CALIPSO Quality Statements |
This document provides a high-level quality assessment of the level 1B lidar data products, as described in Section 2.1 of the CALIPSO Data Products Catalog (Version 3.0) (PDF). As such, it represents the minimum information needed by scientists and researchers for appropriate and successful use of these data products. We strongly suggest that all authors, researchers, and reviewers of research papers review this document for the latest status before publishing any scientific papers using these data products.
The purpose of these data quality summaries is to inform users of the accuracy of CALIOP data products as determined by the CALIPSO Science Team and Lidar Science Working Group (LSWG). This document is intended to briefly summarize key validation results; provide cautions in those areas where users might easily misinterpret the data; supply links to further information about the data products and the algorithms used to generate them; and offer information about planned algorithm revisions and data improvements
The CALIOP Level 1B data product contains a half orbit (day or night) of calibrated and geolocated single-shot (highest resolution) lidar profiles, including 532 nm and 1064 nm attenuated backscatter and depolarization ratio at 532 nm. The product released contains data from nominal science mode measurement.
The CALIOP Level 1B product also contains additional data not found in the Level 0 lidar input file, including post processed ephemeris data, celestial data, and converted payload status data. The major categories of lidar Level 1B data are:
To make proper use of the CALIOP Level 1B products, all users must be aware of the uncertainties inherent in the data products. The data quality of each product is summarized briefly below:
The 532 nm attenuated backscatter coefficients are reported for each laser pulse as an array of 583 elements that have been registered to a constant altitude grid defined by the Lidar Data Altitude field.
Note that to reduce the downlink data volume, an on-board averaging scheme is applied using different horizontal and vertical resolutions for different altitude regimes, as shown in the following table.
Altitude Range (km) | Bin Number | Horizontal Resolution (km) |
532 nm Vertical Resolution (m) |
1064 nm Vertical Resolution (m) |
Altitude Region |
---|---|---|---|---|---|
30.1 to 40.0 | 1-33 | 5 | 300 | N/A | 5 |
20.2 to 30.1 | 34-88 | 5/3 | 180 | 180 | 4 |
8.3 to 20.2 | 89-288 | 1 | 60 | 60 | 3 |
-0.5 to 8.3 | 289-578 | 1/3 | 30 | 60 | 2 |
-2.0 to -0.5 | 579-583 | 1/3 | 300 | 300 | 1 |
Uncertainties for the attenuated backscatter are not explicitly reported in the CALIOP Level 1 data products to save data volume, which would otherwise approximately double the Level 1 data volume. If needed, users can compute random errors for the attenuated backscatter products as described in Uncertainties for Attenuated Backscatter (PDF). IDL code for computing the attenuated backscatter uncertainties is contained in IDL Code for Uncertainty Calculations (PDF).
For the nighttime portion of an orbit, the 532 nm calibration constant is determined for each 55 km averaged profile (11 frames) by comparing the 532-parallel signals in 30 km to 34 km altitude range to a scattering model derived from molecular and ozone number densities provided by NASA's Global Modeling and Assimilation Office (GMAO). This calculation uses equation 4.7 in Section 4.1.2.1 of the CALIPSO Lidar Level I ATBD (PDF). The computed 532 nm calibration constants are then smoothed over an interval of 1485 km using equation 4.8. A constant value of the calibration constant is applied to all single-shot profiles in each 55 km averaging region.
The calibration technique used during the nighttime cannot be used in the daytime portions of the orbits, because the noise associated with solar background signals (i.e., sunlight) degrades the backscatter signal-to-noise ratio (SNR) in the calibration region below usable levels. Therefore, for the daytime portion of the orbit, the calibration constants are derived by interpolating between values derived in the adjacent nighttime portions of the orbits.For each granule (day or night) a single, constant value (granule mean) for C1064 is derived by averaging all individual calibration constant estimates that were obtained. This granule mean serves as the calibration constant that is subsequently applied to all 1064 nm profiles in the granule.
We note that the procedure used in the 532 nm calibration cannot be applied for the 1064 nm measurements, because the molecular scattering at 1064 nm is ~16 times weaker than at 532 nm, and because the avalanche photodiode (APD) detector used in the 1064 nm channel has significantly higher dark noise than photomultiplier tube (PMTs) used in the 532 nm channels.For nighttime calibrations, the uncertainty due to noise is estimated to be typically smaller than 1%. Additional systematic errors may arise from aerosol contamination of the calibration region (less than a few percent), and from large signal spikes seen frequently in the South Atlantic Anomaly (SAA) and occasionally outside the SAA region.
A stratospheric aerosol model is currently being developed to correct for the aerosol present in the calibration region. Upon completion, this model will be applied to calibration processing for subsequent data releases.
Large noise spikes can be present both in the lidar return signals and in the baseline signals. Baseline signals are determined on-board by calculating the mean signal value over 15000 data points (1000 15 meter samples in the 65 to 80 km altitude region from each of the 15 shots within a frame). This calculation is performed for each frame, and the resulting value is subtracted from each sample of all profiles in that frame. The presence of large outliers -- i.e., "spikes" -- in the backscatter signals in the calibration region tends to bias the calibration constant toward a larger value. On the other hand, the spikes present in the baseline region can cause and erroneous overestimate of the measured baseline signal, and the subsequent subtraction of this baseline value will thus introduce a bias in all data within the frame, causing it to be lower than it otherwise should be. This in turn tends to bias the calibration constant toward a smaller value. Threshold-based data filtering schemes are applied to 532 nm data to remove large spikes in the lidar signal prior to performing the nighttime calibration. Two threshold boundaries - a maximum and a minimum - are set. By excluding values outside this range, large signal excursions are effectively removed. Spikes with smaller magnitudes may remain, depending on the selection of the maximum threshold value. Perturbations to the calibration due to spikes in the baseline region can be only partly eliminated by this kind of threshold-based filtering scheme. However, by properly selecting the threshold limits, the impacts of spikes in the calibration region and the baseline region will cancel each other out to some degree. Preliminary comparisons of CALIOP's 532 nm attenuated backscatter coefficients, which are critically dependent on the accuracy of the calibration, with validation measurements acquired by the LaRC airborne high-spectral-resolution lidar (HSRL) and Goddard's airborne Cloud Physics Lidar (CPL) show consistency to within a few percent.
Because the daytime calibration constants are interpolated from nighttime values, the uncertainties contained in the nighttime calibration are transferred to daytime. Additional error may arise from the selection of interpolation scheme. In general, the uncertainty for daytime calibration constants is somewhat higher than the uncertainty for the nighttime values.If a sufficient number of cirrus clouds are present in any granule, the uncertainty due to noise in the granule mean of the 1064 nm calibration constant can be very small. Larger systematic errors may arise from the assumption that the cirrus color ratio (the ratio of backscatter coefficients at 1064 nm and 532 nm) has a constant value of 1.0. A very preliminary study on the ratio of gain and energy-normalized, range-corrected signals (i.e., the quantity X defined in equations 3.7 and 3.8 in the CALIPSO Lidar Level I ATBD (PDF)) at 1064 nm and 532 nm in selected dense cirrus clouds shows a distribution having a width of exceeding 10% of the mean value.
The CALIPSO lidar data are averaged on-board the satellite, prior to being downlinked, using the variable averaging scheme shown in Table 1. Regions 1 and 2 contain single shot data (albeit at different vertical resolutions). In regions 3, 4, and 5, the downlinked data have been averaged to horizontal resolutions of, respectively, 3 shots, 5 shots, and 15 shots. The level 1 processing constructs Pseudo Single Shot Profiles (PSSP) by replicating the data from regions 3, 4, and 5, and then stacking data arrays from the different averaging regions. Two sets of QC flags, as shown in Tables 2 and 3, are computed for each one of these pseudo single shot profiles.
The laser energy assessments reported in QC Flag #1 are computed as follows:For example, suppose that (a) the energies for shot #5 in a 15 shot frame fail the data quality threshold tests but are above the 'near zero' threshold; and (b) the energies for all other shots in the frame are normal. In this case, bits 5, 6, 13, and 14 in profiles 1-4 and 6-15 are all set to zero, to indicate acceptable laser energy. In profile #5, bits 5 and 6 are zero (because the energies were above the 'near zero' threshold) and bits 13 and 14 are one (because the energies were below the data quality threshold). Because the averaged data in region 3 of PSSP #4, #5, and #6 was constructed using the low energy data recorded for shot #5, bits 15 and 16 in these three profiles are set to one, whereas in the remaining profiles bits 15 and 16 are set to zero. Similarly, bits 17 and 18 are set to one for PSSP #1 - #5, and bit 19 is set to one for all profiles in the 15 shot frame.
Bits | Interpretation |
---|---|
1 | 532 nm parallel channel missing |
2 | 532 nm perpendicular channel missing |
3 | 1064 nm channel missing |
4 | Not geolocated |
5 | Single shot 532 laser energy below calibration threshold (near zero energy) |
6 | Single shot 1064 laser energy below calibration threshold (near zero energy) |
7 | Historical value used for the depolarization gain ratio |
8 | Historical calibration constant used, 532 nm parallel channel |
9 | Historical calibration constant used, 532 nm perpendicular channel |
10 | Historical calibration constant used, 1064 nm channel |
11 | Averaged calibration constant used, 532 nm parallel channel |
12 | Averaged calibration constant used, 532 nm perpendicular channel |
13 | Single shot 532 laser energy below data quality threshold (low energy) |
14 | Single shot 1064 laser energy below data quality threshold (low energy) |
15 | Near zero 532 nm laser energy profile included in region 3 average |
16 | Near zero 1064 nm laser energy profile included in region 3 average |
17 | Near zero 532 nm laser energy profile included in region 4 average |
18 | Near zero 1064 nm laser energy profile included in region 4 average |
19 | Near zero 532 nm laser energy profile included in region 5 average |
20 | Low 532 nm laser energy profile included in region 3 average |
21 | Low 1064 nm laser energy profile included in region 3 average |
22 | Low 532 nm laser energy profile included in region 4 average |
23 | Low 1064 nm laser energy profile included in region 4 average |
24 | Low 532 nm laser energy profile included in region 5 average |
25-32 | Spare |
Bits | Interpretation |
---|---|
1 | Reserve |
2 | Excessive underflows, 532 nm parallel channel in region 6* |
3 | Excessive underflows, 532 nm perpendicular parallel channel, region 6* |
4 | Excessive underflows, 1064 nm channel, region 6* |
5 | Excessive overflows, 532 nm parallel channel, region 6* |
6 | Excessive overflows, 532 nm perpendicular parallel channel, region 6* |
7 | Excessive overflows, 1064 nm channel, region 6* |
8 | Excessive overflows, 532 nm parallel channel, region 2 |
9 | Excessive overflows, 532 nm perpendicular parallel channel, region 2 |
10 | Excessive overflows, 1064 nm channel, region 2 |
11 | LRE Flags in SAD packet indicate bad data, 532 nm parallel channel |
12 | LRE Flags in SAD packet indicate bad data, 532 nm perpendicular channel |
13 | LRE Flags in SAD packet indicate bad data, 1064 nm channel |
14 | Quality Flags in SAD packet indicate bad data, 532 nm parallel channel |
15 | Quality Flags in SAD packet indicate bad data, 532 nm perpendicular channel |
16 | Quality Flags in SAD packet indicate bad data, 1064 nm channel |
17 | Suspicious offset calculation, 532 nm parallel channel |
18 | Suspicious offset calculation, 532 nm perpendicular channel |
19 | Suspicious offset calculation, 1064 nm channel |
20 | Suspicious mean signal value, 532 nm parallel channel (any/all regions) |
21 | Suspicious mean signal value, 532 nm perpendicular channel (any/all regions) |
22 | Suspicious mean signal value, 1064 nm channel (any/all regions) |
23 | RMS noise out of range, 532 nm parallel channel |
24 | RMS noise out of range, 532 nm perpendicular parallel channel |
25 | RMS noise out of range, 1064 nm channel |
26-32 | Spare |
Lidar Level 1B Profiles Information Half orbit (Night and Day) geolocated, calibrated Lidar Profiles and Viewing Geometry Products |
|||
---|---|---|---|
Release Date | Version | Data Date Range | Maturity Level |
June 2009 | 3.00 | Initial release: March 12 - May 11, 2009 Future release: June 13, 2006 to present |
Validated Stage 1 |
Version 3.00 includes algorithm improvements, modifications to existing data parameters, and new data parameters. The maturity level of Version 3.00 Lidar Level 1B data product is assigned Validated Stage 1. In this stage, all obvious errors have been identified and corrected, and intercomparisons of attenuated backscatter products have been performed at selected locations and times.
Algorithm improvements were implemented for:532-nm daytime calibration
The revised 532-nm daytime calibration algorithm produces improved
corrections to the thermally-induced drift in signal level that occurs
over the course of the daytime orbit segment. In Version 3.00,
the empirically determined correction factors are applied using a
34-point linear approximation as compared to the 5-point linear
approximation implemented in Versions 2.01 and 2.02. This allows for better
characterization of the small scale changes in signal level that take place
over the daytime orbit segment. Comparisons of nighttime and newly
calibrated daytime clear-air, attenuated scattering ratios over 8-12 km in
altitude were made for multiple seasons of LOM 2 (first laser) operation and
for the first three months of LOM 1 (backup laser) operation. In all cases
the agreement between night and day was within 5% for the entire orbit
segment.
laser energy calculations and signal normalization by laser
energy
Two updates to the 532-nm and 1064-nm laser energy calculation algorithm
were implemented in Version 3.00 in order to reduce errors in both
calibration and the processing of signal profiles for low energy laser shots.
The first update uses new laser energy conversion coefficients to improve
the accuracy of the laser energy calculation.
In the second update, the signal normalization by laser energy is changed
to normalize by averaging region instead of by shot. That is, for each
averaging region, normalize all averaged shots by the corresponding average
energy for that region.
Data are averaged on-board over 15, 5, 3, or 1 shot(s) before downlink,
with the amount of averaging depending upon the altitude.
Application of the new normalization scheme improves the signal
normalization for frames with low energy laser shots and has little
effect on frames with nominal laser energies.
interpolation of GMAO gridded data products to the CALIPSO orbit
tracks
Corrections were made to the code used to interpolate the GMAO gridded data
products to the CALIPSO orbit tracks. In Versions 2.01 and 2.02, two
bracketing GMAO files were used to derive meteorological parameters.
In some cases, the CALIPSO measurement times fell outside of the bracketing
file times causing parameters to be extrapolated.
In Version 3.00, this problem was rectified by selecting three GMAO files
for each orbit track segment.
This assures the orbit track times are completely contained within the
GMAO data file times.
The modified parameters are:
Met_Data_Altitude
The altitudes reported in the parameter Met_Data_Altitude were modified so
they are now coincident with an altitude reported in the Lidar_Data_Altitude
array.
QC_Flag and QC_Flag_2
The quality flags QC_Flag and QC_Flag_2 were updated to identify the
profiles that were normalized using low energy.
New data parameters describe the orbit and path number and are included in the file metadata. The following six parameters were added:
Obtain a Portable Document Format (PDF) reader: Adobe Acrobat Reader | PDF Reader software list