Please see the Data Summaries Web page for a complete list.
     
  2. What are the different release numbers of GLAS data?

    Changes in input data and algorithms result in a new release number, set by the GLAS Software Development Team. Products generated with a new release may be processed forward in time or reprocessed from earlier data. If reprocessed, NSIDC either deletes or hides data granules from previous releases, once they have been replaced by the current release. The first data release was Release-12; there were no additional products with release numbers less than 12. For a complete list of the current data release, see the Current Release Schedule Web page. For a complete list of past data releases, see the Past Release Schedule Web page.
     
  3. In what format is the GLAS data?

    All products are in a flat binary format.
     
  4. How often will data be collected for a given geographic area?

    The first several months of ICESat operations were in an orbit that repeated ground tracks every eight days for calibration and validation purposes. During the subsequent mission phase, ICESat has been in an orbit that repeats ground tracks every 91 days. See the ICESat Reference Orbit Ground Tracks Web page for more information.
     
  5. Why is the center point 0.00° latitude and longitude for all data granules?

    Most data products consist of several global orbits, so there is no real "center point."
     
  6. How does GLAS geolocate data and provide off-nadir pointing?

    Analysis of altimetric data acquired by the GLAS instrument requires accurate determination of the laser spot location on the Earth's surface (ice, land, water, clouds) or geolocation of the laser spot. The spot location with respect to the Earth's center of mass (geocenter) is determined by both the orbital location of GLAS in an appropriate reference frame and the direction of the laser beam described in the same reference frame. With these two position vectors, the location of the laser spot can be inferred in typical geodetic coordinates (geodetic latitude, longitude, and height above a reference ellipsoid) using a Terrestrial Reference Frame whose origin is coincident with the center of mass of the Earth.

    The nominal laser pointing direction is the geodetic nadir (perpendicular to a surface defined by an ellipsoidal model of the Earth), but off-nadir pointing up to 5° is a requirement. ICESat's normal mission consists of a repeat orbit with a 91-day period. A special 8-day repeat orbit was flown for calibration and validation immediately after launch and will be flown as required during its lifespan. For nominal nadir pointing (off-nadir angle less than or equal to an undetermined degree), the 8-day and 91-day repeats are divided into tracks (each track begins at the ascending node) such that one track is a complete revolution around the earth. The orbit altitude is about 600 km, the perigee is fixed in an average sense near the north pole, and the inclination is near 94°. Note: The laser never points directly at nadir because of saturation effects that occurr at nadir, so it points just off nadir where each degree off-nadir corresponds to 10 km on the Earth.
     
  7. How does the Reverb tool select which granules contain data in a specific geographic region?

    The Backtrack Orbit Search Algorithm (BOSA) is used so that Reverb can select which granules contain data in a specific geographic region. The backtrack algorithm exploits the fact that Earth science satellites are in nearly circular orbits. By modeling an orbit as a great circle under which the Earth rotates greatly simplifies the orbit model .  The simplicity of the model allows backtrack to be more efficient than orbit propagator methods, which are designed to work with any satellite.  The simplified orbit model relies on only three parameters: inclination, period, and swath width. 

    As the name implies, Backtrack works by tracing the orbit backwards.  Backtrack starts with the area of interest and answers the question, “In order for the sensor to have seen this area, where must the satellite have crossed the equator?”  There is no time dependence, so the speed of the algorithm is independent of the time range searched.  There is no cumulative error because Backtrack backs up at most one orbit.  There is no performance hit from using a lookup table because Backtrack calculates the actual equatorial crossing range.  The subsequent search is a simple, fast, boolean search on that crossing range (Spatial Search of Orbital Swath Data, Swick and Knowles, 2005).

  8. What are the limits of spatial coverage for GLAS data?

    The orbit inclination chosen for the ICESat/GLAS mission provides coverage between 86° N and 86° S.
     
  9. How does the GLAS data improve upon past ice sheet altimetry data?

    Seasat data collected in 1978 provided scientists with 20 km elevation grids that resolved only the largest surface undulations. Geosat (1985-86) was designed to track better over variable surfaces, with 10 km gridded elevations plotted over the ice sheets. Initial ERS-1 satellite altimetry data sets (with 35-day repeat cycles), collected from April 1994 to March 1995, provided less dense coverage than Geosat, but better spatial coverage of the polar ice sheets - between 81.5°N and 81.5°S. ERS-1 and ERS-2 data (with 168-day repeat cycles) were eventually combined with previous Geosat data to create a more dense gridding of elevation data for the ice sheets.

    The laser altimeter on the GLAS instrument will improve upon this progression by measuring height from the spacecraft to the ice sheet with an intrinsic precision of better than 10 cm with a 60 m surface spot size, thereby providing the capability to measure subdecadal changes in ice sheet thickness of only a few tens of centimeters. Mass balance models will be greatly improved as well as topographic detail of the ice sheets. The accuracy of height determinations over land is yet to be determined, but will be assessed using ground slope and roughness.
     
  10. What is GRACE and how does it contribute to GLAS?

    The Gravity Recovery and Climate Experiment (GRACE) satellite was launched on 17 March 2002. The mission consists of two satellites in low earth orbit, working together to measure the global gravitational field and document gravity anomalies. The earth's gravity field will be measured more accurately and with higher resolution with GRACE, than from any other previous satellite. Temporal variations inferred from GRACE gravity measurements, combined with GLAS altimetry data, provided a better estimate of changes in ice sheet mass balance for Greenland and Antarctica. Potential error sources in measuring mass balance include uncertainties in postglacial rebound, and under sampling and compaction errors. By combining GLAS and GRACE data, the postglacial rebound error is expected to be reduced (Wahr, Wingham, and Bentley 2000). See Improving GLAS data using GRACE for more information.

    Wahr, J., D. Wingham, C. Bentley. 2000. A method of combining ICESat and GRACE satellite data to constrain Antarctic mass balance. Journal of Geophysical Research 105 (B7): 16,279-16,294.

Atmospheric products

  1. All of the "i_atmQF" values in Release-12 data indicate that the "Flag has not been set -- DO NOT USE." I cannot tell where the clouds are in the data. Is this flag computed in the laser #2 data?

    An algorithm has not yet been developed to set this flag. The GLAS science team is currently working on this.
     
  2. What do values of 14 mean in "i_LayHgt_Flag" (GLA08)?

    Values of 1-13 in "i_LayHgt_Flag" describe confidence levels of the quality of the height retrieval; "1" indicating very low confidence and "13" very high. A value of "14" indicates bad PBL height retrieval because of either bad input data or a mistake in the retrieval algorithm.

    The PBL quality flag is computed from the ratio of the average signal (attenuated, calibrated backscatter) within the PBL to the average signal 500 m above the PBL. Normally, the backscatter increases significantly at the top of the PBL and remains higher within the PBL unless it is attenuated by a cloud, extremely dense dust, or smoke; thus, the quality flag is proportional to the magnitude of the gradient of scattering at the PBL top. The larger this gradient, the easier it is to find the PBL top and hence the higher confidence in its detection.
     
  3. Since two different algorithms are used in GLA08 and GLA09, are the aerosol layers different between the two products?

    GLA09 should not contain elevated aerosol layers, only cloud heights. If the GLA09 algorithm detects a layer in the data, a cloud/aerosol discrimination routine is called. If the layer is considered cloud, then it is output to GLA09. If it is aerosol, then it is output to GLA08. The only separate algorithm used in GLA08 is for PBL height retrieval.
     
  4. GLA08 and GLA09 were originally designed to use only the 532 nm profiles from GLA07; however, these two products use the 1064 nm channel data in addition to 532 nm. Where do the 1064 nm layers come from, and do the 532 nm and 1064 nm algorithms differ?

    The GLAS science team added the 1064 nm-based layer detection just a few months before ICESat launched, because they discovered potentially serious problems in the 532 nm channel. The initial design of the software only used the 532 nm channel because of its superior data quality; however, when it became apparent that data would not be available from the 532 nm channel, the science team designed algorithms to use the 1064 nm channel for layer detection, while adhering to the original data structure for GLA08 and GLA09. This explains why these products have both 532 nm and 1064 nm layers in the same array. The 1064 nm layer heights are determined from a separate but similar algorithm to that of the 532 nm channel layer heights.
     
  5. How different are the cloud layer heights in GLA09 identified by the 532 nm and 1064 nm channels?

    Users may note duplication of cloud layers in GLA09, meaning that a given layer is reported twice: once from 532 nm and once from 1064 nm (assuming that both channels are operating). If this occurs, the top and bottom of the layer should be slightly different, because two completely different algorithms are used to locate the layers. In general, the 532 nm and 1064 nm layer top for the same cloud layer will be within 150 m (2 bins) of each other. Most of the time, the 532 nm top height will be slightly higher than the 1064 nm top height. The cloud bottom height will at times be significantly different, especially for geometrically thick clouds. This is because the 1064 nm channel suffers from "signal droop," which causes loss of signal (above and beyond signal loss due to attenuation) as it penetrates down into a cloud. This effect will cause the 1064 nm cloud bottom to be significantly higher than that of 532 nm. Since the droop effect is very dependent on cloud optical and geometric depth, how much higher depends on the characteristics of the cloud itself.
     
  6. What do values of 14 mean in "i_LRCL_Flag" (GLA09)?

    A value of 14 occurs when no ground signal is detected. In this case, it is assumed that the laser pulse became fully attenuated before it penetrated the lowest layer of clouds or aerosol; therefore, the value given for the bottom of the lowest layer is the height where the pulse is extinguished, which is likely different than the actual bottom height of the cloud.
     
  7. In GLA09, is there a cloud bottom for every cloud top? How do you match them?

    For every top of layer given, there is definitely a bottom of layer given in the same corresponding location in the vector. The tops and bottoms vectors are ordered top-down for each channel (532 nm or 1064 nm) independently. Thus, if Cld_top and Cld_bot are the 10-element arrays that hold the top and bottom heights of the cloud layers, then Cld_top(1) is the height of the highest cloud layer and Cld_bot(1) is the height of the bottom of the same cloud layer. Cld_top(2) and Cld_bot(2) are then the top and bottom heights of the next layer down (if there is one detected). A critical fact to remember when using GLA09 is that the vectors holding the tops and bottoms of the layers are used for both 532 nm and 1064 nm detection results. The use flags must be decoded to determine which channel was used to find a particular layer. As an example, suppose that the cloud availability flag (this specifies the total number (532+1064) of cloud layers present) has a value of 5 and the use flags corresponding to the five layers are 0,0,0,1,1. This does not mean that five cloud layers were detected, but rather that Cld_top(1....5) and Cld_bot(1....5) have defined values, and the first three layers (use flag = 0) were obtained from the 532 channel, and the last two layers were obtained from the 1064 nm channel. Further, the layers detected with the 532 nm channel will always precede those detected by the 1064 nm channel (unless of course the 532 channel was inoperative in which case only layers detected by the 1064 nm channel will be present).
     
  8. What do the values of i_SolAng represent in GLA07?

    The variable "i_SolAng" in GLA07 ranges from -90° to 90°. It is zero when the sun is at the horizon (sunrise/sunset), 90 when the sun is directly overhead, and negative when the sun is below the horizon.

Altimetry products

  1. What is the format of the regional mask ancillary files, and what values do they contain?

    The ancillary files "anc27_001_0001.dat" and "anc27_001_0000.dat" are FORTRAN direct-access files.

    The file "anc27_001_0001.dat" has a 1440-byte header record that lists individual regions the mask files used to create the final grid, along with other housekeeping information. The header is in ASCII format, which you can view with the Unix "head" command. Data consist of one 4-byte record for each 1° latitude/longitude cell. The value in the record is a 4-byte integer. If the surface type is constant throughout the cell, the record contains the negative of the surface type (see Surface Type Coding below). If the surface type is not constant throughout the coarse cell, the record contains the starting location in the fine grid file of the subgrid with the surface types for this coarse cell. The location is given as the byte offset from the beginning of the fine grid file. Records start at the North Pole, with 0° - 360° longitude, before stepping to the next latitude band where the sequence is repeated.

    The file "anc27_001_0000.dat" has a 900-byte record followed by one 30x20 byte array. Each cell of this array specifies the surface type for one fine-grid cell within the coarse grid to which the array corresponds (see Surface Type Coding below). The first byte of the array contains the surface type for the northwest corner of the large cell. Location changes most rapidly with longitude.

    Surface Type Coding

    Valid values for the surface type are 1 to 15. The surface types are bit-coded with any combination of surface type allowed. The bit coding is as follows:

    1 = ice sheet
    2 = sea ice
    4 = land
    8 = ocean
     
  2. Does GLA14 cover the ice sheets, and if so, does it use a different algorithm over ice sheets?

    Yes, the land mask (and thus, GLA14) includes ice sheets, although it applies the same algorithm to compute elevation regardless of whether the surface is ice or not. You should use GLA12 for more accurate ice sheet elevations.
     
  3. Why is spatial searching not enabled for GLA01 to GLA04?

    The orbit for GLA01 and GLA02 is a predicted orbit and does not show any target-of-opportunity pointing. The GLAS science team decided not to enable spatial searching for these two products in Reverb because of a potentially large number of false negatives. If you order GLA05 to GLA15 data granules (which support spatial searching) and you require matching GLA01 granules, please note the data times and/or file names from the GLA05 to GLA15 granules. Use these times to perform your search for GLA01 and GLA02 granules.

    GLA03 and GLA04 are engineering and satellite housekeeping data, so spatial searching is not applicable.
     
  4. For GLA13, all of the "i_RufSeaIce" and "i_AvgRuf" values are 32767 (invalid). Was the sea ice surface roughness not computed in the first release of data?

    They are purposely invalid; the GLAS science team may release sea ice surface roughness in the next release of data. More information will be available soon.
     
  5. In preliminary data from laser #1, why does the "i_surfType" record indicate surface classifications that are simultaneously land, sea ice, ocean, and ice sheet?

    This is not an error. The masks for these surfaces purposely overlap in order to not miss any data; the masks are used to determine which data goes in GLA12-15 (the regional products). Most of the ice sheets are considered both ice sheet and land.
     
  6. How do you compute sea ice freeboard using GLA13 data?

    "Freeboard" is a measure of the height of sea ice above the ocean surface. The geoid height field ("i_gdHt") cannot be effectively used as a proxy for sea level in order to estimate freeboard, because current models of the Earth's geoid are too coarse to account for the local variations in geoid height that ICESat/GLAS actually detects.

    A suggested method is to use a high-pass filter to remove the geoid and ocean dynamics effects, essentially filtering out the low-frequency, long-period variability (from geoid and ocean dynamics) and leaving the high-frequency, short-scale variability from sea ice. Some biases still exist, so an offset may be required to make the lowest elevation sea level, while assuming at some point there will be open water. This method is not suggested over a long section of a transect, but it works reasonably well within a given region.

    Kwok, et al. used synthetic aperture radar (SAR) data to find open water and confirm sea level at specific locations, and estimated snow cover.

    Kwok, R., H.J. Zwally, and D. Yi. 2004. ICESat observations of Arctic sea ice: A first look. Geophysical Research Letters 31, L16401, doi:10.1029/2004GL020309.
     
  7. How does the GLAS ellipsoid compare with WGS 84?

    ICESat/GLAS geolocated products are given in terms of geodetic latitude, longitude, and height (elevation) above a reference ellipsoid. ICESat/GLAS uses the same ellipsoid as TOPEX/Poseidon and Jason-1 where the equatorial radius is 6378136.30 m and reciprocal flattening (1/f) is 298.257. Differences between the ellipsoid used by ICESat/GLAS and the WGS 84 ellipsoid are summarized as follows:
     
      ICESat/GLAS WGS 84
    Equatorial radius (a) 6378136.300000 6378137.000000
    Polar radius (b) 6356751.600563 6356752.314245
    Reciprocal flattening (1/f) 298.25700000 298.25722356
    Eccentricity (e) 0.081819221456 0.081819190843

    b = a * (1 - f)
    1/f = 1 / (1 - b/a)
    e = sqrt(2 * f - f * f)

    Thus, the ellipsoid used by ICESat/GLAS is about 70 cm smaller than the WGS 84 ellipsoid. As a consequence, comparison of GLAS elevations to those obtained from other sources must take into account the potential effect of ellipsoid differences. The 3-D coordinates of the same point on the Earth's surface can be described using different ellipsoids, but the coordinates will be different. The dominant difference will be in geodetic height (elevation), with GLAS elevations higher than those obtained using the WGS 84 ellipsoid; however, the differences in geodetic latitude and longitude will produce a horizontal displacement of less than a meter. The horizontal displacement caused by different ellipsoids is well below the GLAS accuracy in horizontal geolocation (Release 14 is 15 m; expected to improve to 5 m), so it can be ignored. The user should ascertain the characteristics of the ellipsoid used with non-GLAS data (GPS, airborne lidar surveys, etc.) before comparing those data to GLAS elevations. The adjustment to elevation to account for different ellipsoids with adequate accuracy is a straightforward function of latitude. Interactive Data Language (IDL) programs to convert between ellipsoids are available via FTP.
     
  8. In GLA13, values of "reflctUncorr" are greater than 1. Is it possible for uncorrected reflectivity to be greater than 1?

    It is possible for reflectivity to exceed 100% from specular reflection. In fact, the Antarctic ice sheet routinely shows 150-200% reflectance. But the echo pulse energy calculation can also show greater than 20% error. The GLAS science team also realized there was a one-shot gain offset in previously released data, but since GLAS was not designed as a radiometer or reflectometer, these issues have been of low priority and will be addressed in the future.
     
  9. In GLA13, values of "surfType" all indicate combinations of surface types, with no values for pure sea ice or pure land. Is this correct?

    Yes, surface type flags are never purely sea ice; ocean is always combined with sea ice. Since GLA13 is a sea ice product, if the sea ice bit is not turned on in "surfType," it will not be in the product. This explains why there are no "pure land" returns. Also, the mask resolution is sufficient to flag many values as multiple surface types.
     
  10. In GLA01 waveform plots, what does the "relative time" axis mean?

    The relative time axis represents two-way travel time. Two adjacent returns separated by a relative time difference (delta_t) in nanoseconds correspond to a height difference of delta_h in centimeters such that:

    delta_h = c / 2 * delta_t

    where:

    c is the speed of light in cm/nsec
    c = (3 x 1010 cm/sec) / (1 x 109 nsec/sec)
    c = 30 cm/nsec

    So, a relative time difference of 1 nsec corresponds to a height difference of 15 cm.
     

  11. What is the basic difference between a standard and alternate fit and when would you normally choose one over the other?

    The standard fit is optimized for "ice sheet-like" returns. The majority of these are single-peaked, narrow waveforms. The standard fit allows for only two peaks. Alternate fitting is meant to capture up to six peaks. More peaks are necessary for more complex waveforms over land, vegetated surfaces and crevassed parts of ice sheets and glaciers. Another key difference between standard and alternate fit concerns the noise floor threshold. Initially, the alternate fitting algorithm used a lower threshold, meant to capture lower-energy peaks than the standard ice sheet waveform. Starting with Laser 2D the thresholds were set lower to account for weaker laser power, resulting in the alternate and standard fits having more similar signal area. Specifically, starting with L2D the standard fit was adjusted from 15 to 9.5, and starting with L2E the alternate fit was adjusted from 3.5 to 7.5. See also: GLAS Altimetry Product Usage Guidance, and GLAS/ICESat L1 and L2 Global Altimetry Data.