Description of Data Releases
This page summarizes changes and known limitations associated with
the most current release of ICESat/GLAS data: Release-29 and Release-28.
Note: ICESat/GLAS data is currently being reprocessed to Release-29 for all laser campaigns. During this reprocessing, both Release-29 and Release-28 data will be available. As a new Release-29 campaign is received, the corresponding Release-28 campaign will no longer be available for ordering.
For information about releases prior to the most current one, see the Description of Past Data Releases Web page.
A new release is created when changes occur in the input data or when improvements are made to the processing algorithms. The GLAS Software Development Team determines the release number.
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have been replaced by the new release. The first data release
was Release-12; there were no products with release numbers
lower than that. We recommend you work with the latest release of
data which is the highest release number and includes improvements to
product algorithms. See the following table for more information.
The following table provides links to NSIDC and other Web pages that provide additional information about GLAS data releases.
Release29
This release incorporates significant and extensive changes to the atmosphere processing, additions and corrections to the waveform and elevation processing, and the incorporation of new tide models.
Overview
The format of several products has changed, but those changes were incorporated using unused variables and spares to minimize impact on pre-GSAS 5.4 product code. Significant improvements were made to the data product format documentation, and this release is accompanied with releases of the “Altimetry Data Products User Guide” and “Atmosphere Data Products User Guide”.
Release-29 products have a YXX release number of 428. Refer to the YXX Release Numbers Web page for more information about the release number convention.
Note: Use caution when comparing different versions or releases of GLAS data.
Notable changes to the data in Release-29 are summarized on this page. For more detailed information, see the following
for more information:
Level-1A Products (GLA01, 02, 03, and 04)
Changes/Improvements
- Changed a GPS-time interval limit on GLAS_L0proc to improve the quality and quantity of ANC32 records produced during GPS boundary condition periods.
Known Problems
None.
Altimetry Products (GLA05, 06, 12, 13, 14, and 15)
Additions
- A blowing snow range delay estimate and confidence flag from atmosphere processing has been added to GLA06, GLA12 to 14.
- 40Hz transmit pulse energy was added to GLA05, GLA06 and GLA012 -15.
- Long period (equilibrium) tides are now being applied and are available at 2Hz on the
record
Changes/Improvements
- The ocean tide model was changed from the GOT99.2 model to the TPX07.1 model on GLA06, and 12 to 15.
- The reflectivity estimate is now calibrated for several instrument effects. These values are passed on to GLA06, GLA12 to 15. Previous versions contained errors and were not calibrated.
Bug Fixes
- An error was fixed in the use of the compression index in waveform processing in the computation of received energy.
- The saturation range correction for gain equal to 13 was fixed to a maximum of 2.2m. A software bug in v 5.3 had limited it to 1.5m.
- Solar angle and azimuth is now calculated after a pre-geolocation process in the elevation manager. In earlier versions, it was before the pre-geolocation, and would use passthrough values from GLA05 that were not always updated based on latest pointing
information.
Known Problems
- Reflectivity, not corrected for atmospheric effects, is calculated as Refl = R/T,
where R is the received energy after it has been scaled for range, and T is the
transmitted energy. i_reflctUncorr has also been calibrated for gain non-linearity
(only for non-saturated waveforms), ground truth calibration and boresight shift
shadowing (BSS). It is not corrected for saturation effects. If the shot is saturated
(satindex above 2) then to correct for saturation the reflectivity estimate needs to
be multiplied by the ratio of the corrected energy to the uncorrected energy (sat
corrected reflectivity = i_reflctUncorr * (i_RecNrgAll + i_satNrgCorr)/i_RecNrgAll)
- The atmospheric corrected reflectivity may be calculated from this uncorrected
reflectivity by multiplying it by d_reflCor_atm.
- The saturation energy correction has not be applied to the received energy
before the computation of reflectance. The saturation energy corrections have not been fully verified and need additional investigation. A few of concenrs are:
- If the saturation energy correction is very large compared to the received
energy the data is questionable and often a large correction will cause reflectivity
values greater than 1 to be computed.
- The saturation energy corrections were determined using narrow waveforms
and may not be valid for wide over-land waveforms.
- For about 20% of land data there is saturation for which the current saturation
energy correction does not provide a valid correction.
Atmosphere Products (GLA07, 08, 09, 10, and 11)
There are many changes, additions, and bug fixes incorporated into the version 5.4 GSAS atmospheric data products.
Additions
- Blowing snow has been added to the GLA09 data product. The following are stored at the
5 Hz rate on GLA09: blowing snow layer height, optical depth, range delay and
confidence flag. The latter two parameters are also stored on the GLA11 product at the 1
Hz rate.
- 1064 Total Column Optical Depth (over oceans only at 40 Hz and 1 Hz) has been added
to GLA11. This is based on the measured (from altimetry channel) surface reflectance
over ocean and the computed surface reflectance value (from the Cox-Munk model) as a
function of surface wind speed. The computed (from the surface wind) surface
reflectance value is also stored on GLA11.
- Diurnal Cloud Flag added to GLA09. This new parameter identifies when a given cloud
layer detected at night could have been detected during daytime. A space existed in the
GLA09 data product for this parameter in prior releases, but its value was not computed
until now.
- Browse Products are now available for GLA07-11.
- GLA08 now contains a fairly reliable aerosol layer height product. This provides top and
bottom height of at most 2 aerosol layers detected from the 1064 channel at a 4 second
resolution.
Changes/Improvements
- Improved 532 calibration during daytime – affects laser 2 and 3 GLA07 532 backscatter
profiles
- Changed (improved) calibration for 1064 laser 3 operation periods – affects GLA07 1064
backscatter profiles
- Changed the extinction retrieval method from one that operated only within detected
cloud and aerosol layers to a method that computes extinction for the entire vertical
profile (20 km to the top of the first cloud) – affects GLA10
- Incorporated GMAO (Goddard Modeling and Assimilation Office) GEOS-5 aerosol
transport model calculations of aerosol type as a function of geographic location and time
to improve aerosol optical depth calculation – affects GLA11.
- Improved multiple scattering correction and multiple scattering range delay calculation –
affects GLA10 and GLA11
- Added 532 total column aerosol optical depth and use flag to GLA11 and the bottom
height of cloud free troposphere to GLA10
- No longer use a calculated extinction to backscatter value for cirrus optical depth during
daytime (affects some cloud optical depths on GLA11).
- When the 1064 laser energy is less than 25 mJ, then the following occurs: 1) 40 Hz 1064
cloud search is not executed – the 40 Hz cloud top value is set to 0.0 and the quality flag
set to 15. 2) the 1 and 4 second resolution 1064 cloud layer quality flag is set to one to
indicate lowest quality
- When the 532 laser energy falls below 1.5 mJ, 532 processing is halted. All 532 data
products will be invalid for both day and night.
Bug Fixes
- Fixed 532 calibration error induced by altimetry ocean scans and around the world scans.
- Corrected a bug in the 532 extinction retrieval that caused an incomplete computation of
extinction profiles and an error in optical depth.
- Fixed a bug in cloud/aerosol discrimination that caused a step-decrease in retrieved
aerosol amount poleward of 60 degrees.
Utility Changes
- QA-associated metadata values are now available in the GLAS .MET files. These data
will be available for use by customers selecting data from NSIDC.
- Created a new utility (ANC32_QA) to perform post-L0 processing QA on ANC32 files.
This utility provides improved sanity checking on time calibration during the production
of Level 1 data products.
Known Problems
- Background correction for (some) 532 daytime data is still problematic. After being hit
with a large amount of background light, the 532 detector has a time dependent response
which is a non-linear function of the background level. This will result in the GLA07
attenuated backscatter profiles to be miss-calibrated especially in areas of very bright or
rapidly changing background.
- The 1064 cloud heights (all resolutions, but especially the 40 Hz cloud heights) suffer
from false positives (especially) in daytime for operation periods after L3E. This
becomes very noticeable when the 1064 laser energy falls below 30 mJ. It also affects the
nighttime data as laser energy continues to fall.
Release-28
This release incorporates significant changes to the Level-1B and Level-2A product parameters. There are now improved corrections for saturation effects and improved flags to aid in data selection.
Overview
Note: The saturation elevation correction needs to be added to the elevations by the user, but the correction is not generally recommended for land surfaces (GLA14).
Many parameters were improved with updated algorithms, and some are now being computed for the first time. The data dictionary and flag descriptions were updated as well. A significant effort was made to improve descriptions of the altimetry products. As a result of applying the updated oscillator frequency to the initial time tag estimates, the record index may have changed from the previous release by one or two counts. All Release-28 products have consistent time tags; the sample times have not changed. The surface slope and roughness are output as invalid values for this release.
A new release number designation was adopted to denote accuracy of orbit and laser pointing knowledge which contribute to elevation accuracy. Refer to the YXX Release Numbers Web page for more information. Release-28 products have a YXX release number of 428 in the new designation.
Note: Use caution when comparing different versions or releases of GLAS data.
Notable changes to the data in Release-28 are summarized on this page. For more detailed information, see the following
for more information:
For this release, file naming has changed. Please see the following for
more information about the new file naming convention:
Altimetry Products (GLA05, 06, 12, 13, 14, and 15)
Please read the Notes About Saturation
Elevation Correction below. The changes to altimetry products
include:
- Corrections are now applied for oscillator frequency as determined
by the GLAS engineering team. All elevations are shifted about
2 cm in the horizontal, and the relative elevations between operations
periods shift less than 3 to 8 mm. This change means products
from version 5.3 (Release-28) and higher should not be used in
combination with products from earlier releases in any research.
- The saturation range correction
has been renamed to saturation elevation correction (SatElevCorr)
to reflect the proper sign of the correction. More importantly,
it is provided when it is believed to be correct and is invalid
when a correction is needed but algorithm can not provide a value.
Also the flag (satCorrFlg) has been improved for better data usage
determination. The saturation correction has not been applied
to the elevations, so users should apply this correction when
it is valid. When it is not valid, data should be used with care
as it may be biased. For this release the percent saturation for
the flag to be inconsequential is 2%. This low value was picked
so studies can be made of the low saturation correction. Since
these are small values this may either improve the data or add
a small noise to the results.
- A correction to the received
energy computed from saturated waveforms is provided (satNrgCorr).
It has not been applied to the data.
- The algorithm for setting
the saturation index (satNdx) was changed. The index is still
the number of samples above a threshold but this threshold is
now receiver gain dependent.
- The fit standard deviations (wfFitSDev) may have been computed
from either the raw waveform or a normalized waveform. For this
release, the standard fit uses the raw waveform and the alternate
fit uses the normalized waveform. The maximum amplitude has been
added to most products to allow users to rescale the standard
deviations to the base they want to use.
- The fit
to the transmit pulse changed slightly. This may induce a few
mm change in its location at times and is reflected in the range
and elevations.
- A new parameter (pctSAT) to express the percent
of waveform saturated has been added. The science team is still
evaluating this parameter regarding its usefulness in data quality
selection.
- The waveform quality flag word has changed. A bit
was added to indicate if waveforms are clipped at maximum values.
Also some of the flags were found to be confusing and were removed.
- The alternate fit algorithm has been tuned to provide fewer
peaks in narrow return echoes. These should be more meaningful
to true surface features. Most saturated echoes are now fit with
a single peak. Also it was found that some fits were not keeping
the last peak when they should have been. For this release, the
last peak from the initial estimates is kept during the fit process.
If its amplitude goes to zero it will be deleted from the output.
(In this release, parameters are set so it should never get to
zero amplitude, but still may be insufficient.)
- The sea ice
roughness algorithm has been changed (GLA13 RufSeaIce).
- Some GLA15 algorithms have been updated and additional parameters
have been added. Several previously unfilled parameters are now
filled appropriately.
- The high resolution SRTM DEM has been
updated to use the “finished” data files The averaging
scheme to fill in missing data is no longer being used. Data products
now have more complete coverage when not pointing at the reference
ground track.
- The transmit energy algorithm was updated to
match the GLAS engineering team calibrated values.
- An error
in the Gaussian filter smoothing was fixed. As a result, the first
estimated peak information may be different. This leads to some
initial estimates being different and can result in different
final peak estimates on the products.
- Products should be more
consistent with corrections being invalid if the parameter is
invalid. This may induce changes to editing rules for some users.
Notes About Saturation Elevation Correction
The Release-28 products contain an improved correction
to the surface elevation for saturated returns. The corrections
are not applied to the elevations on the products. Detector saturation,
caused by higher than predicted energy returns, occurred frequently
and is most pervasive in the early operation periods of each laser
(Laser 1, Laser 2A, Laser 2B, Laser 3A, and Laser 3B). The saturation elevation correction
mitigates this problem. Ice sheet crossovers for the Laser 3A and Laser 3B
operation periods show a reduction in the ice sheet crossover standard
deviation from 20 cm down to below 10 cm. The user is encouraged
to apply the correction by adding it to the surface elevation when
the correction value is valid. Unsaturated returns will have a correction
value of zero.
The saturation elevation correction has two known
problems in the Release-28 data.
- The first is that a cutoff of
1.5 m was used for all gain values. The correct cutoff would have
allowed the correction to go as high as 2.25 m for the case when
the gain is 13. In this case, the saturation elevation correction
is set to invalid and the elevations
should be discarded or used with the understanding that a meter-level
error may exist.
- The second possible problem is that determination
of which returns are saturated may have discounted some very slightly
saturated waveforms.
The effect of both of these issues is that
the correction will not be “defined” for all possible
saturated returns. It is still recommended that the correction
be used when valid. These issues will be corrected in an
upcoming release.
Atmospheric Products (GLA07, 08, 09, 10 and 11)
The changes to atmospheric products include:
- Boundary Layer Height (GLA08): The PBL height retrieval over
oceans was modified such that the result can not exceed 3600 m.
Prior versions would produce erroneously high values over ocean
due to clouds. For latitudes less than -65 degrees, the boundary
layer height quality flag (qf) found in i_LayHgt_Flag (byte offset
301), is set to 1, indicating lowest quality. This was done to
flag the PBL height retrievals over Antarctica which are of poor
quality.
- Cloud/Aerosol Discrimination (GLA08 and 09): A problem in
prior releases was the misidentification of an aerosol layer with
an embedded cloud as a cloud layer. The cloud/aerosol discrimination
routine was changed such that it could separate out the two layers
into the aerosol and cloud components. However, the identification
and removal of false positive cloud and aerosol layers was removing
too many valid layers. This was adjusted so as to keep more good
layers while eliminating the invalid layers.
- Aerosol Extinction
and Optical Depth (GLA10 and 11): A change to the optical processing
routines was made so that it would not process the boundary layer
if the quality flag (qf) is set to one. This means that no boundary
layer optical depth or extinction retrievals are made for any
of the measurements below -65 degrees latitude (where the qf is
set to one everywhere). This was done to rectify a problem with
high aerosol optical depth over Antarctica being caused by low
cirrus clouds being mistaken for boundary layer aerosol. In other
areas of the globe where the qf is one (set by the ratio of PBL
average signal to the avgas signal above PBL), this should have
little effect as retrievals with qf=1 as they consist of low
confidence retrievals with weak PBL backscatter (and very low
optical depth).
Known Problems with Release-28
- Altimetry parameter descriptions are still being improved in
the data dictionary.
- As with
prior versions, daytime 532 data from laser 2A and early 2B operating
periods suffer from a background problem related to the detectors.
Essentially, the background is a function of range and it is difficult
to accurately measure and compensate for this effect. Consequently,
the daytime 532 data are sometimes poorly calibrated. This problem
becomes worse as the laser energy decreases in the laser 2B operating
timeframe.
- Clouds that are embedded within aerosol layers will
be detected at the 4 second (low) resolution, but it is possible
that under some conditions they will be missed at the higher resolutions.
This would usually occur with cumulus clouds embedded within a
3-5 km thick aerosol layer in the lower troposphere.
- We are still
investigating the GLA04 IST time tag alignments.
- The science team
created a new estimated surface slope algorithm for flat surfaces.
The new estimated slopes on flat surfaces at low slopes were found
to be larger than expected. The science team is investigating
the reason why the received and transmit signal widths would be
causing this problem. The surface slope and roughness for this
release have been changed to output as invalid values.
- Please read the Notes About
Saturation Elevation Correction in the description of
changes to altimetry products for Release-28.
