Cloud Clearing and Cloud Properties

This section describes how cloud clearing affects the interpretation of CLIMCAPS soundings.

In this section, we describe how cloud clearing affects the interpretation of CLIMCAPS soundings. CLIMCAPS retrieves atmospheric variables from cloud cleared radiances in a series of steps. Once, when all variables have been retrieved, CLIMCAPS calculates cloud cleared radiances once more (according to the AIRS V7 method) and writes it out as a separate product (CCR). 


Does CLIMCAPS retrieve cloud properties?

Yes – cloud top pressure (CTP) and cloud fraction (α). CLIMCAPS retrieves cloud fraction for each field-of-view (FOV; instrument footprint) and a two-layer cloud top pressure for each field-of-regard (FOR; CLIMCAPS footprint). There are nine FOVs in each FOR (see Figure 1). This means that CLIMCAPS retrieves a total of 18 cloud parameters for each retrieval scene—9 α’s for every CTP layer—from a subset of channels sensitive to clouds in the troposphere. CLIMCAPS does not retrieve any cloud microphysical properties.

Does CLIMCAPS retrieve soundings from clear-sky scenes only?

No. CLIMCAPS retrieves atmospheric profile variables from clear and partly cloudy scenes. In fact, CLIMCAPS retrievals pass quality control in scenes with as much as 80–90% cloud cover. This said, it is important to distinguish that CLIMCAPS does not retrieve sounding profiles through cloud fields, but instead uses a technique known as ‘cloud clearing’ to remove the radiative effects of clouds from the infrared measurements. One can understand CLIMCAPS retrievals as representing the clear portion of the atmosphere past or around cloud fields in the target scene.

What is cloud clearing?

Cloud clearing is a linear extrapolation technique that aggregates spectral channels from 9 FOVs with varying degrees of cloud cover into a single set of channels that represents the clear state of the atmosphere around clouds in the target scene (or FOR).

Cloud clearing requires no prior knowledge of clouds in a target scene, nor does it depend on radiative transfer calculations through clouds. It is a simple technique that uses the spatial variability in cloud cover among 9 FOVs as information content to linearly derive a set of cloud-free infrared channels. If there is no variability in cloud cover among the 9 FOVs in a FOR, then their spatial information content is zero and cloud clearing fails. In turn, the higher the cloud heterogeneity among a cluster of FOVs, the higher the spatial information content and more accurate the cloud clearing.

Does cloud clearing impact measurement information content?

Yes, but only for the infrared channels. Elsewhere (e.g., Smith and Barnet, 2020) we describe how CLIMCAPS information content is a function of instrument noise, measurement and scene-specific uncertainty, channel selection and channel weighting functions. Cloud clearing impacts CLIMCAPS information content by affecting the random instrument noise (NEN) and scene-specific uncertainty. We quantify this as ampl_eta and etarej (Section 2), propagate them into subsequent retrievals as described in Smith and Barnet (2019) and report them in the CLIMCPAS product file as diagnostic metrics for use during analysis.

1. Brief summary of methods

Cloud clearing uses the spatial heterogeneity in cloud cover from an array of 3 × 3 FOVs as information content to derive a set of infrared spectral channels that represent the cloud-cleared (or cloud-free) atmosphere at a target scene (FOR). This technique was first developed by (Chahine, 1977, 1982; Smith, 1968) and later adopted by the AIRS Science Team for use in the AIRS retrieval systems (Susskind et al., 2003, 2014, 2017) as well as CLIMCAPS (Smith and Barnet, 2019, 2020). This technique is well documented in peer review publications and reports. Our goal here is to summarize the main steps only.      

The main purpose of cloud clearing is to improve retrieval yield and allow observation of the vertical atmospheric state in partly cloudy conditions. CLIMCAPS is a global product and cloud clearing allows the successful retrieval of sounding variables from ~75% of all IR+MW measurements made in one day.

The purpose of CLIMCAPS cloud parameter retrievals (cloud fraction and cloud top pressure) is to determine which channels to ‘cloud clear’ and which to simply average. We write the retrieved cloud parameters to the Level 2 product file to represent the full atmospheric state and enhance diagnostic and data filtering capability in subsequent analyses.

We gave a graphic depiction of the CLIMCAPS retrieval flow and elaborate here on those steps that mention cloud top pressure (CTP), cloud fraction (α) and cloud clearing (CC).

  1. Define the a-priori for two cloud layers within each CLIMCAPS retrieval footprint (FOR) as follows: [α= 0.5, CTP1 = 350], [α= 0.25, CTP2 = 800].
  2. Retrieve [α1, CTP1], [α, CTP2] for each FOR from a subset of infrared channels using the a-priori variables as defined for temperature (T(p)), water vapor (H2O), trace gases and surface variables.
  3. Based on the values of α and CTP, determine which channels to ‘cloud clear’ and which to simply average.
  4. Once all atmospheric state variables have been retrieved, retrieve α for each FOV, CTP for each FOR and derive cloud cleared radiance channels for each FOR.


  • CLIMCAPS cloud clearing does NOT depend on coincident microwave measurements (from AMSU or ATMS) to cloud clear its channel sets. Instead, CLIMCAPS uses MERRA2 (Gelaro et al. 2017) as a-priori to determine the clear-sky state of temperature and water vapor. 
  • Unlike AIRS V7, CLIMCAPS does NOT iterate cloud clearing and performs it only once before retrieving the atmospheric state variables.
  • Cloud fraction and cloud top pressure have the same radiative effect in top-of-atmosphere infrared measurements. This makes them difficult to retrieve. We constrain our solution to retrieving only two layers of CTP for the entire CLIMCAPS footprint (FOR), 9 × α for each FOV.
  • CLIMCAPS uses the same channel set to retrieve both CTP and α, which is a subset of the channels used in cloud clearing.
  • Channels in the long-wave infrared window region (~10 μm) are always cloud cleared, even if their cloud fraction retrievals are close to or equal to zero. This is a precautionary measure since space-based infrared measurements have, by definition, reduced observing capability in the boundary layer.
  • All other channels are sometimes cloud cleared, sometimes not, depending on the scene-specific CTP and α values.
  • Cloud clearing is more accurate the higher the cloud contrast among FOVs in a retrieval footprint (Figure 1).

Figure 1. CLIMCAPS retrieval footprint

Figure 1: The CLIMCAPS retrieval footprint is the field-of-regard (FOR; grey dashed circles) that consists of 3 × 3 instrument fields-of-view (FOV; black solid circles). CLIMCAPS aggregates the 9 FOVs into a single spectrum from which it then retrieves a set of atmospheric profile variables. Cloud fraction is the only variable that CLIMCAPS retrieves for each FOV (9 per FOR); all other retrieval variables represent conditions within the FOR (~50 km at nadir, ~150 km at edge-of-scan). Here we illustrate four typical cloudy scenarios encountered by CLIMCAPS: (a) partly cloudy FOR where all FOVs have cloud fraction > 0.0 (i.e., not clear) but no two FOVs has the same cloud fraction, (b) partly cloudy FOR where some FOVs have no clouds (cloud fraction = 0), (c) partly cloudy where each FOV has the exact same cloud fraction (no contrast), and (d) overcast FOR where all FOVs have cloud fraction = 1.0. Cloud clearing is accurate (i.e., small brightness temperature residuals with low etarej values) in (top row) spatial heterogeneous retrieval footprints, but fails (i.e., large brightness temperature residuals with high etarej values) in (bottom row) spatial homogeneous scenes.

Readers can refer to studies that compared AIRS cloud retrievals to a host of other observations (Kahn et al., 2014, 2015; Nasiri et al., 2011; Weisz et al., 2007; Wong et al., 2015, 2015; Wu et al., 2009; Yue et al., 2011). There are also studies that demonstrate the value of AIRS cloud cleared radiances in data assimilation (Reale et al., 2008, 2018).

2. Relevant CLIMCAPS product fields

Note that we refer, here, to the CLIMCAPS Level 2 retrieval product. The cloud cleared radiances are written to and distributed as a separate product not discussed here.

Within the netCDF files, we highlight a few fields that are relevant to clouds. The fields have dimensions that correspond to the following variables:

  • atrack = 30 (number of retrieval footprints along an instrument scanline)
  • xtrack = 45 (number of scanlines grouped together in a CLIMCAPS file)
  • fov = 9 (number of fields of view in a CLIMCAPS footprint)
  • cld_lay = 2 (number of cloud retrieval layers)
  • air_pres_lay = 100 (number of profile retrieval layers)

Retrieved variables

  • cld_frac(atrack, xtrack, fov, cld_lay)  Cloud fraction retrievals for each field-of-view (AIRS or CrIS instrument footprint) and up to two cloud layers from a subset of infrared channels.
  •  aux/for_cld_top_pres_2lay(atrack, xtrack, cld_lay) Cloud top pressure retrievals for up to two layers of clouds on each CLIMCAPS footprint (or field-of-regard) from a subset of infrared channels. 
  • mw_cld_phase (atrack, xtrack, air_pres_lay) Cloud ice detection flag for every retrieval layer using information in microwave channels; 0 means the CLIMCAPS footprint at a target layer has only liquid clouds or is cloud free, while 1 means that ice clouds were detected.

Derived variables

The methods used in deriving these cloud variables are the same as those used in the AIRS retrieval system for AIRS/AMSU and CrIS/ATMS. We refer the reader to Susskind et al. (2017) for a full description.

  • cld_top_pres(atrack, xtrack, fov, cld_lay) This is the for_cld_top_pres_2lay retrieval but reported on every field-of-view (FOV) to allow easy match-ups with the cld_frac field.
  • cld_top_temp(atrac, xtrac, fov, cld_lay) This is the value from air_temp (retrieved temperature profile) that corresponds to the cloud top pressure retrieval (for_cld_top_pres_2lay). Even though this field is reported on each FOV, it represents the temperature at the FOR (CLIMCAPS retrieval footprint) scale. The lack of variation in cld_top_temp across the FOVs as reported in this field does, therefore, not mean a real lack of variation across the FOR.
  • num_cld(atrack, xtrack, fov) Number of cloud layers with nonzero cloud fraction as depicted by cld_frac for each FOV.
  • aux/for_cld_frac_tot(atrack, xtrack) Cloud fraction across all cloud layers and FOVs to represent the total cloud cover for the target retrieval scene.
  • aux/for_cld_top_pres_tot (atrack, xtrack) Cloud top pressure for the retrieval scene (FOR) as the total for all cloud layers and FOVs. It is calculated as the weighted sum of the cloud top pressure from both cloud layers, divided by the sum of the cloud fraction from both cloud layers as follows:
ctp_wght = (cld_top_pres(1) ⁎ cld_frac(1)) + (cld_top_pres (2) ⁎ cld_frac(2))

for_cld_top_pres_tot = ctp_wght / (cld_frac(1) + cld_frac(2)). 

This method was developed by Lena Iredell et al. and is the same method also used in the CHART as described on p. 33 in this report

  • aux/for_cld_frac_2lay(atrac, xtrack, cld_lay) Total cloud fraction across all FOVs for two layers. This is similar to for_cld_frac_tot.
  • cldfrac_500(atrack, xtrack) The total cloud fraction of all clouds below 500 hPa over the retrieval footprint (FOR). This is similar to for_cld_frac_tot but only for those clouds in the lower troposphere.

Uncertainty metrics

  • aux/etarej(atrack,xtrack) The cloud clearing radiance error in brightness temperature units [Kelvin] calculated as the difference between a simulated clear-sky spectrum and the derived cloud cleared spectrum at the target retrieval scene. Etarej quantifies the quality of cloud clearing by indicating how well the cloud-cleared radiance represents the clear-sky state around the clouds at that scene. Smaller values of etarej indicate successful cloud clearing and a high confidence in the removal of clouds from the infrared radiance measurements. Higher values of etarej indicate a lower confidence in cloud clearing and retrievals should be interpreted as being ‘contaminated’ by residual undetected clouds.
  • aux/ampl_eta(atrack,xtrack) The amplification factor (ampl_eta) quantifies  how much the random instrument noise (NEN or NEdT in units Kelvin) was amplified (ampl_eta(i,j) > 1) or damped (ampl_eta(i,j) < 1) as a result cloud clearing.
  • aux/aeff_end(atrack,xtrack) The effective amplification factor (aeff_end) is a compound metric that combines random instrument noise as scaled by the ampl_eta with systematic uncertainty due to spectral correlation.

3. References

  • Chahine, M. T.: Retemote sounding of cloudy parameters.II. Multiple cloud formations, J. Atmos. Sci., 34, 744–757, 1977.
  • Chahine, M. T.: Remote sensing of cloud parameters, J. Atmos. Sci., 39, 159–170, 1982.
  • Gelaro, R., McCarty, W., Suárez, M. J., Todling, R., Molod, A., Takacs, L., Randles, C. A., Darmenov, A., Bosilovich, M. G., Reichle, R., Wargan, K., Coy, L., Cullather, R., Draper, C., Akella, S., Buchard, V., Conaty, A., da Silva, A. M., Gu, W., Kim, G.-K., Koster, R., Lucchesi, R., Merkova, D., Nielsen, J. E., Partyka, G., Pawson, S., Putman, W., Rienecker, M., Schubert, S. D., Sienkiewicz, M. and Zhao, B.: The Modern-Era Retrospective Analysis for Research and Applications, Version 2 (MERRA-2), Journal of Climate, 30(14), 5419–5454, doi:10.1175/JCLI-D-16-0758.1, 2017
  • Kahn, B. H., Irion, F. W., Dang, V. T., Manning, E. M., Nasiri, S. L., Naud, C. M., Blaisdell, J. M., Schreier, M. M., Yue, Q., Bowman, K. W., Fetzer, E. J., Hulley, G. C., Liou, K. N., Lubin, D., Ou, S. C., Susskind, J., Takano, Y., Tian, B. and Worden, J. R.: The Atmospheric Infrared Sounder version 6 cloud products, Atmospheric Chemistry and Physics, 14(1), 399–426, doi:10.5194/acp-14-399-2014, 2014.
  • Kahn, B. H., Schreier, M. M., Yue, Q., Fetzer, E. J., Irion, F. W., Platnick, S., Wang, C., Nasiri, S. L. and L’Ecuyer, T. S.: Pixel-scale assessment and uncertainty analysis of AIRS and MODIS ice cloud optical thickness and effective radius: AIRS AND MODIS ICE CLOUDS, Journal of Geophysical Research: Atmospheres, 120(22), 11,669-11,689, doi:10.1002/2015JD023950, 2015.
  • Nasiri, S. L., Van T. Dang, H., Kahn, B. H., Fetzer, E. J., Manning, E. M., Schreier, M. M. and Frey, R. A.: Comparing MODIS and AIRS Infrared-Based Cloud Retrievals, Journal of Applied Meteorology and Climatology, 50(5), 1057–1072, doi:10.1175/2010JAMC2603.1, 2011.
  • Reale, O., Susskind, J., Rosenberg, R., Brin, E., Liu, E., Riishojgaard, L. P., Terry, J. and Jusem, J. C.: Improving forecast skill by assimilation of quality-controlled AIRS temperature retrievals under partially cloudy conditions, Geophysical Research Letters, 35(8), doi:10.1029/2007GL033002, 2008.
  • Reale, O., McGrath-Spangler, E. L., McCarty, W., Holdaway, D. and Gelaro, R.: Impact of Adaptively Thinned AIRS Cloud-Cleared Radiances on Tropical Cyclone Representation in a Global Data Assimilation and Forecast System, Weather and Forecasting, 33(4), 909–931, doi:10.1175/WAF-D-17-0175.1, 2018.
  • Smith, N. and Barnet, C. D.: Uncertainty Characterization and Propagation in the Community Long-Term Infrared Microwave Combined Atmospheric Product System (CLIMCAPS), Remote Sensing, 11(10), 1227, doi:10.3390/rs11101227, 2019.
  • Smith, N. and Barnet, C. D.: CLIMCAPS Observing Capability for Temperature, Moisture and Trace Gases from AIRS/AMSU and CrIS/ATMS, Atmospheric Measurement Techniques, 13, 4437–4458, doi:10.5194/amt-13-4437-2020, 2020.
  • Smith, W. L.: An improved method for calculating tropospheric temperature and moisture from satellite radiometer measurements, Monthly Weather Review, 96(6), 387–396, doi:10.1175/1520-0493(1968)096<0387:AIMFCT>2.0.CO;2, 1968.
  • Susskind, J., Barnet, C. D. and Blaisdell, J. M.: Retrieval of atmospheric and surface parameters from AIRS/AMSU/HSB data in the presence of clouds, IEEE TGRS, 41, 390–409, 2003.
  • Susskind, J., Blaisdell, J. M. and Iredell, L.: Improved methodology for surface and atmospheric soundings, error estimates, and quality control procedures: the atmospheric infrared sounder science team version-6 retrieval algorithm, Journal of Applied Remote Sensing, 8(1), 084994, doi:10.1117/1.JRS.8.084994, 2014.
  • Susskind, J., Blaisdell, J., Kouvaris, L. and Iredell, L.: CrIS CHART Retrieval Algorithm ATBD, Algorithm Theoretical Basis Document, NASA GSFC. [online] Available from:, 2017.
  • Weisz, E., Li, J., Menzel, W. P., Heidinger, A. K., Kahn, B. H. and Liu, C.-Y.: Comparison of AIRS, MODIS, CloudSat and CALIPSO cloud top height retrievals, Geophysical Research Letters, 34(17), doi:10.1029/2007GL030676, 2007.
  • Wong, S., Fetzer, E. J., Schreier, M., Manipon, G., Fishbein, E. F., Kahn, B. H., Yue, Q. and Irion, F. W.: Cloud-induced uncertainties in AIRS and ECMWF temperature and specific humidity: Cloud-dependent AIRS V6 validation, Journal of Geophysical Research: Atmospheres, 120(5), 1880–1901, doi:10.1002/2014JD022440, 2015.
  • Wu, D. L., Ackerman, S. A., Davies, R., Diner, D. J., Garay, M. J., Kahn, B. H., Maddux, B. C., Moroney, C. M., Stephens, G. L., Veefkind, J. P. and Vaughan, M. A.: Vertical distributions and relationships of cloud occurrence frequency as observed by MISR, AIRS, MODIS, OMI, CALIPSO, and CloudSat, Geophysical Research Letters, 36(9), doi:10.1029/2009GL037464, 2009.
  • Yue, Q., Kahn, B. H., Fetzer, E. J. and Teixeira, J.: Relationship between marine boundary layer clouds and lower tropospheric stability observed by AIRS, CloudSat, and CALIOP, Journal of Geophysical Research, 116(D18), doi:10.1029/2011JD016136, 2011.