• Ackerman, S., , K. Strabala, , W. Menzel, , R. Frey, , C. Moeller, , and L. Gumley, 1998: Discriminating clear sky from clouds with MODIS. J. Geophys. Res., 103, 32 14132 157.

    • Search Google Scholar
    • Export Citation
  • Aumann, H. H., and Coauthors, 2003: AIRS/AMSU/HSB on the Aqua mission: Design, science objectives, data products and processing systems. IEEE Trans. Geosci. Remote Sens., 41, 253264.

    • Search Google Scholar
    • Export Citation
  • Baran, A. J., , and P. N. Francis, 2004: On the radiative properties of cirrus cloud at solar and thermal wavelengths: A test of model consistency using high-resolution airborne radiance measurements. Quart. J. Roy. Meteor. Soc., 130, 763778.

    • Search Google Scholar
    • Export Citation
  • Barnes, W. L., , T. S. Pagano, , and V. V. Salomonson, 1998: Prelaunch characteristics of the Moderate Resolution Imaging Spectroradiometer (MODIS) on EOS-AM1. IEEE Trans. Geosci. Remote Sens., 36, 10881100.

    • Search Google Scholar
    • Export Citation
  • Chahine, M. T., 1974: Remote sounding of cloudy atmospheres. I. The single cloud layer. J. Atmos. Sci., 31, 233243.

  • Comstock, J. M., , T. P. Ackerman, , and G. G. Mace, 2002: Ground-based lidar and radar remote sensing of tropical cirrus clouds at Nauru Island: Cloud statistics and radiative impacts. J. Geophys. Res., 107, 4714, doi:10.1029/2002JD002203.

    • Search Google Scholar
    • Export Citation
  • Dong, S., , S. T. Gille, , J. Sprintall, , and E. J. Fetzer, 2010: Assessing the potential of the Atmospheric Infrared Sounder (AIRS) surface temperature and specific humidity in turbulent heat flux estimates in the Southern Ocean. J. Geophys. Res., 115, C05013, doi:10.1029/2009JC005542.

    • Search Google Scholar
    • Export Citation
  • Fetzer, E. J., , J. Teixeira, , E. T. Olsen, , and E. F. Fishbein, 2004: Satellite remote sounding of atmospheric boundary layer temperature inversions over the subtropical eastern Pacific. Geophys. Res. Lett., 31, L17102, doi:10.1029/2004GL020174.

    • Search Google Scholar
    • Export Citation
  • Fetzer, E. J., , B. H. Lambrigtsen, , A. Eldering, , H. H. Aumann, , and M. T. Chahine, 2006: Biases in total precipitable water vapor climatologies from Atmospheric Infrared Sounder and Advanced Microwave Scanning Radiometer. J. Geophys. Res., 111, D09S16, doi:10.1029/2005JD006598.

    • Search Google Scholar
    • Export Citation
  • Fetzer, E. J., and Coauthors, 2008: Comparison of upper tropospheric water vapor observations from the Microwave Limb Sounder and Atmospheric Infrared Sounder. J. Geophys. Res., 113, D22110, doi:10.1029/2008JD010000.

    • Search Google Scholar
    • Export Citation
  • Frey, R. A., , S. A. Ackerman, , Y. Liu, , K. I. Strabala, , H. Zhang, , J. R. Key, , and X. Wang, 2008: Cloud detection with MODIS. Part I: Improvements in the MODIS cloud mask for collection 5. J. Atmos. Oceanic Technol., 25, 10571072.

    • Search Google Scholar
    • Export Citation
  • Garay, M. J., , S. P. de Szoeke, , and C. M. Moroney, 2008: Comparison of marine stratocumulus cloud top heights in the southeastern Pacific retrieved from satellites with coincident ship‐based observations. J. Geophys. Res., 113, D18204, doi:10.1029/2008JD009975.

    • Search Google Scholar
    • Export Citation
  • Ham, S. H., , B. J. Sohn, , P. Yang, , and B. A. Baum, 2009: Assessment of the quality of MODIS cloud products from radiance simulations. J. Appl. Meteor. Climatol., 48, 15911612.

    • Search Google Scholar
    • Export Citation
  • Holz, R. E., , S. A. Ackerman, , F. W. Nagle, , R. Frey, , S. Dutcher, , R. E. Kuehn, , M. A. Vaughan, , and B. Baum, 2008: Global Moderate Resolution Imaging Spectroradiometer (MODIS) cloud detection and height evaluation using CALIOP. J. Geophys. Res., 113, D00A19, doi:10.1029/2008JD009837.

    • Search Google Scholar
    • Export Citation
  • Johnson, R. H., , T. M. Richenbach, , S. A. Rutledge, , P. E. Ciesielski, , and W. H. Schubert, 1999: Trimodal characteristics of tropical convection. J. Climate, 12, 23972418.

    • Search Google Scholar
    • Export Citation
  • Kahn, B. H., , A. Eldering, , A. J. Braverman, , E. J. Fetzer, , J. H. Jiang, , E. Fishbein, , and D. L. Wu, 2007a: Toward the characterization of upper tropospheric clouds using Atmospheric Infrared Sounder and Microwave Limb Sounder observations. J. Geophys. Res., 112, D05202, doi:10.1029/2006JD007336.

    • Search Google Scholar
    • Export Citation
  • Kahn, B. H., , E. Fishbein, , S. L. Nasiri, , A. Eldering, , E. J. Fetzer, , M. J. Garay, , and S.-Y. Lee, 2007b: The radiative consistency of Atmospheric Infrared Sounder and Moderate Resolution Imaging Spectroradiometer cloud retrievals. J. Geophys. Res., 112, D09201, doi:10.1029/2006JD007486.

    • Search Google Scholar
    • Export Citation
  • Kahn, B. H., and Coauthors, 2008: Cloud-type comparisons of AIRS, CloudSat, and CALIPSO cloud height and amount. Atmos. Chem. Phys., 8, 12311248.

    • Search Google Scholar
    • Export Citation
  • L’Ecuyer, T. S., , P. Gabriel, , K. Leesman, , S. J. Cooper, , and G. L. Stephens, 2006: Objective assessment of the information content of visible and infrared radiance measurements for cloud microphysical property retrievals over the global oceans. Part I: Liquid clouds. J. Appl. Meteor. Climatol., 45, 2041.

    • Search Google Scholar
    • Export Citation
  • Li, J., and Coauthors, 2005: Retrieval of cloud microphysical properties from MODIS and AIRS. J. Appl. Meteor., 44, 15261543.

  • Mace, G. G., , Q. Zhang, , M. Vaughan, , R. Marchand, , G. Stephens, , C. Trepte, , and D. Winker, 2009: A description of hydrometeor layer occurrence statistics derived from the first year of merged CloudSat and CALIPSO data. J. Geophys. Res., 114, D00A26, doi:10.1029/2007JD009755.

    • Search Google Scholar
    • Export Citation
  • Menzel, W. P., and Coauthors, 2008: MODIS global cloud-top pressure and amount estimation: Algorithm description and results. J. Appl. Meteor. Climatol., 47, 11751198.

    • Search Google Scholar
    • Export Citation
  • Overoye, K., , H. H. Aumann, , M. H. Weiler, , G. W. Gigioli, , W. Shaw, , E. Frost, , and T. McKay, 1999: Test and calibration of the AIRS instrument. Infrared Spaceborne Remote Sensing VII, M. Strojnik and B. F. Andresen, Eds., International Society for Optical Engineering (SPIE Proceedings Vol. 3759), 254–265.

    • Search Google Scholar
    • Export Citation
  • Platnick, S., , M. D. King, , S. A. Ackerman, , W. P. Menzel, , B. A. Baum, , J. C. Riédi, , and R. A. Frey, 2003: The MODIS cloud products: Algorithms and examples from Terra. IEEE Trans. Geosci. Remote Sens., 41, 459473.

    • Search Google Scholar
    • Export Citation
  • Rossow, W. B., and Coauthors, 1985: ISCCP cloud algorithm intercomparisons. J. Climate Appl. Meteor., 24, 877903.

  • Schreier, M. M., , B. H. Kahn, , A. Eldering, , D. A. Elliott, , E. Fishbein, , F. W. Irion, , and T. S. Pagano, 2010: Radiance comparisons of MODIS and AIRS using spatial response information. J. Atmos. Oceanic Technol., 27, 13311342.

    • Search Google Scholar
    • Export Citation
  • Susskind, J., , C. D. Barnet, , and J. M. Blaisdell, 2003: Retrieval of atmospheric and surface parameters from AIRS/AMSU/HSB data in the presence of clouds. IEEE Trans. Geosci. Remote Sens., 41, 390409.

    • Search Google Scholar
    • Export Citation
  • Susskind, J., , C. D. Barnet, , J. M. Blaisdell, , L. Iredell, , F. Keita, , L. Kouvaris, , G. Molnar, , and M. Chahine, 2006: Accuracy of geophysical parameters derived from Atmospheric Infrared Sounder/Advanced Microwave Sounding Unit as a function of fractional cloud cover. J. Geophys. Res., 111, D09S17, doi:10.1029/2005JD006272.

    • Search Google Scholar
    • Export Citation
  • Tobin, D. C., , H. E. Revercomb, , C. C. Moeller, , and T. S. Pagano, 2006: Use of AIRS high spectral resolution spectra to assess the calibration of MODIS on EOS Aqua. J. Geophys. Res., 111, D09S05, doi:10.1029/2005JD006095.

    • Search Google Scholar
    • Export Citation
  • Wei, H., , P. Yang, , J. Li, , B. A. Baum, , H.-L. Huang, , S. Platnick, , Y. Hu, , and L. Strow, 2004: Retrieval of semitransparent ice cloud optical thickness from Atmospheric Infrared Sounder (AIRS) measurements. IEEE Trans. Geosci. Remote Sens., 42, 22542267.

    • Search Google Scholar
    • Export Citation
  • Yue, Q., , and K. N. Liou, 2009: Cirrus cloud optical and microphysical properties determined from AIRS infrared spectra. Geophys. Res. Lett., 36, L05810, doi:10.1029/2008GL036502679.

    • Search Google Scholar
    • Export Citation
  • View in gallery

    Comparisons between retrieved MODIS and AIRS (left) effective brightness temperature and (right) cloud-top temperature for 1 Jan 2005. The color scale indicates the number of observations. The 1-to-1 line is shown in black, and the line of best fit is in red. The (a),(b) all-observations, (c),(d) single-layer, and (e),(f) two-layer data are based on AIRS effective cloud fraction as described in the text.

  • View in gallery

    Comparisons between measured brightness temperature and calculated effective brightness temperature for (left) MODIS and (right) AIRS: (a) (11 μm) vs for individual MODIS retrieval, (c) (11 μm) vs for AIRS–AMSU-A averaged MODIS values, (e) as in (c) but for (12 μm), (b) (8.121 μm) vs , (d) as in (b) but for (10.410 μm), and (f) as in (b) but for (11.668 μm).

  • View in gallery

    Cumulative distribution functions for MODIS–AIRS (a) effective brightness temperature, (b) retrieved cloud-top temperature, and (c) assumed surface temperature.

  • View in gallery

    Cases in which ≥60% of the MODIS retrievals within the AIRS FOR used the (left) CO2-slicing and (right) window retrieval methods for (a),(b) all-observations, (c),(d) single-layer, and (e),(f) two-layer data.

  • View in gallery

    Histograms of (a) , (b) fA, (c) , and (d) fM are shown for the single-layer case.

  • View in gallery

    Cumulative distribution plots of the dependence of on f for (a),(b) all-observations, (c),(d) single-layer, and (e),(f) two-layer data. (left) Curves vary with fA; (right) curves vary with fM.

  • View in gallery

    As in Fig. 6, but for MODIS − AIRS effective brightness temperature.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 27 27 4
PDF Downloads 19 19 5

Comparing MODIS and AIRS Infrared-Based Cloud Retrievals

View More View Less
  • 1 Department of Atmospheric Science, Texas A&M University, College Station, Texas
  • | 2 Jet Propulsion Laboratory, California Institute of Technology, Pasadena, California
  • | 3 Jet Propulsion Laboratory, California Institute of Technology, Pasadena, and Joint Institute for Regional Earth System Science and Engineering, University of California, Los Angeles, Los Angeles, California
  • | 4 Cooperative Institute for Meteorological Satellite Studies, Madison, Wisconsin
© Get Permissions
Full access

Abstract

Comparisons are described for infrared-derived cloud products retrieved from the Moderate Resolution Imaging Spectroradiometer (MODIS) and the Atmospheric Infrared Sounder (AIRS) using measured spatial response functions obtained from prelaunch AIRS calibration. One full day (1 January 2005) of global collection-5 MODIS and version-5 AIRS retrievals of cloud-top temperature Tc, effective cloud fraction f, and derived effective brightness temperature Tb,e is investigated. Comparisons of Tb,e demonstrate that MODIS and AIRS are essentially radiatively consistent and that MODIS Tb,e is 0.62 K higher than AIRS Tb,e for all scenes, increasing to 1.43 K for cloud described by AIRS as single layer and decreasing to 0.50 K for two-layer clouds. Somewhat larger differences in Tc and f are observed between the two instruments. The magnitudes of differences depend partly on whether MODIS uses a CO2-slicing or 11-μm brightness temperature window retrieval method. Some cloud- and regime-type differences and similarities between AIRS and MODIS cloud products are traceable to the assumptions made about the number of cloud layers in AIRS and also to the MODIS retrieval method. This (partially) holistic comparison approach should be useful for ongoing algorithm refinements, rigorous assessments of climate applicability, and establishment of the capability of synergistic MODIS and AIRS retrievals for improved cloud quantities and also should be useful for future observations to be made by the National Polar-Orbiting Operational Environmental Satellite System (NPOESS) Preparatory Project (NPP).

Corresponding author address: Dr. Shaima L. Nasiri, Dept. of Atmospheric Science, Texas A&M University, College Station, TX 77843-3150. E-mail: snasiri@tamu.edu

Abstract

Comparisons are described for infrared-derived cloud products retrieved from the Moderate Resolution Imaging Spectroradiometer (MODIS) and the Atmospheric Infrared Sounder (AIRS) using measured spatial response functions obtained from prelaunch AIRS calibration. One full day (1 January 2005) of global collection-5 MODIS and version-5 AIRS retrievals of cloud-top temperature Tc, effective cloud fraction f, and derived effective brightness temperature Tb,e is investigated. Comparisons of Tb,e demonstrate that MODIS and AIRS are essentially radiatively consistent and that MODIS Tb,e is 0.62 K higher than AIRS Tb,e for all scenes, increasing to 1.43 K for cloud described by AIRS as single layer and decreasing to 0.50 K for two-layer clouds. Somewhat larger differences in Tc and f are observed between the two instruments. The magnitudes of differences depend partly on whether MODIS uses a CO2-slicing or 11-μm brightness temperature window retrieval method. Some cloud- and regime-type differences and similarities between AIRS and MODIS cloud products are traceable to the assumptions made about the number of cloud layers in AIRS and also to the MODIS retrieval method. This (partially) holistic comparison approach should be useful for ongoing algorithm refinements, rigorous assessments of climate applicability, and establishment of the capability of synergistic MODIS and AIRS retrievals for improved cloud quantities and also should be useful for future observations to be made by the National Polar-Orbiting Operational Environmental Satellite System (NPOESS) Preparatory Project (NPP).

Corresponding author address: Dr. Shaima L. Nasiri, Dept. of Atmospheric Science, Texas A&M University, College Station, TX 77843-3150. E-mail: snasiri@tamu.edu

1. Introduction

The launch of the Earth Observing System Aqua satellite in May of 2002 initiated a multiyear record of spatially and temporally collocated high-spectral-resolution infrared radiances from the Atmospheric Infrared Sounder (AIRS; Aumann et al. 2003) and high-spatial-resolution visible-to-midinfrared radiances from the Moderate Resolution Imaging Spectroradiometer (MODIS; Barnes et al. 1998). With the continued operation of both instruments into the foreseeable future and the eventual follow-on of the Cross-Track Infrared Sounder (CrIS), the Advanced Technology Microwave Sounder (ATMS), and the Visible/Infrared Imager/Radiometer Suite (VIIRS) instruments on the National Polar-Orbiting Operational Environmental Satellite System (NPOESS) Preparatory Project (NPP), further advances in monitoring climate and quantifying atmospheric processes are expected from the synergistic use of high-spectral-resolution sounders and high-spatial-resolution imagers. However, the ability of each instrument to sample the physical state within some types of cloud regimes is still poorly understood (Fetzer et al. 2006, 2008) and must be compared further before the derived geophysical variables reach a nominal level of maturity for application to climate monitoring and trends.

Furthermore, before AIRS and MODIS can be combined together in joint retrievals of surface and/or atmospheric properties (e.g., L’Ecuyer et al. 2006; Li et al. 2005), a rigorous intercomparison of retrieval products common between the instruments (e.g., infrared-derived cloud temperature and amount) must be performed and the strengths and weaknesses of each instrument must be better understood. Most validation and intercomparison studies to date have focused in isolation on particular retrieval products such as cloud detection, cloud-top height and temperature, cloud amount, and, more recently, microphysical and optical properties (e.g., Wei et al. 2004; Yue and Liou 2009; Kahn et al. 2007a). Very few studies have taken a holistic multiparameter view to, for instance, quantify compensating errors between retrieval products or assess radiative consistency (e.g., Kahn et al. 2007b; Ham et al. 2009).

The AIRS and MODIS instruments have many similarities and differences. AIRS is a grating spectrometer with 2378 infrared channels, spectral resolution λλ of ~1200, and spectral coverage between 3.7 and 15.4 μm (with gaps from 4.6 to 6.2 μm and from 8.2 to 8.8 μm). The trade-off for AIRS’s high infrared spectral resolution is its nadir spatial resolution of 13.5 km. Throughout this paper, an observation by AIRS is referred to as a field of view (FOV). AIRS scans to ±49.5° off nadir, leading to 90 FOVs across track and 135 FOVs along track in a 6-min data granule. Many of the AIRS level-2 (L2) data products rely on information from the Advanced Microwave Sounder Unit A (AMSU-A), coaligned with AIRS on Aqua. There are 3 × 3 AIRS FOVs within an approximately 45-km AMSU-A FOV. The AIRS cloud products that are the focus of this paper, as well as the AIRS temperature and humidity soundings, are produced at the combined AMSU-A spatial resolution, which we refer to in this paper as the combined AIRS–AMSU-A field of regard (FOR).

MODIS is a 36-channel imaging radiometer. It has a higher spatial resolution of 1 km in the infrared (over 100 spatial footprints for each AIRS footprint), but only 16 of its 36 channels are in the infrared between 3.6 and 14.4 μm. MODIS scans to ±55° off nadir, leading to 1354 1-km FOVs across track and 2030 1-km FOVs along track in a 5-min data granule. The difference in swath width between MODIS and AIRS means that MODIS data at high viewing angles are not included in comparisons with AIRS. Several MODIS IR retrievals, such as the cloud properties in this study, are retrieved using 5 × 5 arrays of MODIS 1-km IR data. In this paper, these 5 × 5 MODIS retrievals are simply referred to as MODIS retrievals to avoid confusion with the AIRS FOVs and combined AIRS–AMSU-A FORs.

Infrared-based retrievals of cloud-top temperature Tc and effective cloud fraction f, also called effective emissivity, are individually derived from both MODIS and AIRS. Fundamental differences between the two instruments make comparisons of the retrieved cloud properties challenging, however. This challenge arises primarily from differences in the spectral and spatial resolutions of the two instruments; neither has both high spectral and high spatial resolution. Differences in instrument characteristics, observational capabilities, and primary mission objectives require fundamentally different cloud-property retrieval algorithms. The operational AIRS L2 retrieval is based on the cloud-clearing method (Chahine 1974; Susskind et al. 2003), which combines the infrared and microwave radiances of AIRS and AMSU-A, respectively. The MODIS retrieval algorithm is based on a chain of logic that starts with cloud detection and the use of thresholds (Ackerman et al. 1998; Frey et al. 2008) and then utilizes a combination of four carbon dioxide (CO2)-slicing channels and an 11-μm window retrieval (Platnick et al. 2003; Menzel et al. 2008).

Despite the inherent challenges in comparisons of cloud-top properties from AIRS and MODIS, cross-algorithm and cross-instrument comparisons are necessary for several reasons. First, because AIRS and MODIS view essentially identical scenes, pixel-scale cross comparisons are highly useful in assessing prelaunch calibration (Tobin et al. 2006; Schreier et al. 2010) and in assessing the retrieval-algorithm performance of both instruments. Second, reconciling the two datasets is a necessary (but by no means sufficient) requirement for inferring short-term weather variability from either instrument. Third, any long-term climate variability or trends in retrievals from one instrument can be tested by the other instrument, and a lack of consistency must be explained by known instrumental, algorithm, sensitivity, sampling, or calibration shortcomings. A few of those shortcomings (and associated strengths) are readily apparent in the comparison study described here. Last, given the absence of CO2-slicing channels on VIIRS for NPP, the ability of CrIS and ATMS to obtain cloud products at coarser spatial resolution must be assessed using MODIS and AIRS as a comparison baseline.

An initial set of comparisons between AIRS- and MODIS-retrieved IR cloud properties was made by Kahn et al. (2007b; henceforth abbreviated as K07). Working with a small set of granules (several minutes of observations each) from 6 September 2002, K07 compared MODIS and AIRS retrievals of Tc, f, and a quantity termed effective brightness temperature, or Tb,e. The Tb,e was defined to test the radiative consistency between MODIS and AIRS in the presence of clouds while taking into account the different spatial and vertical resolutions of the retrievals. In essence, Tb,e is an effective cloud fraction–weighted brightness temperature Tb. Agreement in Tb,e is a necessary, but not a sufficient, condition for agreement between MODIS and AIRS Tc and f retrievals. One shortcoming of K07 was the use of an approximate FOV representation of the AIRS footprint, which has been substantially improved with the application of AIRS spatial response functions (Schreier et al. 2010); the improved version is used in this work.

In this paper, comprehensive comparisons are presented between AIRS and MODIS IR retrievals of Tc, f, and Tb,e. In section 2, the cloud-property comparison method is introduced. Section 3 details the results of comparing a full day of AIRS and MODIS retrievals over the entire globe. Section 4 summarizes and discusses the conclusions of this work.

2. Method

Infrared-based cloud-property retrievals from the current MODIS collection 5 are obtained under the assumption of a single cloud layer within a 5 km × 5 km retrieval; thus, Tb,e for a single MODIS retrieval is defined as
e1
where the subscript 1 on the MODIS Tb,e denotes the single 5 km × 5 km retrieval, f is the effective cloud fraction of the MODIS cloud layer, Tc is the retrieved cloud-top temperature, and Ts is the surface temperature following the notation of K07. When comparing MODIS with AIRS , the MODIS single-retrieval Tb,e values within the AIRS–AMSU-A FOR must be averaged. Therefore, is at the AIRS–AMSU-A spatial resolution. In K07, the AIRS–AMSU-A FOR was approximated as being circular at near nadir and as an increasingly distorted ellipse with increasing scan angle, and each MODIS retrieval within the AIRS–AMSU-A FOR was weighted equally. As shown by Schreier et al. (2010), the K07 geolocation approach was imprecise and did not account for the dependence of the AIRS spatial response functions on AIRS channel and scan angle. In this work, we explicitly account for the AIRS spatial response functions.
Here, MODIS retrievals are collocated within individual AIRS FOVs following the technique described by Schreier et al. (2010). In a typical case, four MODIS cloud retrievals per AIRS FOV meet the collocation criteria. Because there are 3 × 3 AIRS FOVs within the AIRS–AMSU-A FOR, there are typically 36 MODIS retrievals within each AIRS–AMSU-A FOR. The value of is calculated by weighting each individual MODIS collocated within the AIRS–AMSU-A FOR by an AIRS spatial response function, represented here as w:
e2
AIRS retrieves properties for up to two cloud layers within a single FOR; therefore, is defined as
e3
where the subscripts 1 and 2 refer to the upper (1) and lower (2) cloud layers. It is important to note that when AIRS retrieves two cloud layers there is no information on whether those layers are horizontally separated or are overlapping geometrically.

K07 demonstrated that Tb,e is useful for testing the radiative consistency of cloud retrievals from different algorithms and/or instruments. The K07 results showed that, in general, compared well but with a noticeable warm bias for Tb,e between 270 and 295 K. K07 show that this warm AIRS bias is correlated with a high variability in retrieved MODIS Tc, overlapping cloud scenes, and scenes with a large fraction of MODIS retrievals made using the 11-μm Tb within the AIRS–AMSU-A FOR. Although Tb,e generally compared well between MODIS and AIRS in K07, the Tc, cloud-top pressure Pc, and f retrieval comparisons in isolation were not as promising, demonstrating the imprecise nature of the K07 geolocation for the former and the nature of compensating errors among Ts, Tc, and f for the latter. One reason for the lack of promise in MODIS and AIRS Tc comparisons is that K07 focused on comparing the MODIS retrievals averaged over the AIRS–AMSU-A FOR with each of the AIRS cloud retrievals separately. Because of this assumption, it is not surprising that, in general, AIRS upper-level clouds were higher, colder, and more transparent than the corresponding MODIS retrievals.

The AIRS and MODIS processing versions in this work are version 5 (V5) and collection 5 (C5), respectively, whereas the K07 comparisons were performed on AIRS version 4 (V4) and MODIS collection 4 (C4) data. One difference between AIRS V4 and V5 is in the quality-assessment information. K07 discuss in detail how the quality of the V4 MODIS and AIRS Tc and Tb,e comparisons depends strongly on the AIRS V4 “RetQAFlag,” which, in turn, depend on the cloud amount, opacity, and occurrence of precipitation within the field of regard (Susskind et al. 2006). As described by Menzel et al. (2008), the MODIS Tc retrieval algorithm underwent significant changes from C4 to C5. These modifications, including a radiance calibration adjustment and a top-down retrieval approach, have led to better agreement between MODIS and lidar cloud-top heights and, in general, higher cloud heights for thin clouds.

Other studies have shown that Tc retrievals tend to be more precise for opaque clouds with larger f. The K07 results show that comparisons between (upper layer) and are best for clouds with higher f that cover the entire FOR, whereas clouds with low f tend to be poorer. In basic terms, the more tenuous the cloud is, the poorer is the agreement between IR-derived cloud products and those derived from other IR or active profiling instruments (Kahn et al. 2007a,b, 2008; Holz et al. 2008). This is not a problem for only MODIS and AIRS (see Rossow et al. 1985); in general, larger compensating errors in geophysical products occur in tenuous clouds than in opaque clouds.

K07 demonstrated radiative consistency under cloudy-sky conditions for MODIS and AIRS, but there are several shortcomings to that study. The main limitation is the relatively small number of comparisons—only a few granules (roughly 30 min of observations from a single day in a record that now extends for nearly a decade)—and the lack of detailed statistics on correlations or biases. As a consequence, comparisons between MODIS-retrieved Tc and f were secondary. As important is that in K07 MODIS Tc, Pc, and f were compared with only the upper-layer AIRS properties. The more stringent comparisons between AIRS and MODIS retrievals for scenes described by AIRS as containing single-layer clouds were not quantified separately from all cloud scenes in K07. The comparisons were also made using the now obsolete AIRS V4 and MODIS C4 retrievals. Last, all MODIS retrievals within the AIRS–AMSU-A FOR were given equal weighting in K07 without taking into account the much more precise observational geometry determined from the AIRS spatial response functions obtained from prelaunch calibration activities (Overoye et al. 1999). Recent work using the AIRS spatial response functions has shown that, by considering the spatial contributions of the radiance within an individual AIRS FOV, significant reduction in the variability and skewness, but somewhat less so in the mean bias, is obtained for radiance differences between AIRS and MODIS (Schreier et al. 2010).

3. Comparisons

In this study, comparisons are made between collocated MODIS and AIRS cloud retrievals. These comparisons are significantly refined from the original K07 study in a few ways. First, as discussed earlier, the matching method of Schreier et al. (2010) is applied to AIRS and MODIS. Second, we will show that improved averaging of MODIS cloud properties within the AIRS–AMSU-A FOR allows much more robust comparisons between Tc and f as well as Tb,e. Last, the global comparison dataset is much larger than the one used in K07 and provides a more robust set of statistics. Comparisons were made for five separate full days in 2005. The overall conclusions are the same for all days, and therefore only results from 1 January 2005 are shown in this study. For these comparisons, all AIRS–AMSU-A cloud retrievals over the approximately 45-km (near nadir) FOR are compared with averages of MODIS retrievals within the AIRS–AMSU-A FOR using the method of Schreier et al. (2010). Individual MODIS retrievals within the AIRS–AMSU-A FOR are considered only when the effects of different cloud-property retrieval methods are relevant (e.g., CO2-slicing vs 11-μm Tb window methods).

a. Sorting by the number of AIRS cloud layers

The AIRS–MODIS comparison dataset is sorted into three categories: 1) all observations, 2) AIRS-retrieved single-layer clouds (henceforth referred to as single layer), and 3) AIRS-retrieved two-layer clouds (henceforth referred to as two layer). The all-observations category includes all matched FORs, including those with AIRS and/or MODIS f = 0 (clear sky), with AIRS retrieval quality-control variable “Qual_Cloud_OLR” < 2. Note that, for a few rare cases with fA − 1 = 1 × 10−6, fA has been set = 1. The single-layer category is (conservatively) defined to have , where 0.02 is the value below which it is assumed that some, but not all, cloudiness may be erroneous based on comparisons with the Cloud–Aerosol Lidar and Infrared Pathfinder Satellite Observation (CALIPSO) lidar (Kahn et al. 2008). The two-layer category has to ensure nearly all two-layer cases contain cloud in both layers. It is important to keep in mind that a two-layer AIRS scene does not necessarily mean that there are two cloud layers overlapping geometrically. There may be two horizontally separated cloud layers, or there may be cloud layers that partially overlay geometrically. Another possibility is that of a geometrically thick single-layered cloud with a tenuous upper part, such as nimbostratus (Kahn et al. 2008). In this situation, would be greater than . It is also possible for to be much greater than . Total effective cloud fraction for AIRS is defined to be .

AIRS and MODIS Tb,e are compared to quantify the overall radiative consistency between the cloud retrievals (Fig. 1). The left side of Fig. 1 shows versus for 1 January 2005. Very good agreement between MODIS and AIRS Tb,e is shown in Fig. 1a. The correlation between is high, with R = 0.978 and an average warm MODIS bias of 0.62 K. (Note that if we assume an average lapse rate of 7 K km−1 and f = 1.0 then this bias implies MODIS average cloud tops are less than 100 m below those of AIRS.) The high correlation and low bias imply that the quality of the radiative agreement between MODIS and AIRS for all observations is excellent and is much improved over K07. An interesting note is that, although MODIS has an overall slight warm bias relative to AIRS (or AIRS has a cold bias relative to MODIS—neither instrument is assumed to be the “truth,” and “bias” is used interchangeably), tends to be warmer for the small populations of points (dark blue color) at the warm and cold Tb,e extremes. In addition, a very small cold population with and near 280 K exists, but the rarity of these points is reflected in the mean bias of only 0.62 K. These points correspond to poor AIRS–AMSU-A retrieval quality in the presence of heavy precipitation (Kahn et al. 2007a).

Fig. 1.
Fig. 1.

Comparisons between retrieved MODIS and AIRS (left) effective brightness temperature and (right) cloud-top temperature for 1 Jan 2005. The color scale indicates the number of observations. The 1-to-1 line is shown in black, and the line of best fit is in red. The (a),(b) all-observations, (c),(d) single-layer, and (e),(f) two-layer data are based on AIRS effective cloud fraction as described in the text.

Citation: Journal of Applied Meteorology and Climatology 50, 5; 10.1175/2010JAMC2603.1

About 19% of the total comparisons fulfill the single-layer criteria. Because AIRS is the baseline for the number of cloud layers in this study, it is not surprising that a smaller number of collocated MODIS scenes match AIRS criteria. Because many single-layered clouds fall into the two-layer category for some cloud types (Kahn et al. 2008), the frequency of single-layer cases is expected to be an underestimate of actual single-layered cloud frequency (Mace et al. 2009). Table 1 shows the number of AIRS and MODIS scenes satisfying the single-layer and two-layer criteria, with each row representing a different range of f retrieved from each instrument. The single-layer points in Fig. 1c tend to have warm Tb,e values, as evidenced by the large concentration of values near 300 K. The single-layer uncertainties are described in detail in the appendix. In this region with warmer Tb,e, tends to be larger than , accounting for the majority of the 1.43-K warm MODIS bias in the single-layer category. Some of the warm MODIS bias near Tb,e = 300 K appears to be related to MODIS reliance upon IR window channels for retrievals in some cases; this will be discussed later in detail. That the majority of the single-layer category has high Tb,e values implies that AIRS most commonly finds only one cloud layer within the ~45-km FOR for either warm low clouds or optically thin high clouds over warm surfaces. In contrast to the single-layer Tb,e values, the two-layer cloud points (Fig. 1e) tend to have colder Tb,e values, with the largest frequency clustered within the range 240 ≤ Tb,e ≤ 260 K. It is found that 49.7% of both the AIRS and MODIS points fit into the two-layer category. The correlation is very high for these scenes, with R = 0.979 and a warm MODIS bias of only 0.5 K.

Table 1.

Percentage of AIRS and MODIS scenes meeting the single-layer and two-layer criteria. Whether a scene has one or two cloud layers is based on the AIRS effective cloud fraction for the upper and lower layers: , respectively. Single-layer means . Two-layer means . The effective cloud fraction column refers to total effective cloud fraction for either MODIS or AIRS. Percentages are based on total numbers of AIRS and MODIS points.

Table 1.

MODIS and AIRS Tb,e values are highly correlated with brightness temperature observation in the infrared window, as shown in Fig. 2 and Table 2. Figure 2a shows the relationship between the MODIS-observed brightness temperature at 11 μm [(11 μm)] and calculated for the individual MODIS retrievals. Note that (11 μm) is averaged over the 5 km × 5 km MODIS retrieval footprint. This comparison of more than 24 million data points has a correlation coefficient R of 0.994; on average, (11 μm) is colder than by 0.38 K. This level of agreement at the MODIS retrieval scale is not surprising because the MODIS retrieval algorithm requires agreement among observed 11-μm radiances, retrieved cloud-top pressure and effective cloud fraction, and calculated clear-sky radiance at 11 μm (Menzel et al. 2008). When MODIS (11 μm) is averaged over the AIRS–AMSU-A FOR and compared with computed on that scale, the agreement degrades only slightly, with a cold bias of 0.66 K, as shown in Fig. 2c. When (12 μm), averaged over the AIRS–AMSU-A FOR, is compared with (Fig. 2e), the cold bias increases to 1.46 K. These effects are likely due to increased sensitivity to atmospheric water vapor in the MODIS 12-μm channel. Figures 2b, 2d, and 2f compare three single-channel window-region brightness temperatures, (8.121 μm), (10.410 μm), and (11.668 μm), respectively, with . Although the AIRS retrieval also requires agreement between measured and calculated radiances, the amount of agreement shown in Fig. 2 is channel dependent.

Fig. 2.
Fig. 2.

Comparisons between measured brightness temperature and calculated effective brightness temperature for (left) MODIS and (right) AIRS: (a) (11 μm) vs for individual MODIS retrieval, (c) (11 μm) vs for AIRS–AMSU-A averaged MODIS values, (e) as in (c) but for (12 μm), (b) (8.121 μm) vs , (d) as in (b) but for (10.410 μm), and (f) as in (b) but for (11.668 μm).

Citation: Journal of Applied Meteorology and Climatology 50, 5; 10.1175/2010JAMC2603.1

Table 2.

Comparisons of measured brightness temperature with calculated effective brightness temperature for several MODIS and AIRS channels as shown in Fig. 2. The first row shows comparisons for MODIS single-retrieval values, and the rest compare averaged values at the AIRS–AMSU-A FOR.

Table 2.

MODIS and AIRS cloud retrievals show a high level of radiative consistency, but the agreement in Tc is not as good. The AIRS layer-averaged Tc is defined as
e4
where f1 is the effective cloud fraction of the upper layer and Tc1 is the retrieved cloud-top temperature of the upper layer. If AIRS reports two layers, it is impossible to directly compare them with the single layer reported from MODIS. Thus, is weighted by each layer’s f. For those cases that fulfill the single-layer AIRS cloud criteria, f2 is assumed to be 0 in the computation of . Because our definition of single-layer clouds allows a second AIRS cloud layer as long as f2 ≤ 0.02, there may be a small contribution to from a second cloud layer, even in single-layer cases.
Although MODIS only retrieves single-layer cloud properties per individual retrieval, there are typically 36 MODIS cloud retrievals within the AIRS–AMSU-A FOR. Because this means that MODIS can, in theory, have information on up to 36 cloud layers, we focus on comparing AIRS-averaged cloud-top temperatures with MODIS-averaged cloud top temperatures. This is a substantial change from the comparison method used in K07. In this study, the MODIS cloud-top temperature is a FOR-averaged value defined as
e5
where, as in Eq. (2), the subscript j refers to an individual MODIS retrieval and wj is the spatial weighting of the MODIS retrieval within the AIRS–AMSU-A FOR. We investigated the effect of including in the numerator and denominator of Eq. (5) but found that it had little impact on the calculation of averaged MODIS Tc within the AIRS–AMSU-A FOR.

Comparisons of MODIS and AIRS Tc retrievals are shown for the all-observations (Fig. 1b), single-layer (Fig. 1d), and two-layer (Fig. 1f) categories. AIRS layer-averaged Tc and MODIS averaged Tc for all observations reveal a warm MODIS bias of 5.25 K and R = 0.78 (Fig. 1b). Of interest is that MODIS Tc is nearly always warmer than AIRS Tc for the single-layer case (Fig. 1d), having a warm MODIS bias of 12.3 K and R = 0.80. The most obvious feature of the plot of single-layer versus is the red bull’s-eye centered near and . This shows that the warm in Fig. 1c correspond, in general, to warm clouds, rather than to thin cold clouds. The Tc agreement for the two-layer category (Fig. 1f) is much better than for single-layer cases, and, due in part to compensating errors, the bias is <0.01 K.

Although the plots in Fig. 1 demonstrate the general similarities and differences of the Tb,e and Tc populations for the three categories, the impact of the lower population regions is difficult to estimate visually. To aid in quantifying the differences described above, the cumulative distributions functions for are plotted in Fig. 3a for the same three cases as in Fig. 1. Figure 3a shows that the vast majority of points in all three cases have −5 < ΔTb,e < 5 K (~90% of the two-layer cases, ~85% of the all-data cases, and ~78% of the single-layer cases) and that over 95% of the points have −10 < ΔTb,e < 10 K. In addition, Fig. 3a shows that the warm MODIS bias is small in magnitude but is greatest for the single-layer cases.

Fig. 3.
Fig. 3.

Cumulative distribution functions for MODIS–AIRS (a) effective brightness temperature, (b) retrieved cloud-top temperature, and (c) assumed surface temperature.

Citation: Journal of Applied Meteorology and Climatology 50, 5; 10.1175/2010JAMC2603.1

Taken together, Figs. 1 and 3a demonstrate the high degree of radiative consistency between MODIS and AIRS cloud retrievals despite the significant differences in instrument spatial and spectral resolutions, surface temperature assumptions, effective emissivity determination algorithms, cloud detection criteria, and cloud-height assignment. Differences in measured clear radiances between MODIS and AIRS are generally within 0.1 K on the basis of the values determined by Tobin et al. (2006) and Schreier et al. (2010).

Although ΔTb,e in Fig. 3a is generally near zero, differences in retrieved Tc can be substantially larger. Figure 3b shows the cumulative distribution functions for the cloud-top temperature differences between MODIS and AIRS. Figure 3 shows −5 < ΔTc < 5 K in just over 50% of the two-layer cases, in approximately 45% of the all-observations cases, and in only 25% of the single-layer cases. Between 5% and 20% of the cases have ΔTc > 20 K, with the greatest differences occurring for single-layer cases. Figure 3b shows that approximately 50% of the two-layer cases show MODIS with warmer Tc than AIRS and more than 80% of the single-layer cases show MODIS warmer than AIRS.

The differences in Tc and f between MODIS and AIRS have the largest impacts on ΔTb,e, but the effects of Ts have also been investigated. In MODIS collection 5, Ts are obtained from a blend of National Centers for Environmental Prediction Global Forecast System model output and Reynolds sea surface temperature analyses (Menzel et al. 2008), whereas in AIRS version 5, Ts are retrieved along with atmospheric profile and cloud properties as part of the cloud-clearing process (Susskind et al. 2003). The discrepancies in Ts are similar to those in Tc but are somewhat reduced in magnitude (Fig. 3b). The absolute value of the Ts differences between MODIS and AIRS is greater than 5 K in more than 20% of the cases. MODIS Ts values tends to be warmer than those from AIRS, and there is a larger difference between MODIS and AIRS Ts for two-layer cases than for single-layer cases. This is consistent with cool AIRS Ts biases related to the more complex cloud structure with which cloud clearing has difficulties (Susskind et al. 2006) and the complicated overlying structures of temperature and water vapor, such as temperature inversions, that further complicate the retrieval scene (Fetzer et al. 2004; Dong et al. 2010).

b. Sorting by MODIS cloud retrieval method

The magnitude of ΔTc appears to be related in part to MODIS’s dependence on two different retrieval techniques: the CO2-slicing method and the 11-μm-Tb window method. The window method is employed when there is not enough contrast between CO2-channel radiance pairs, such as when cloud tops are below 700 hPa (Menzel et al. 2008). The window method proceeds under the assumption of an opaque cloud. Holz et al. (2008) found that, for transmissive high clouds, MODIS cloud-top-height retrievals using the window method can be over 6 km higher than those from CALIPSO. In these cases, the opaque cloud assumption will cause these retrieved cloud-top temperatures to be too warm.

The effects of these two MODIS methods on ΔTc are now examined. Figure 4 shows versus for the all-data, single-layer, and two-layer cases (similar to Fig. 1) but is parsed into data points for which MODIS cloud retrievals rely on either the slicing or window methods for 60% or more of the MODIS retrievals within the AIRS–AMSU-A FOR. In comparing ΔTc in Fig. 1 to Fig. 4, one sees that many of the prominent features in ΔTc are related to a particular method. For example, the “hook” at warm temperatures in Fig. 1b is due to window-method retrievals, primarily from single-layer cases. (See Table 3 for the percentage of data points fitting each criterion, as well as their correlations and biases.)

Fig. 4.
Fig. 4.

Cases in which ≥60% of the MODIS retrievals within the AIRS FOR used the (left) CO2-slicing and (right) window retrieval methods for (a),(b) all-observations, (c),(d) single-layer, and (e),(f) two-layer data.

Citation: Journal of Applied Meteorology and Climatology 50, 5; 10.1175/2010JAMC2603.1

Table 3.

Percentage of comparisons, out of the total number of AIRS cases (Table 1), and basic statistics of retrieved Tc when 60% of individual MODIS retrievals within AIRS–AMSU-A FOR use either the CO2-slicing or window method. These values correspond to Fig. 4.

Table 3.

One obvious feature in Fig. 1b is that while a large number of Tc fall along the 1-to-1 line there is a large region with a low population of . This feature mainly occurs in single-layer cases (Fig. 1d) in which the window method was used (Fig. 4d). The window method applied in the presence of two-layer clouds (as defined by AIRS) gives much better agreement (Fig. 4f). More scenes are near the 1-to-1 line for between approximately 240 and 260 K. In addition, MODIS retrievals tend to be warmer than AIRS when the window method predominates because MODIS sets fM to unity in these cases while AIRS does not. It is highly likely that MODIS assigns a much more realistic distribution of Tc over these low cloud types than AIRS does (Garay et al. 2008; Kahn et al. 2008). MODIS predominantly uses the slicing method when Tc is usually <270 K. Another interesting feature is that nearly all of the points with occur when the slicing method predominates. Although not shown here, the features discussed above are still robust when the method fraction within the AIRS–AMSU-A FOR is increased from 60% to 90%. Figure 4 underlines an important point: the agreement between MODIS and AIRS Tc retrievals is not determined solely by whether MODIS uses the window retrieval method or the slicing retrieval method but also depends on the cloud type and the number of cloud layers present, as well as the magnitude of Tc, f, and Tb,e.

Retrievals of Tc are widely used, but retrievals of f are less frequently used and are not understood well. To achieve the level of radiative consistency between MODIS and AIRS retrievals already shown, differences in retrieved f must balance large differences in retrieved ΔTc. Figure 5 shows histograms of , fA, , and fM for the single-layer case. The AIRS Tc distribution shows expected peaks for high cold clouds (e.g., Comstock et al. 2002), midtemperature or midlevel clouds (e.g., Johnson et al. 1999), and warm low clouds although they show somewhat of a cold bias (e.g., Kahn et al. 2008). The MODIS Tc distribution is very different from the AIRS distribution, with many fewer cold clouds and an increasing frequency with temperature. The mean of the retrievals is 263.1 K, and the mean of the retrievals is 251.3 K. Most of the single-layer AIRS scenes have low f, implying either thin or broken clouds, or both. The fA frequency decreases with increasing cloud frequency until fA = 0.9; then, there is a spike in frequency at fA = 1. The drop between fA = 0.9 and 1.0 is due to the AIRS algorithm defaulting to fA = 1 if the effective cloud fractions of all nine of the AIRS FOVs within the AIRS–AMSU-A FOR are >0.9. On the other hand, the fM frequency is bimodal with peaks at 0 and 1 and is fairly evenly distributed for values between the peaks. The spike in fM at 0 occurs because our definition of cloudiness for this study is based on fA and AIRS detects some thin cirrus that MODIS does not (Kahn et al. 2008). The mean of the fM retrievals is 0.57, and the mean of the fA retrievals is 0.42. Figure 5 demonstrates that the MODIS and AIRS f distributions are largely different, especially for the single-layer case, and compensate Tc such that Tb,e are in much better agreement than either Tc or f alone.

Fig. 5.
Fig. 5.

Histograms of (a) , (b) fA, (c) , and (d) fM are shown for the single-layer case.

Citation: Journal of Applied Meteorology and Climatology 50, 5; 10.1175/2010JAMC2603.1

The dependence of ΔTc on f is shown in cumulative distribution plots in Fig. 6. These plots clearly show that ΔTc is strongly related to fA and is to a lesser extent related to fM. The number of data points in each f bin is listed in Table 1. The single-layer and two-layer cases show for most fA bins, except for two-layer cases with fA ≥ 0.75. This effect can also be seen in the all-observations case since two-layer clouds are 2.5 times as frequent as single-layer clouds in AIRS retrievals. The warm bias of increases with decreasing fA. Between single-layer and two-layer cases, effective cloud fraction varies significantly: fA < 0.25 for 33% of single-layer cases, but fA < 0.25 for less than 1% of two-layer cases. The dependence on fM is weak, although reduced ΔTc is found for fM ≥ 0.75. Larger fractions of data points fall into the fM ≥ 0.75 category than into the fA ≥ 0.75 category, due in part to the assignment of fM = 1 when the window method is used. Effects of the layer averaging of retrievals can also be seen for two layers. The compares better to the upper-layer AIRS, Tc1, when but better to the AIRS Tc2 when .

Fig. 6.
Fig. 6.

Cumulative distribution plots of the dependence of on f for (a),(b) all-observations, (c),(d) single-layer, and (e),(f) two-layer data. (left) Curves vary with fA; (right) curves vary with fM.

Citation: Journal of Applied Meteorology and Climatology 50, 5; 10.1175/2010JAMC2603.1

Although fA strongly affects the magnitude of ΔTc, it only has a small effect on the agreement in Tb,e (Fig. 7). The most noticeable differences in f are only seen when fA < 0.25 for single-layer cases (6.5% of AIRS data) and fM < 0.25 for two-layer cases (3.8% of MODIS data). There is a small dependence on f such that for the all-observations case the cumulative distribution functions tend to be nearly symmetric for f ≥ 0.75, and MODIS shows a warm Tb,e bias for f < 0.75.

Fig. 7.
Fig. 7.

As in Fig. 6, but for MODIS − AIRS effective brightness temperature.

Citation: Journal of Applied Meteorology and Climatology 50, 5; 10.1175/2010JAMC2603.1

4. Discussion and conclusions

Comparisons are described for infrared-derived cloud products retrieved from the collection-5 Moderate Resolution Imaging Spectraradiometer (Barnes et al. 1998) and the version-5 Atmospheric Infrared Sounder (Aumann et al. 2003) algorithms using the measured spatial response functions of the AIRS instrument obtained from prelaunch calibration as the basis for collocation (Schreier et al. 2010). This study compares one full day (1 January 2005) of global MODIS and AIRS retrievals of cloud-top temperature Tc, effective cloud fraction f, and the derived effective brightness temperature Tb,e (Kahn et al. 2007b) for the two instruments. MODIS and AIRS are both on the Aqua platform and observe the same scenes simultaneously, but differences in spectral resolution, spatial resolution, and cloud remote sensing algorithms mean that the two instruments retrieve Tc and f values that are not necessarily the same for all cloud types and scenes. The approach taken here is to compare AIRS layer-averaged cloud retrievals with MODIS retrievals that have been averaged over the much larger AIRS–AMSU-A footprint.

A warm AIRS Tb,e bias for warm Tb,e scenes was found by Kahn et al. (2007b), with significant scatter in Tb,e noted. This bias is substantially reduced along with the scatter—a result that is mostly attributable to the improved collocation approach. Our results show a warm, but much smaller, MODIS Tb,e bias of 0.62 K. This bias increases to 1.43 K when only single-layer clouds, as determined by AIRS, are considered and decreases to 0.50 K for two-layer clouds. A small warm AIRS Tb,e bias can be seen for a subset of clouds that are cold and opaque. Because collection 4 products are no longer available, it was not determined whether the overall sign change in the biases is due to the small dataset used in Kahn et al. (2007b) or is due to fundamental differences between the C4/V4 and C5/V5 retrievals. Much tighter agreement in Tb,e is apparent in this work, however, when compared with Kahn et al. (2007b).

In addition, MODIS and AIRS Tb,e values are highly correlated with brightness temperature observations in the infrared window, but the amount of agreement is spectrally dependent. MODIS Tb,e compares best to at 11 μm at the individual retrieval scale, which is not surprising because of the nature of the MODIS retrieval algorithm. Differences increase when averaging up to the AIRS–AMSU-A scale and when comparing with at 12 μm. The best comparison to AIRS Tb,e occurs for observations of at 11.668 μm.

Regardless, the Tb,e comparisons demonstrate that the MODIS and AIRS infrared-based cloud products are essentially radiatively consistent, but there are somewhat larger differences in Tc and f between the two instruments. The magnitudes of the differences are strongly related to whether MODIS uses a CO2-slicing or 11-μm brightness temperature window method retrieval. On one hand, MODIS is known to default inappropriately sometimes to the window method in the case of clouds that are both cold and optically thin. On the other hand, the AIRS retrieval algorithm has been shown to yield biased Tc and f in the presence of some cloud types. For example, the AIRS single-layer cases tend to occur in stratus and stratocumulus regimes. Kahn et al. (2008) show that in these cloud regimes AIRS tends to place clouds too high. There are three important points regarding Tc comparisons as a function of MODIS retrieval method: 1) MODIS and AIRS average Tc retrievals do not always disagree when the window method is attempted, 2) when there is disagreement for single-layer clouds, the window method is typically used, and 3) MODIS Tc tends to be warmer than AIRS Tc when the window method is used but the reverse is true when the CO2-slicing method is used.

Differences between retrieved MODIS and AIRS Tc depend strongly on AIRS f but only weakly on MODIS f. This lack of dependence on MODIS f is likely due to impacts of the opaque cloud assumption associated with the window retrievals. The relative lack of dependence of the Tb,e retrievals on f further demonstrates that differences in Tc and f compensate to produce very tight radiative agreement in most situations.

There are several reasons to quantitatively cross compare MODIS and AIRS infrared cloud products. Comparisons can demonstrate impacts of algorithm refinement. Comparing the results of this work with K07, we see that the algorithm changes to cloud retrieval algorithms between MODIS collections 4 and 5 and AIRS versions 4 and 5 resulted in a change of sign in the MODIS–AIRS Tb,e bias. Comparisons can also help in the interpretation of results when MODIS and AIRS cloud products are used in scientific studies. For example, the cloudy-sky radiative consistency demonstrated here implies that top-of-the-atmosphere radiative studies using MODIS and AIRS cloud products might be assumed to be similar, yet the results of cloud process studies and/or cloud climatological studies might differ more. In addition, understanding in which situations MODIS and AIRS cloud retrieval agree and disagree will help in the development of synergistic MODIS–AIRS cloud products such as ice crystal size and habit distribution (e.g., Baran and Francis 2004). Furthermore, the potential of the CrIS and ATMS to obtain similar products not available from the VIIRS instrument on NPP must be quantified because NPP VIIRS will not have CO2-slicing channels and therefore potentially inferior cloud products will be produced.

These results show that there are some cloud- and regime-type differences and similarities between AIRS and MODIS that are traceable to the assumptions made about single- and two-layered clouds in AIRS and also to the method of retrieval (slicing or window methods). Despite the cloud- and regime-type similarities and differences in the individual cloud products, the radiative consistency of the retrievals from the two instruments is very high. This demonstrates that each instrument contributes certain strengths and weaknesses to cloud quantification and also suggests that further improvements in infrared-based cloud products can be attained with advancements in multiple-instrument cloud retrieval algorithms.

Acknowledgments

BHK and MMS were partially funded by NASA Award NNX08AI09G. MODIS data were obtained through the Level-1 and Atmosphere Archive and Distribution System (LAADS; online at http://ladsweb.nascom.nasa.gov/). AIRS data were obtained through the Goddard Earth Sciences Data and Information Services Center (online at http://daac.gsfc.nasa.gov). A portion of this work was performed within the Joint Institute for Regional Earth System Science and Engineering (JIFRESSE) of the University of California, Los Angeles (UCLA), and at the Jet Propulsion Laboratory, California Institute of Technology, under contract with NASA.

APPENDIX

Details of the Single-Layer Uncertainties

The high correlation between AIRS and MODIS Tb,e (Fig. 1) in both single- and multilayered cases is largely a result of very good radiative fits between the simulated and observed radiances, as shown in Fig. 2. It does not follow that the cloud products themselves agree as well individually (as shown in Figs. 1, 3, and 4). To illustrate this point further, two types of error perturbation experiments applied to single-layered clouds were performed. First, to quantify the impact in assumed biases from each term in Eq. (3), we perturbed Tc by ±7 K (equivalent to 1 km assuming a lapse rate of 7 K km−1), Ts by ±2 K, and f by ±0.1 individually. These values are somewhat conservative, perhaps by a factor of 2 or more, when compared with validation studies of AIRS and MODIS Tc biases (e.g., Kahn et al. 2008; Holz et al. 2008) and AIRS Ts (Dong et al. 2010). Second, the same quantities were perturbed in a randomization experiment with the values above set to 1 standard deviation of a normal distribution, where each point in Fig. 1a was randomly perturbed once. In either case, f is constrained to remain within the range 0.0–1.0. The results are shown in Table A1.

Table A1.

“Error” experiments for Ts, Tc, and f limited to single-layered cases. The first six experiment rows perturb each quantity individually by a fixed bias, either negatively or positively. The last row assumes that these errors are 1-σ values of a normal distribution and are randomly, and simultaneously, chosen.

Table A1.

The changes among the three physical quantities in the bias experiment are significantly different from each other. For instance, the average bias for the baseline single-layer case is +1.42 K. When perturbing Tc, the bias is changed by approximately ±2.9 K; similar results are found for f. In the case of Ts, the change is smaller and is on the order of ±1.2 K. The standard deviation of the absolute average difference (third column in Table A1) is marginally smaller than that of the baseline case only for the Ts = −2 K perturbation experiment, whereas worse agreement is shown between AIRS and MODIS cloud products in all other cases. The fairly large values of Tb,e bias shown in Table A1 for conservative estimates of Tc, Ts, and f biases suggest that these three parameters are largely compensating to yield good fits in the cloud retrievals. This is highlighted further in the randomization experiment shown in the last row of Table A1. The correlation is substantially reduced, and all other columns indicate worse agreement except for the average bias (which is expected because no bias was added).

These results also suggest that they are, to leading order, the three most important parameters for obtaining agreement in AIRS and MODIS Tb,e. In Kahn et al. (2007a), the AIRS standard L2 product uncertainties that are reported for Tc show qualitative consistency with absolute differences between AIRS and Atmospheric Radiation Measurement Program active cloud radar observations. The absolute magnitudes of these uncertainties remain unvalidated, however, and there is some suggestion that they are overestimated in thin clouds and underestimated in opaque clouds (cf. Figs. 1 and 4 in Kahn et al. 2007a). It is not clear whether these uncertainties should be treated as systematic biases, random error, or some combination thereof. More robust error characterization may come from advancements in, for example, optimal estimation retrievals of geophysical parameters from AIRS radiances. In summary, the two error perturbation experiments described here show that the uncertainties in Tc, Ts, and f are highly correlated with each other and compensate to produce radiatively consistent Tb,e—a finding that is also supported by the very good agreement between Tb,e and Tb in Fig. 2.

REFERENCES

  • Ackerman, S., , K. Strabala, , W. Menzel, , R. Frey, , C. Moeller, , and L. Gumley, 1998: Discriminating clear sky from clouds with MODIS. J. Geophys. Res., 103, 32 14132 157.

    • Search Google Scholar
    • Export Citation
  • Aumann, H. H., and Coauthors, 2003: AIRS/AMSU/HSB on the Aqua mission: Design, science objectives, data products and processing systems. IEEE Trans. Geosci. Remote Sens., 41, 253264.

    • Search Google Scholar
    • Export Citation
  • Baran, A. J., , and P. N. Francis, 2004: On the radiative properties of cirrus cloud at solar and thermal wavelengths: A test of model consistency using high-resolution airborne radiance measurements. Quart. J. Roy. Meteor. Soc., 130, 763778.

    • Search Google Scholar
    • Export Citation
  • Barnes, W. L., , T. S. Pagano, , and V. V. Salomonson, 1998: Prelaunch characteristics of the Moderate Resolution Imaging Spectroradiometer (MODIS) on EOS-AM1. IEEE Trans. Geosci. Remote Sens., 36, 10881100.

    • Search Google Scholar
    • Export Citation
  • Chahine, M. T., 1974: Remote sounding of cloudy atmospheres. I. The single cloud layer. J. Atmos. Sci., 31, 233243.

  • Comstock, J. M., , T. P. Ackerman, , and G. G. Mace, 2002: Ground-based lidar and radar remote sensing of tropical cirrus clouds at Nauru Island: Cloud statistics and radiative impacts. J. Geophys. Res., 107, 4714, doi:10.1029/2002JD002203.

    • Search Google Scholar
    • Export Citation
  • Dong, S., , S. T. Gille, , J. Sprintall, , and E. J. Fetzer, 2010: Assessing the potential of the Atmospheric Infrared Sounder (AIRS) surface temperature and specific humidity in turbulent heat flux estimates in the Southern Ocean. J. Geophys. Res., 115, C05013, doi:10.1029/2009JC005542.

    • Search Google Scholar
    • Export Citation
  • Fetzer, E. J., , J. Teixeira, , E. T. Olsen, , and E. F. Fishbein, 2004: Satellite remote sounding of atmospheric boundary layer temperature inversions over the subtropical eastern Pacific. Geophys. Res. Lett., 31, L17102, doi:10.1029/2004GL020174.

    • Search Google Scholar
    • Export Citation
  • Fetzer, E. J., , B. H. Lambrigtsen, , A. Eldering, , H. H. Aumann, , and M. T. Chahine, 2006: Biases in total precipitable water vapor climatologies from Atmospheric Infrared Sounder and Advanced Microwave Scanning Radiometer. J. Geophys. Res., 111, D09S16, doi:10.1029/2005JD006598.

    • Search Google Scholar
    • Export Citation
  • Fetzer, E. J., and Coauthors, 2008: Comparison of upper tropospheric water vapor observations from the Microwave Limb Sounder and Atmospheric Infrared Sounder. J. Geophys. Res., 113, D22110, doi:10.1029/2008JD010000.

    • Search Google Scholar
    • Export Citation
  • Frey, R. A., , S. A. Ackerman, , Y. Liu, , K. I. Strabala, , H. Zhang, , J. R. Key, , and X. Wang, 2008: Cloud detection with MODIS. Part I: Improvements in the MODIS cloud mask for collection 5. J. Atmos. Oceanic Technol., 25, 10571072.

    • Search Google Scholar
    • Export Citation
  • Garay, M. J., , S. P. de Szoeke, , and C. M. Moroney, 2008: Comparison of marine stratocumulus cloud top heights in the southeastern Pacific retrieved from satellites with coincident ship‐based observations. J. Geophys. Res., 113, D18204, doi:10.1029/2008JD009975.

    • Search Google Scholar
    • Export Citation
  • Ham, S. H., , B. J. Sohn, , P. Yang, , and B. A. Baum, 2009: Assessment of the quality of MODIS cloud products from radiance simulations. J. Appl. Meteor. Climatol., 48, 15911612.

    • Search Google Scholar
    • Export Citation
  • Holz, R. E., , S. A. Ackerman, , F. W. Nagle, , R. Frey, , S. Dutcher, , R. E. Kuehn, , M. A. Vaughan, , and B. Baum, 2008: Global Moderate Resolution Imaging Spectroradiometer (MODIS) cloud detection and height evaluation using CALIOP. J. Geophys. Res., 113, D00A19, doi:10.1029/2008JD009837.

    • Search Google Scholar
    • Export Citation
  • Johnson, R. H., , T. M. Richenbach, , S. A. Rutledge, , P. E. Ciesielski, , and W. H. Schubert, 1999: Trimodal characteristics of tropical convection. J. Climate, 12, 23972418.

    • Search Google Scholar
    • Export Citation
  • Kahn, B. H., , A. Eldering, , A. J. Braverman, , E. J. Fetzer, , J. H. Jiang, , E. Fishbein, , and D. L. Wu, 2007a: Toward the characterization of upper tropospheric clouds using Atmospheric Infrared Sounder and Microwave Limb Sounder observations. J. Geophys. Res., 112, D05202, doi:10.1029/2006JD007336.

    • Search Google Scholar
    • Export Citation
  • Kahn, B. H., , E. Fishbein, , S. L. Nasiri, , A. Eldering, , E. J. Fetzer, , M. J. Garay, , and S.-Y. Lee, 2007b: The radiative consistency of Atmospheric Infrared Sounder and Moderate Resolution Imaging Spectroradiometer cloud retrievals. J. Geophys. Res., 112, D09201, doi:10.1029/2006JD007486.

    • Search Google Scholar
    • Export Citation
  • Kahn, B. H., and Coauthors, 2008: Cloud-type comparisons of AIRS, CloudSat, and CALIPSO cloud height and amount. Atmos. Chem. Phys., 8, 12311248.

    • Search Google Scholar
    • Export Citation
  • L’Ecuyer, T. S., , P. Gabriel, , K. Leesman, , S. J. Cooper, , and G. L. Stephens, 2006: Objective assessment of the information content of visible and infrared radiance measurements for cloud microphysical property retrievals over the global oceans. Part I: Liquid clouds. J. Appl. Meteor. Climatol., 45, 2041.

    • Search Google Scholar
    • Export Citation
  • Li, J., and Coauthors, 2005: Retrieval of cloud microphysical properties from MODIS and AIRS. J. Appl. Meteor., 44, 15261543.

  • Mace, G. G., , Q. Zhang, , M. Vaughan, , R. Marchand, , G. Stephens, , C. Trepte, , and D. Winker, 2009: A description of hydrometeor layer occurrence statistics derived from the first year of merged CloudSat and CALIPSO data. J. Geophys. Res., 114, D00A26, doi:10.1029/2007JD009755.

    • Search Google Scholar
    • Export Citation
  • Menzel, W. P., and Coauthors, 2008: MODIS global cloud-top pressure and amount estimation: Algorithm description and results. J. Appl. Meteor. Climatol., 47, 11751198.

    • Search Google Scholar
    • Export Citation
  • Overoye, K., , H. H. Aumann, , M. H. Weiler, , G. W. Gigioli, , W. Shaw, , E. Frost, , and T. McKay, 1999: Test and calibration of the AIRS instrument. Infrared Spaceborne Remote Sensing VII, M. Strojnik and B. F. Andresen, Eds., International Society for Optical Engineering (SPIE Proceedings Vol. 3759), 254–265.

    • Search Google Scholar
    • Export Citation
  • Platnick, S., , M. D. King, , S. A. Ackerman, , W. P. Menzel, , B. A. Baum, , J. C. Riédi, , and R. A. Frey, 2003: The MODIS cloud products: Algorithms and examples from Terra. IEEE Trans. Geosci. Remote Sens., 41, 459473.

    • Search Google Scholar
    • Export Citation
  • Rossow, W. B., and Coauthors, 1985: ISCCP cloud algorithm intercomparisons. J. Climate Appl. Meteor., 24, 877903.

  • Schreier, M. M., , B. H. Kahn, , A. Eldering, , D. A. Elliott, , E. Fishbein, , F. W. Irion, , and T. S. Pagano, 2010: Radiance comparisons of MODIS and AIRS using spatial response information. J. Atmos. Oceanic Technol., 27, 13311342.

    • Search Google Scholar
    • Export Citation
  • Susskind, J., , C. D. Barnet, , and J. M. Blaisdell, 2003: Retrieval of atmospheric and surface parameters from AIRS/AMSU/HSB data in the presence of clouds. IEEE Trans. Geosci. Remote Sens., 41, 390409.

    • Search Google Scholar
    • Export Citation
  • Susskind, J., , C. D. Barnet, , J. M. Blaisdell, , L. Iredell, , F. Keita, , L. Kouvaris, , G. Molnar, , and M. Chahine, 2006: Accuracy of geophysical parameters derived from Atmospheric Infrared Sounder/Advanced Microwave Sounding Unit as a function of fractional cloud cover. J. Geophys. Res., 111, D09S17, doi:10.1029/2005JD006272.

    • Search Google Scholar
    • Export Citation
  • Tobin, D. C., , H. E. Revercomb, , C. C. Moeller, , and T. S. Pagano, 2006: Use of AIRS high spectral resolution spectra to assess the calibration of MODIS on EOS Aqua. J. Geophys. Res., 111, D09S05, doi:10.1029/2005JD006095.

    • Search Google Scholar
    • Export Citation
  • Wei, H., , P. Yang, , J. Li, , B. A. Baum, , H.-L. Huang, , S. Platnick, , Y. Hu, , and L. Strow, 2004: Retrieval of semitransparent ice cloud optical thickness from Atmospheric Infrared Sounder (AIRS) measurements. IEEE Trans. Geosci. Remote Sens., 42, 22542267.

    • Search Google Scholar
    • Export Citation
  • Yue, Q., , and K. N. Liou, 2009: Cirrus cloud optical and microphysical properties determined from AIRS infrared spectra. Geophys. Res. Lett., 36, L05810, doi:10.1029/2008GL036502679.

    • Search Google Scholar
    • Export Citation
Save