NASA - National Aeronautics and Space Administration

Detailed Quality Document for the CALIPSO Version 4.20 IIR Level 2 Track and Swath Data Products

Last Updated: April 28, 2020
Last Updated: October 11, 2020

Data Version: 4.20     Data Version: 4.21
Data Release Date: April 27, 2020     Data Release Date: October 12, 2020
Data Date Range: June 13, 2006 to June 30, 2020     Data Date Range: July 01, 2020 to present

10-01-2020: Version 4.21
A minor version bump (+0.01) has been applied to all CALIPSO data products due to a required upgrade to the operating system on the CALIPSO production cluster. All algorithms were re-compiled to process in this new environment with no change to the underlying science algorithms or inputs.

The IIR Level 2 algorithm, described in details in Garnier et al. (2012, 2013), takes full advantage of the co-located characterization of the atmosphere provided by CALIOP. It is applied to suitable types of scenes classified using the V4.20 CALIOP 5-km Cloud and Aerosol layer products. The retrievals are applied to the so-called upper level cloud which includes either one single semi-transparent or opaque layer, or several semi-transparent layers.

Effective emissivity retrievals require i) a correction for the contribution from the “background” and ii) the determination of the radiative temperature of the upper level cloud. When the lowermost of at least two individual cloud layers in the column is opaque to CALIOP, the background reference is from this opaque layer, which is called lower level. Otherwise, the background reference is from the Earth surface in clear air conditions. Both in V3 and in V4, the first step into the computation of the radiative temperature is to determine the centroid altitude of the upper level cloud.

As in V3, ice cloud effective diameter (De) retrievals use the concept of microphysical index (ßeff) applied to the IIR pairs of channels (12.05, 10.6) and (12.05, 8.65), with ßeff12/10 and ßeff12/08 defined as the 12.05-to-10.6 and 12.05-to-08.65 ratios of the effective absorption optical depths, respectively. The microphysical indices are interpreted in terms of De by using Look-Up Tables (LUTs) built for several models. The ice water path is then determined from calculated effective emissivities and De.

The most significant changes in the V4 algorithm include:

  • A refined scene classification
  • An updated radiative transfer model analysis for clear air reference with new surface emissivity values and corrected land surface temperature
  • A revised centroid altitude in multi-layer cases
  • Refined radiative temperature estimates in ice clouds
  • Updated and extended microphysical retrievals in ice clouds
  • Addition of microphysical retrievals in liquid water clouds
  • Revised uncertainty estimates

Refined Scene Classification


In V4, a new “Was_Cleared_Flag_1km” parameter reports the number of clouds seen by the 1-km IIR pixel that were detected by CALIOP at single shot resolution and that were cleared from the CALIOP 5-km layer products. These clouds have a top altitude < 4 km. In V3, these single-shot cleared clouds were not reported in the CALIOP 5-km layer products and they were ignored by the IIR algorithm.

In V4, the cloud-free (clear sky or aerosol only) scenes must have Was_Cleared_Flag_1km = 0. When Was_Cleared_Flag_1km ≠ 0, these scenes, which were thought to be cloud-free in V3, are assigned new types of scenes in V4.

For effective emissivity retrievals, the background radiance is computed regardless of Was_Cleared_Flag_1km. Unless the bias in the background radiance due to the cleared clouds can be estimated a posteriori and can be deemed acceptable, it is recommended to ensure that Was_Cleared_Flag_1km = 0 for detailed analyses of retrievals in optically thin clouds.

Absorbing Aerosols

For the scenes that fall into the “clouds” category, the semi-transparent aerosol layers, if any, are ignored when computing the cloud effective emissivity. In V4, we require that the “clouds” scenes that contain only one or two semi-transparent cloud layers do not contain high (typically stratospheric) aerosol layers, because these high aerosol layers could absorb in the IIR channels.

In addition, the presence of potentially absorbing aerosol layers (i.e. tropospheric dust or stratospheric aerosols) is now reported in a new Dust_Stratospheric_Aerosol_Flag and a new Dust_Stratospheric_Aerosol_Flag_QA so that the retrievals can be filtered using these flags.

Updated Radiative Transfer Model Analysis for Clear Air Reference

When no relevant neighboring observations can be found, the background reference is computed using a radiative transfer model, meteorological profiles and surface temperature from MERRA-2 re-analyses, and surface emissivity values.


The radiative transfer model has been updated in V4 to better simulate absorption by water vapor.

Surface Category, Surface Emissivity, and Surface Temperature

In V3, the surface emissivity values in each IIR channel were theoretical static values assigned to each IGBP Surface Type. When the daily Snow_Ice_Surface_type indicated the presence of snow or sea ice, IGBP Surface Type was changed to IGBP = 15 (snow/ice).

In V4, the characterization of the surface has been refined, and is reported in a new “TGeotype” parameter. TGeotype is assigned using IGBP Surface Type and the co-located Surface_532_Integrated_Depolarization_Ratio, now reported in the V4 CALIOP products, and Snow_Ice_Surface_type if Surface_532_Integrated_Depolarization_Ratio is not available. Surface_532_Integrated_Depolarization_Ratio is also used to refine the land/water classification around coastlines. The TGeotype values fall into five main categories: water, water/sea ice transition, sea ice, snow, and snow-free land.

Surface emissivity values used in V4 have been determined empirically from the analysis of two years of IIR data in clear sky conditions. The V4 algorithm continues to use static values for the water, water/sea ice transition, sea ice, and snow categories. For the snow-free land category, the V3 static values per IGBP type at 08.65 μm and 10.6 μm are replaced with monthly daytime and nighttime maps (resolution: latitude x longitude = 1° x 2°), while a constant value (0.975) is taken at 12.05 μm. In case of snow-free land, the surface temperature is the initial MERRA 2 surface temperature further corrected using monthly daytime and nighttime correction maps that were derived using the 12.05 μm reference channel by reconciling calculations and IIR observations.

Summary of Clear Air Reference Changes from V3 to V4

The changes described above improve the accuracy of effective emissivity and microphysical retrievals in optically thin layers when the clear air background radiance is derived from the model. Subsequently, retrievals derived from observed and computed background radiances are more consistent in V4 than in V3.

Over oceans, the change from V3 to V4 is for the most part the reduction of systematic effective emissivity biases at 08.65 μm, which improves the reliability of effective diameter retrievals. Even though less accurate than over oceans, retrievals over land are improved in V4, in particular in desertic areas and in polar regions.

Revised Centroid Altitude in Multi-layer Cases

Because of an error in the V3 algorithm, the centroid altitude of multi-layer systems could be too low by up to 3 km in V3. This has been fixed in V4.

Refined Radiative Temperature Estimates in Ice Clouds

In V3, the radiative temperature was set to the temperature at the lidar centroid altitude for any cloud or aerosol system. The approach is the same in V4, except when all the layers are classified as ice by the V4 CALIOP ice/water phase algorithm. In the latter case, the initial centroid temperature estimate (which was used in V3) has been shown to be cold (Garnier et al., 2015). It is now refined in V4 using parameterized correction functions. Blackbody brightness temperatures derived from both the centroid temperature and the V4 radiative temperature are reported in the V4 product.

The correction applied in V4 has a negligible impact on effective emissivity values smaller than about 0.3. The largest impact is for opaque clouds, the correction being more efficient for nighttime than for daytime data. For instance, for oceanic opaque ice clouds in January 2008, nighttime and daytime distributions of effective emissivity at 12.05 μm peak respectively at 0.99 and 0.97 in V4, compared to 0.94 in V3.

Updated Microphysical Retrievals in Ice Clouds

Optical Properties

The three ice crystal models used in V3 have been replaced with two ice crystal models in V4, namely severely rough “single hexagonal column” and “8-element column aggregate” elaborated from new ice properties (Yang et al., 2013; Bi and Yang, 2017). In addition, a gamma particle size distribution with effective variance equal to 0.1 is taken in V4, whereas there was no size distribution in V3. As a result of these changes, De is increased from V3 to V4.

Four independent sets of De derived from a second approach are also reported in V4. This approach uses analytical functions relating ßeff12/10 and De as derived from in situ measurements at tropical and mid-latitudes performed during the TC4 and SPARTICUS field experiments (Mitchell et al., 2018).

Ice Water Path

For ice clouds, Ice Water Path is estimated from De and visible extinction optical depth (τvis) derived from calculated effective emissivities. In V3, τvis was approximated to 2τa,12, where τa,12 is the IIR affective absorption optical depth at 12.05 μm. In V4, τvis is approximated to τa,12 + τa,10, where τa,10 is the IIR affective absorption optical depth at 10.6 μm. The V4 approximate reduces the dependence on De and improves the τvis estimate by up 10 % for De > 20 μm.

Addition of Microphysical Retrievals In Liquid Water Clouds

Unlike in V3, microphysical retrievals in liquid water clouds are provided in V4.

For De retrievals, the LUTs are computed using the Mie theory with refractive indices from Hale and Querry (1973) and again using a gamma particle size distribution (PSD) with effective variance equal to 0.1. No temperature dependence of infrared absorption is included.

Liquid Water Path is estimated from De, τa,12, and Qa,12(De), where the latter is the effective absorption efficiency at 12.05 μm, whose variation with De ≤ 20 μm is represented using a fourth-degree polynomial. In agreement with Pinnick et al. (1979), Qa,12 increases quasi-linearly with De < 10 μm up to about 1, and then increases slowly up to 1.15 as De increases from 10 to 20 μm.

Revised Uncertainty Estimates

Effective Emissivity Uncertainty

For each IIR channel, the estimated uncertainty in the effective emissivity is composed of three terms associated to

  • an error dTm in the measured brightness temperature,
  • an error dTBG in the background reference brightness temperature,
  • an error dTBB in the blackbody temperature.

In V3, the reported uncertainties were computed using dTm = dTBG = dTBB = 1 K. The following temperature errors are used in V4:

  Water Surface Non-water Surface
dTm(K) 0.3 0.3
dTBG(K) 1 3
dTBB(K) 2 2

In V4, the three terms used to compute the estimated uncertainty in the effective emissivity are reported.

Effective Diameter Uncertainty

The approach chosen to report the estimated uncertainty in De has been changed in V4. In V4, it is simply inferred from the estimated uncertainty in the microphysical indices and from the LUT used for the retrievals.

Ice or Liquid Water Path Uncertainty

Unlike in V3, an estimated uncertainty in the ice or liquid water path is reported in V4. It is computed from the estimated uncertainty in De and in the effective absorption optical depth(s).

Last Updated: November 22, 2021
Curator: Charles R. Trepte
NASA Official: Charles R. Trepte

+ No Fear Acts
+ Privacy

+ Office of Inspector General
+ Office of Special Councel
+ Agency Financial Reports

+ Contact NASA