|
||||||
|
|
||||||
This page summarizes changes and known limitations associated with past releases of ICESat/GLAS data. For information about the most current release, see the Description of Current Data Release Web page.
The first data release was Release-12; there were no products with release numbers lower than 12. We recommend you work with the latest release of data which is the highest release number and includes improvements to product algorithms.
Click on a release number below to view a summary of its release notes.
Release-26
Release-24
Release-22
Release-21
Release-19
Release-18
Release-17 Atmospheric Products (GLA02, 07-09)
Release-14 Land Products (GLA14)
Release-13 and -14 Period Data (Oct./Nov. 2003)
Release-12 Cal/Val Period Data
The data dictionaries and flag descriptions were significantly updated for this release. There were no L1A changes.
Notable changes are summarized here. See a list of product format and definition changes associated with Release-26. Please refer to the NASA Wallops Flight Facility's GSAS v5.2 Release Notes for detailed release notes for each product in Release-26.
The changes to altimetry products are:
Release-26 incorporates two significant additions for GLAS 532 nm data products. First, there are improvements to parameter tables used to calculate the aerosol optical depth and cloud multiple scattering factors; these improvements considerably increase accuracy. Second, Release-26 is the first to include data products from the Laser 2B (February 2004) operating period.
Other issues for Release-26 are.
This is a major release. Most changes affect the Level-1B and Level-2 products. Only minor changes were made in Level-1A processing (correction made to the EU conversion of GLA04 IST i_VTBoreH and i_VTBoreV). The following products were updated to Release-24: GLA01 - GLA15.
Notable changes are summarized here. See a list of product format and definition changes associated with Release-24. Please refer to the NASA Wallops Flight Facility's GSAS v5.0 Release Notes for detailed release notes for each product in Release-24.
The key change made to the altimetry products is the lengthening of all of the Level-1B and Level-2 products allowing space for new variables and additional growth. These new variables include:
Changes have been made to both the transmitted and received energy calculations, as suggested by the instrument team. For details of the energy computation, see the waveform ATBD.
Two minor changes have been made to the standard parameterization waveform fitting. The standard fit now uses the absolute peak location change (instead of a percentage change) as the convergence criteria in the Gaussian fitting. An error in the computation of the standard fit sigma has also been corrected. Neither of these changes has significantly affected the elevations on the data.
In the alternate waveform fitting, an option to normalize the waveform based on the peak amplitude was implemented. An error in reporting the alternate fit standard deviation was corrected and the precision was changed on the product to accommodate the smaller numbers. The alternate fit parameters were changed to perform normalization by peak amplitude before fitting, the minimum distance between selected peaks was set to be greater than 15 ns, and the weight of the sigma widths was tuned to provide the best alternate fits on selected land waveforms.
Through I-SIPS Release-23, cloud detection using the GLAS 1064 nm atmospheric channel was limited to 4 and 1 second resolution. This was due mainly to the low inherent signal-to-noise of the 1064 data. Over approximately the last 6 months, we have been striving to produce 1064 cloud top height on a shot-to-shot basis (40 Hz). The new algorithms have been tested and are included in this release of GSAS. The limited amount of testing done to date has shown that we can definitely detect clouds at this resolution, though the exact limits of what can be detected (in terms of optical depth) have not yet been quantified. An assessment will be performed after the laser 2A data has been reprocessed using this release of the GSAS software, by comparing the 1064 nm cloud detection with that of the 532 nm channel. A part of the cloud detection algorithm is based on the integrated 1064 nm signal, which is also stored as a separate quantity on the Release-24 GLA09 product. Testing has shown that when the magnitude of the integrated signal exceeds a certain threshold, it is highly likely (as corroborated by the 532 channel) that there is a cloud present even if the threshold algorithm did not detect a cloud. In this case, the cloud height is set to 10 km and a flag is set to indicate this fact. The algorithm also interrogates the ground return signal width in an effort to detect very low clouds. Over flat terrain such as ice sheets, the 1064 ground return signal has a characteristic width. When this signal is unusually wide, and clouds have not been detected higher up, the cloud top height is set to 100 m and a flag is set to indicate the likely presence of low clouds. While these approaches improve the overall cloud detection, they also have their limits. Extensive testing is needed to determine those limits.
The following products were updated to Release-22: GLA01, GLA02, GLA03, GLA04, GLA05, GLA07, GLA08, and GLA09. This release contains significant improvements to the waveform alternate fitting process, to energy/reflectance calculations, and to atmospheric products. Notable changes are as follows:
The energy calculation improved for GLA01. For GLA05, energy/reflectance calculations improved and a saturation index was added. The alternate fit uses the normalized waveform based on area before functional fit. (Standard fit uses raw voltage waveform.) The alternate fit now also uses up to five largest peaks plus the peak nearest the ground. The algorithm keeps all six peaks, but amplitudes are allowed to go to zero. The alternate fit least-squares exit criteria changed; maximum iterations are now significantly less frequent. If the maximum iteration flag is set, data should not be used (or used with caution). See a list of product format and definition changes associated with release-22.
The 532 nm cloud layer detection improved. The 40 Hz cloud search is now executed for every shot, independent of the results of cloud searches at lower resolutions. Also, the starting height for the cloud search is now 10 km, which means that clouds are now reported when they are found up to 10 km altitude. (In prior releases, 4 km was the maximum altitude). These changes mean that a cloud could be reported for one or more 40 Hz shots of a given second, while no cloud was reported at 1 second or 5 Hz resolution.
The major problems with these data pertain to the 532 channel only. The 1064 backscatter data (GLA07) appear well-calibrated and the 1064 cloud heights are robust both day and night.
The 532 channel had only 5 mJ of energy and a fractured far-field laser beam pattern that reduced the signal strength considerably. As a result, the calibration for daytime 532 data is often poor. The calibration of nighttime data is generally acceptable. The 532 channel signal levels are very low, which severely affects the daytime measurements.
Because of the poor 532 channel signal, elevated aerosol and PBL height data (GLA08) will be unusable during day and only marginal at night. Thicker aerosol and PBL layers can still be detected, but only at night. Cloud height data (GLA09) are good during night, but only thick clouds can be detected during the day.
Optical depth and extinction data (GLA10 and GLA11) are not included in this release but attempts will be made to release nighttime only data in the future.
This release contains the following improvements to the elevation products only (GLA06, GLA12-15). No product format changes were implemented.
Known problems are the same as Release-19, plus flagging does not correctly indicate when GLA05 pass-through PAD is used in place of data from ANC09.
Notable improvements include the following:
Please refer to the NASA Wallops Flight Facility's GSAS v4.1 Release Notes for detailed release notes for each product in Release-19.
Improvements include the following:
ICESat/GLAS Release-18 and earlier data products error in received energy computation:
The 40 Hz received gain values i_gainSet1064 reported on GLA01, and i_gval_rcv reported on GLA05, are offset from the shot to which they actually pertain. The gain used onboard for shot N+1 is contained in the telemetry for shot N. Ground processing should therefore use the gain reported with for shot N in the energy computations for shot N+1.
For Release-18 and earlier ICESat/GLAS data products, this shift was not made; therefore, all parameters that used the gain on GLA01, 5, 6, 12, 13, 14, and 15 have some error. The energy parameters are in error by an amount proportional to the shot-to-shot change in the received gain setting. No error is induced in the elevation measurements from this wrong application of the gain setting.
Parameters that are known to be affected are:
i_RecNrgAll_EU
i_RecNrgLast_EU
i_reflctUncorr
i_reflctuncmxpk
Gain is also used to determine the waveform saturations flags:
i_WFqual (gwi_satFlagLo)
i_WFqual (gwi_satFlagHi)
i_WFqual (gwi_satFlagHiF).
The gain on GLA01 will be shifted to the correct shot in a future release of the data. This will fix all the problems in the ground processing and correct the parameters on the data products.
GLA08 and GLA09: Cloud aerosol discrimination usually results in layers misidentified as aerosol when they should in fact be cloud. The problem is most severe in the high-latitude regions and over the poles, and in regions of multiple cloud layers where the signal attenuation from the upper layer makes the lower layer "look" like aerosol.
GLA07: Background computation problem in areas of very bright light cases signal to be poorly calibrated in some instances where the background is high.
Please refer to the NASA Wallops Flight Facility's GSAS v4.0 Release Notes for detailed release notes for each product in Release-18.
Release-17 data include protection for floating point exceptions, corrections to science software documentation, and improved Quality Assurance Products (QAPs) and browse products. Release-17 products include data from GLAS laser #2. Atmospheric products now have substantially improved PBL height retrieval (GLA08), improved cloud layer detection from 1064 channel, and a new time-dependent calibration for 1064 nm.
The GLAS data product Release-17 is the initial production run of data products for the GLAS 532 nm atmospheric channel. Preliminary data quality and validation activities have been preformed, but additional testing and improvements will be incorporated in further releases.
GLAS is a two channel atmospheric lidar and was operated for two periods in 2003. The instrument was run with only the less sensitive 1064nm channel from February 21 to March 29. GLAS was first fully turned on at the end of September. There is data from both 1064 and 532 nm channels from Sept. 25 to Nov. 18. From Sept. 25 to 28 the filter and alignment for the 532 nm channel were not yet set, and there were also subsequent short periods where the 532 nm data quality was variable when the alignments were reset. For most of the fall period of operation, we obtained very good data.
The GLAS instrument was launched with a number of hardware level problems that effect atmospheric data quality. These include some signal non-linearity and variation of offset under certain conditions. The large majority of errors have been corrected or mitigated in the data processing, but some known issues require further attention.
GLA02 Lidar Signal
The 1064 nm detector circuits introduce “signal droop” resulting in too low values following strong signals. The effect is corrected in the GLA02 software for low and medium strength signals, but due to saturation and other effects, is not fully corrected for strong return signals.
GLA07 Calibrated Observed (Attenuated) Backscatter Cross Section
The 532 backscatter cross section calibration is very good at night and with daytime low to medium backgrounds. At higher background levels (> 10 photons/bin), the detector response varies somewhat with range, causing problems in the background subtraction. which in turn causes the 532 profile to not follow Rayleigh with high accuracy. The problem is estimated to be in about 10% of the daytime data. The 1064 calibration must be based on comparison to the 532 nm data where signals overlap and ground data intercomparisons and is not as accurate as the 532 nm calibration. The 1064 nm calibration in release 17 are now thought to give backscatter values about 5-10% low (calibration constant about 5-10% high), based on further analysis and results from 532 calibration transfers.
GLA08 Planetary Boundary Layer Heights
Planetary boundary layer heights are in general very good. Some problems are:
GLA09 Cloud/Aerosol Heights Including Multi Layers
GLA10 Extinction and Attenuation Corrected Backscatter Cross Sections
GLA11 Thin Cloud and Aerosol Optical Depth
Due to a coding error in the latitudes and longitude which was found after release 17 processing, these products will not be available until Release-18.
Please refer to the NASA Wallops Flight Facility's GSAS v3.9 Release Notes for detailed release notes for each product in Release-17.
Release-13 and Release-14 use data from GLAS laser #2, and include improved waveform fitting and precision-pointing data for altimetry products (GLA05, GLA06, GLA12-15).
Release-13 (GLA01) and Release-14 (GLA05, GLA06, GLA12-15) elevation data products covering October/November 2003 have not been fully calibrated. The fundamental elevation product contains: time tag, geodetic latitude, longitude and height (elevation) above the TOPEX reference ellipsoid. These products result from a combination of round-trip laser pulse travel time, precision orbit determination (POD), and precision attitude determination (PAD). Contributions from various sources lead to the following absolute accuracy assessment of the fundamental elevation products.
In some occasional cases, off-nadir pointing was used at some locations, which introduces an apparent surface slope of up to 5 degrees, depending on the angle used.
ICESat products for land areas (all continents including Antarctica, and ocean islands, including Greenland) are distributed in the GLA14 data product. The GLA14 results are produced using processing results designated as alternate in the GLA05 waveform processing data product, in contrast to the standard processing results used to create the GLA12 ice sheet products. ICESat products have been produced in a number of releases, with improving data acquisition, calibration, and processing techniques. Data covering an 8-day period in March, 2003 were distributed as Release 12. Data covering a 33-day period in October and November, 2003 are distributed as Release 14.
The primary difference between the Release 12 and Release 14 versions of the GLA14 land products is a substantially reduced frequency of saturated returns, achieved by modifying the onboard gain algorithm to better accommodate rapid changes in received signal amplitude caused by changes in atmospheric transmissivity and surface reflectance. The modified gain algorithm was implemented for the fall, 2003 operation period.
An additional difference between the Release 12 and 14 data sets is the threshold level used in ground processing to detect the presence of a return signal. A threshold 5 times the standard deviation of the background noise level was used in Release 12. This threshold detected some weak returns from clouds that were thus included in the GLA14 product. To reduce the number of cloud false-alarm returns, a higher threshold, 7.5 times the background noise standard deviation, was used in Release 14. However, the higher threshold also causes some weak returns from the land surface beneath clouds to be excluded from the GLA14 product.
In addition to detecting the presence of returns, the threshold is used to define waveform extent based on the start and end of the signal corresponding to the first and last crossing of the threshold. The higher 7.5x threshold used in Release 14 underestimates waveform extent as compared to the 5x threshold used in Release 12. The Release 12 designations of signal start and end are a more accurate measure of the highest and lowest detected surfaces within the laser footprint. Furthermore, the elevation reported in the GLA14 product corresponds to the centroid of the received signal between signal start and end. The elevation is therefore sensitive to the threshold used.
Because of the above considerations, future releases of the GLA14 products will revert to the lower 5x threshold in order to detect weak surface returns and more accurately define waveform extent.
Future releases of GLA14 products will include several additional improvements. Fitting of Gaussian distributions to waveform peaks will be refined to more fully characterize multi-peaked distributions. Waveform fitting results in Release 14 are known to have some deficiencies and should be used only as a provisional product. Additionally, Releases 12 and 14 of GLA14 include the location of the signal centroid and end, but not signal start, as defined by the threshold above noise. The signal start currently included in GLA14 is based on a different threshold defined as a percentage of the maximum signal amplitude. This signal start will be replaced by the noise threshold signal start in future releases, so as to be compatible with the centroid and signal end designations. The position of the noise threshold signal start is available in the Release 12 and 14 data sets in the GLA05 waveform processing product.
Finally, for some returns in the data acquired during 2003 the signal extent exceeds the maximum width of the waveform and the upper end of signal is truncated. This occurs for footprint locations with large relief caused by steep topographic slopes and/or tall forest canopies. In mountainous regions, approximately 10% of the waveforms are truncated in this way. During 2003 operations the sampled waveform covers an extent of 81.6 m (544 waveform bins x 0.15 m per bin). Where the signal exceeds this maximum extent, signal start is designated as the first (highest) waveform bin. For operations in 2004 an onboard compression algorithm has been implemented whereby the waveform extent is increased to 150 m, thus minimizing occurrences of signal truncation.
Please refer to the NASA Wallops Flight Facility's GSAS v3.7 Release Notes for detailed release notes for each product in Releases 13 and 14.
Release-12 is the original, at-launch version of all products, using data from GLAS laser #1.
The data calibration and validation activities by the ICESat/GLAS Science Team were interrupted by the premature end of laser-1 operation after about 36 days on March 28, 2003, and were resumed with operation of laser-2 beginning on September 25. The results from ICESat demonstrate the unique capabilities of the laser measurements with unprecedented accuracy. However, the cal/val period data, in particular, should be regarded as not fully calibrated and validated (i.e. with some uncertain errors). Some limitations may be described by flags on the data products. One important limitation (to be corrected in future releases) is the laser-attitude accuracy (at the 10 to 15 arcsec level), which affects the accuracy of the surface elevations (cms to meters depending on surface slope) and geolocations of the measurements (tens of meters).
Another limitation is the distortion of the received-pulse shape due to saturation of the detector amplifier. Type-1 saturation, which non-linearly distorts the top and trailing edge of the waveform, affects much of the data over ice and other strong reflectors (tens of cm range effect) and may be correctable with future algorithms. Type-2 saturation, for which the gain goes to a maximum and stays there awhile, causes an effect in the meters range. Type 2 has been corrected for laser 2 data by changing the instrument gain algorithm. Lidar data on clouds and aerosols from the 1064 nm channel is included, but the backscatter cross-section magnitudes have a large offset that will be corrected. The more-sensitive 532 nm lidar channel was not yet activated, so there are no optical-depth or extinction cross-section products. Other limitations may be discovered as analysis and calibration/validation continues.
Please refer to the NASA Wallops Flight Facility's GSAS v3.6 Release Notes for detailed release notes for each product in Release-12.
|
Supporting Cryospheric Research Since 1976 |
|
