• Ackerman, S. A., , K. I. Strabala, , W. P. Menzel, , R. A. Frey, , C. C. Moeller, , and L. E. Gumley, 1998: Discriminating clear sky from clouds with MODIS. J. Geophys. Res., 103D , 3214132157.

    • Search Google Scholar
    • Export Citation
  • Ackerman, S. A., , R. E. Holz, , R. Frey, , E. W. Eloranta, , B. C. Maddux, , and M. McGill, 2008: Cloud detection with MODIS. Part II: Validation. J. Atmos. Oceanic Technol., 25 , 10731086.

    • Search Google Scholar
    • Export Citation
  • Beesley, J. A., , and R. E. Moritz, 1999: Toward an explanation of the annual cycle of cloudiness over the Arctic Ocean. J. Climate, 12 , 395415.

    • Search Google Scholar
    • Export Citation
  • Birch, C. E., , I. M. Brooks, , M. Tjernstrom, , S. F. Milton, , P. Earnshaw, , S. Soderberg, , and P. O. G. Persson, 2009: The performance of a global and mesoscale model over the central Arctic Ocean during late summer. J. Geophys. Res., 114 , D13104. doi:10.1029/2008JD010790.

    • Search Google Scholar
    • Export Citation
  • Cavalieri, D., , M. Thorsten, , and J. Comiso, cited. 2004: AMSR-E/Aqua daily L3 25-km brightness temperature and SIC polar grids V002, July 2006 to December 2008. National Snow and Ice Data Center. [Available online at http://nsidc.org/data/ae_si25.html].

    • Search Google Scholar
    • Export Citation
  • Comiso, J. C., , C. L. Parkinson, , R. Gersten, , and L. Stock, 2008: Accelerated decline in the Arctic sea ice cover. Geophys. Res. Lett., 35 , L01703. doi:10.1029/2007GL031972.

    • Search Google Scholar
    • Export Citation
  • Curry, J. A., , W. B. Rossow, , D. Randall, , and J. L. Schramm, 1996: Overview of Arctic cloud and radiation characteristics. J. Climate, 9 , 17311764.

    • Search Google Scholar
    • Export Citation
  • Deser, C., , and H. Teng, 2008: Evolution of Arctic sea ice concentration trends and the role of atmospheric circulation forcing, 1979-2007. Geophys. Res. Lett., 35 , L02504. doi:10.1029/2007GL032023.

    • Search Google Scholar
    • Export Citation
  • Eloranta, E. W., , J. P. Garcia, , I. A. Razenkov, , T. Uttal, , and M. Shupe, 2008: Cloud fraction statistics derived from 2-years of high spectral resolution lidar data acquired at Eureka, Canada. Proc. 24th Int. Laser Radar Conf., Boulder, CO, NCAR and NOAA, 555–558.

    • Search Google Scholar
    • Export Citation
  • Evan, A. T., , Y. Liu, , and B. Maddux, 2008: Global cloudiness. Bull. Amer. Meteor. Soc., 89 , (Suppl.). S23S26.

  • 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
  • Gao, B. C., , W. Han, , S. C. Tsay, , and N. F. Larsen, 1998: Cloud detection over the Arctic region using airborne imaging spectrometer data during the daytime. J. Appl. Meteor., 37 , 14211429.

    • Search Google Scholar
    • Export Citation
  • Hanssen, A. W., , and W. J. A. Kuipers, 1965: On the relationship between the frequency of rain and various meteorological parameters. Meded. Verh., 81 , 215.

    • Search Google Scholar
    • Export Citation
  • Holland, M. M., , and C. M. Bitz, 2003: Polar amplification of climate change in coupled models. Climate Dyn., 21 , 221232.

  • 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., 113D , D00A19. doi:10.1029/2008JD009837.

    • Search Google Scholar
    • Export Citation
  • Inoue, J., , J. Liu, , J. O. Pinto, , and J. A. Curry, 2006: Intercomparison of Arctic regional climate models: Modeling clouds and radiation for SHEBA in May 1998. J. Climate, 19 , 41674178.

    • Search Google Scholar
    • Export Citation
  • Intrieri, J. M., , C. W. Fairall, , M. D. Shupe, , P. O. G. Persson, , E. L. Andreas, , P. S. Guest, , and R. E. Moritz, 2002: An annual cycle of Arctic surface cloud forcing at SHEBA. J. Geophys. Res., 107 , 8039. doi:10.1029/2000JC000439.

    • Search Google Scholar
    • Export Citation
  • Jin, X., , J. M. Hanesiak, , and D. G. Barber, 2007: Time series of daily averaged cloud fractions over landfast first-year sea ice from multiple data sources. J. Appl. Meteor. Climatol., 46 , 18181827.

    • Search Google Scholar
    • Export Citation
  • Kato, S., , N. G. Loeb, , P. Minnis, , J. A. Francis, , T. P. Charlock, , D. A. Rutan, , E. E. Clothiaux, , and S. Sun-Mack, 2006: Seasonal and interannual variations of top-of-atmosphere irradiance and cloud cover over polar regions derived from the CERES data set. Geophys. Res. Lett., 33 , L19804. doi:10.1029/2006GL026685.

    • Search Google Scholar
    • Export Citation
  • Key, J., , and R. G. Barry, 1989: Cloud cover analysis with Arctic AVHRR data. 1. Cloud detection. J. Geophys. Res., 94D , 1852118535.

  • King, M. D., and Coauthors, 2003: Cloud and aerosol properties, precipitable water, and profiles of temperature and water vapor from MODIS. IEEE Trans. Geosci. Remote Sens., 41 , 442458.

    • Search Google Scholar
    • Export Citation
  • Lindsay, R. W., , J. Zhang, , A. Schweiger, , M. Steele, , and H. Stern, 2009: Arctic sea ice retreat in 2007 follows thinning trend. J. Climate, 22 , 165176.

    • Search Google Scholar
    • Export Citation
  • Liu, J. P., , J. A. Curry, , and Y. Y. Hu, 2004a: Recent Arctic sea ice variability: Connections to the Arctic Oscillation and the ENSO. Geophys. Res. Lett., 31 , L09211. doi:10.1029/2004GL019858.

    • Search Google Scholar
    • Export Citation
  • Liu, Y., , J. R. Key, , R. A. Frey, , S. A. Ackerman, , and W. P. Menzel, 2004b: Nighttime polar cloud detection with MODIS. Remote Sens. Environ., 92 , 181194.

    • Search Google Scholar
    • Export Citation
  • Liu, Y., , J. R. Key, , J. A. Francis, , and X. Wang, 2007: Possible causes of decreasing cloud cover in the Arctic winter, 1982-2000. Geophys. Res. Lett., 34 , L14705. doi:10.1029/2007GL030042.

    • Search Google Scholar
    • Export Citation
  • Liu, Y., , J. R. Key, , and X. Wang, 2008: The influence of changes in cloud cover on recent surface temperature trends in the Arctic. J. Climate, 21 , 705715.

    • Search Google Scholar
    • Export Citation
  • Lubin, D., , and E. Morrow, 1998: Evaluation of an AVHRR cloud detection and classification method over the central Arctic Ocean. J. Appl. Meteor., 37 , 166183.

    • Search Google Scholar
    • Export Citation
  • Mace, G. G., cited. 2007: Level-2 radar-lidar GEOPROF product version 1.0, process description and interface control document. Jet Propulsion Laboratory. [Available online at http://www.CloudSat.cira.colostate.edu/dataICDlist.php?go=list&path=/2B-GEOPROF-LIDAR].

    • Search Google Scholar
    • Export Citation
  • 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
  • Paluch, I. R., , D. H. Lenschow, , and Q. Wang, 1997: Arctic boundary layer in the fall season over open and frozen sea. J. Geophys. Res., 102D , 2595525971.

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

    • Search Google Scholar
    • Export Citation
  • Quinn, P. K., , G. Shaw, , E. Andrews, , E. G. Dutton, , T. Ruoho-Airola, , and S. L. Gong, 2007: Arctic haze: Current trends and knowledge gaps. Tellus, 59B , 99114.

    • Search Google Scholar
    • Export Citation
  • Rossow, W. B., , and R. A. Schiffer, 1999: Advances in understanding clouds from ISCCP. Bull. Amer. Meteor. Soc., 80 , 22612287.

  • Schweiger, A. J., 2004: Changes in seasonal cloud cover over the Arctic seas from satellite and surface observations. Geophys. Res. Lett., 31 , L12207. doi:10.1029/2004GL020067.

    • Search Google Scholar
    • Export Citation
  • Schweiger, A. J., , and J. R. Key, 1994: Arctic Ocean radiative fluxes and cloud forcing estimated from the ISCCP C2 cloud dataset, 1983–1990. J. Appl. Meteor., 33 , 948963.

    • Search Google Scholar
    • Export Citation
  • Serreze, M. C., , and J. A. Francis, 2006: The Arctic amplification debate. Climatic Change, 76 , 241264.

  • Serreze, M. C., , M. M. Holland, , and J. Stroeve, 2007: Perspectives on the Arctic’s shrinking sea-ice cover. Science, 315 , 15331536.

  • Solomon, S., , D. Qin, , M. Manning, , Z. Chen, , M. Marquis, , K. B. Averyt, , M. Tignor, , and H. L. Miller, 2007: Summary for policymakers. Climate Change 2007: The Physical Science Basis, S. Solomon et al., Eds., Cambridge University Press, 1–18.

    • Search Google Scholar
    • Export Citation
  • Stephens, G. L., and Coauthors, 2002: The CloudSat mission and the a-train—A new dimension of space-based observations of clouds and precipitation. Bull. Amer. Meteor. Soc., 83 , 17711790.

    • Search Google Scholar
    • Export Citation
  • Tjernstrom, M., , J. Sedlar, , and M. D. Shupe, 2008: How well do regional climate models reproduce radiation and clouds in the Arctic? An evaluation of ARCMIP simulations. J. Appl. Meteor. Climatol., 47 , 24052422.

    • Search Google Scholar
    • Export Citation
  • Vavrus, S., 2004: The impact of cloud feedbacks on Arctic climate under greenhouse forcing. J. Climate, 17 , 603615.

  • Vavrus, S., , and D. Waliser, 2008: An improved parameterization for simulating Arctic cloud amount in the CCSM3 climate model. J. Climate, 21 , 56735687.

    • Search Google Scholar
    • Export Citation
  • Walsh, J. E., , V. M. Kattsov, , W. L. Chapman, , V. Govorkova, , and T. Pavlova, 2002: Comparison of Arctic climate simulations by uncoupled and coupled global models. J. Climate, 15 , 14291446.

    • Search Google Scholar
    • Export Citation
  • Walsh, J. E., , S. J. Vavrus, , and W. L. Chapman, 2005: Workshop on modeling of the Arctic atmosphere. Bull. Amer. Meteor. Soc., 86 , 845852.

    • Search Google Scholar
    • Export Citation
  • Wang, X. J., , and J. R. Key, 2003: Recent trends in Arctic surface, cloud, and radiation properties from space. Science, 299 , 17251728.

    • Search Google Scholar
    • Export Citation
  • Wang, X. J., , and J. R. Key, 2005a: Arctic surface, cloud, and radiation properties based on the AVHRR Polar Pathfinder dataset. Part I: Spatial and temporal characteristics. J. Climate, 18 , 25582574.

    • Search Google Scholar
    • Export Citation
  • Wang, X. J., , and J. R. Key, 2005b: Arctic surface, cloud, and radiation properties based on the AVHRR Polar Pathfinder dataset. Part II: Recent trends. J. Climate, 18 , 25752593.

    • Search Google Scholar
    • Export Citation
  • Winker, D. M., , J. Pelon, , and M. P. McCormick, 2003: The CALIPSO mission: Spaceborne lidar for observation of aerosols and clouds. Lidar Remote Sensing for Industry and Environment Monitoring III, U. N. Singh, T. Itabe, and Z. Liu, Eds., International Society for Optical Engineering (SPIE Proceedings, Vol. 4893), 1–11.

    • Search Google Scholar
    • Export Citation
  • Zhang, X., , and J. E. Walsh, 2006: Toward a seasonally ice-covered Arctic Ocean: Scenarios from the IPCC AR4 model simulations. J. Climate, 19 , 17301747.

    • Search Google Scholar
    • Export Citation
  • View in gallery

    (left) The hit rate (solid lines) and (right) skill score (dotted lines) of the MODIS cloud detection algorithm in comparison to CALIOP cloud detection for a year period for open water (red) and sea ice (blue) surfaces in polar night conditions.

  • View in gallery

    Frequency distributions of (a) MODIS daily mean cloud amount (%), (b) GEOPROF-lidar daily mean cloud amount (%), and (c) daily mean cloud amount difference from MODIS and GEOPROF-lidar (%) vs AMSR-E SIC (%). Median values are overlaid as thick black line.

  • View in gallery

    As in Fig. 2, but for daily mean daytime MODIS cloud amount.

  • View in gallery

    As in Fig. 2, but for daily mean nighttime MODIS cloud amount.

  • View in gallery

    (a) Mean cloud amount September to October 2007, (b) difference between September to October 2007 and 2006 mean cloud amount from MODIS, (c) CloudSat–CALIPSO, and (d) AMSR-E SIC anomalies for 2007 vs the 2002–07 time period. White areas for CloudSat–CALIPSO signify an insufficient number of observations, with less than 1800 observations.

  • View in gallery

    Sea ice concentration decadal trends in winter, spring, summer, and autumn from 1982 to 2004.

  • View in gallery

    Anomalous cloud amount decadal trends caused by sea ice concentration changes and leading to changes in satellite detection capabilities. The decadal trends are for winter, spring, summer, and autumn from 1982 to 2004.

  • View in gallery

    Anomalous cloud radiative forcing decadal trend caused by anomalous cloud amount trend (Fig. 7) associated with trends in SIC, that lead to low bias in cloud amount over ice surface. The decadal trends are for winter, spring, summer, and autumn from 1982 to 2004.

  • View in gallery

    Fig. A1. Frequency distribution of cloud amount difference from (left) MODIS CF (%), (right) MODIS CFC (%), and GEOPROF-lidar (%) with regard to AMSR-E SIC (%). Medians are overlaid as thick black line.

  • View in gallery

    Fig. A2. Daily mean cloud amount September 2007 anomaly using (top) (left) MODIS CF and (right) MODIS CFC; (bottom) (left) the difference of the September 2007 anomalies of CF and CFC (CFC minus CF) and (bottom) the difference between CF and CFC (CFC minus CF).

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 62 62 6
PDF Downloads 37 37 3

Errors in Cloud Detection over the Arctic Using a Satellite Imager and Implications for Observing Feedback Mechanisms

View More View Less
  • 1 Cooperative Institute for Meteorological Satellite Studies, University of Wisconsin—Madison, Madison, Wisconsin
  • | 2 Cooperative Institute for Meteorological Satellite Studies, and Department of Atmospheric and Oceanic Sciences, University of Wisconsin—Madison, Madison, Wisconsin
  • | 3 Center for Satellite Applications and Research, NOAA/NESDIS, Madison, Wisconsin
  • | 4 Cooperative Institute for Meteorological Satellite Studies, University of Wisconsin—Madison, Madison, Wisconsin
© Get Permissions
Full access

Abstract

Arctic sea ice extent has decreased dramatically over the last 30 years, and this trend is expected to continue through the twenty-first century. Changes in sea ice extent impact cloud cover, which in turn influences the surface energy budget. Understanding cloud feedback mechanisms requires an accurate determination of cloud cover over the polar regions, which must be obtained from satellite-based measurements. The accuracy of cloud detection using observations from space varies with surface type, complicating any assessment of climate trends as well as the understanding of ice–albedo and cloud–radiative feedback mechanisms. To explore the implications of this dependence on measurement capability, cloud amounts from the Moderate Resolution Imaging Spectroradiometer (MODIS) are compared with those from the CloudSat and Cloud–Aerosol Lidar and Infrared Pathfinder (CALIPSO) satellites in both daytime and nighttime during the time period from July 2006 to December 2008. MODIS is an imager that makes observations in the solar and infrared spectrum. The active sensors of CloudSat and CALIPSO, a radar and lidar, respectively, provide vertical cloud structures along a narrow curtain.

Results clearly indicate that MODIS cloud mask products perform better over open water than over ice. Regional changes in cloud amount from CloudSat/CALIPSO and MODIS are categorized as a function of independent measurements of sea ice concentration (SIC) from the Advanced Microwave Scanning Radiometer for Earth Observing System (AMSR-E). As SIC increases from 10% to 90%, the mean cloud amounts from MODIS and CloudSat–CALIPSO both decrease; water that is more open is associated with increased cloud amount. However, this dependency on SIC is much stronger for MODIS than for CloudSat–CALIPSO, and is likely due to a low bias in MODIS cloud amount. The implications of this on the surface radiative energy budget using historical satellite measurements are discussed. The quantified ice–water difference in MODIS cloud detection can be used to adjust estimated trends in cloud amount in the presence of changing sea ice cover from an independent dataset. It was found that cloud amount trends in the Arctic might be in error by up to 2.7% per decade. The impact of these errors on the surface net cloud radiative effect (“forcing”) of the Arctic can be significant, as high as 8.5%.

Corresponding author address: Yinghui Liu, Cooperative Institute for Meteorological Satellite Studies, University of Wisconsin—Madison, 1225 West Dayton Street, Madison, WI 53706. Email: yinghuil@ssec.wisc.edu

Abstract

Arctic sea ice extent has decreased dramatically over the last 30 years, and this trend is expected to continue through the twenty-first century. Changes in sea ice extent impact cloud cover, which in turn influences the surface energy budget. Understanding cloud feedback mechanisms requires an accurate determination of cloud cover over the polar regions, which must be obtained from satellite-based measurements. The accuracy of cloud detection using observations from space varies with surface type, complicating any assessment of climate trends as well as the understanding of ice–albedo and cloud–radiative feedback mechanisms. To explore the implications of this dependence on measurement capability, cloud amounts from the Moderate Resolution Imaging Spectroradiometer (MODIS) are compared with those from the CloudSat and Cloud–Aerosol Lidar and Infrared Pathfinder (CALIPSO) satellites in both daytime and nighttime during the time period from July 2006 to December 2008. MODIS is an imager that makes observations in the solar and infrared spectrum. The active sensors of CloudSat and CALIPSO, a radar and lidar, respectively, provide vertical cloud structures along a narrow curtain.

Results clearly indicate that MODIS cloud mask products perform better over open water than over ice. Regional changes in cloud amount from CloudSat/CALIPSO and MODIS are categorized as a function of independent measurements of sea ice concentration (SIC) from the Advanced Microwave Scanning Radiometer for Earth Observing System (AMSR-E). As SIC increases from 10% to 90%, the mean cloud amounts from MODIS and CloudSat–CALIPSO both decrease; water that is more open is associated with increased cloud amount. However, this dependency on SIC is much stronger for MODIS than for CloudSat–CALIPSO, and is likely due to a low bias in MODIS cloud amount. The implications of this on the surface radiative energy budget using historical satellite measurements are discussed. The quantified ice–water difference in MODIS cloud detection can be used to adjust estimated trends in cloud amount in the presence of changing sea ice cover from an independent dataset. It was found that cloud amount trends in the Arctic might be in error by up to 2.7% per decade. The impact of these errors on the surface net cloud radiative effect (“forcing”) of the Arctic can be significant, as high as 8.5%.

Corresponding author address: Yinghui Liu, Cooperative Institute for Meteorological Satellite Studies, University of Wisconsin—Madison, 1225 West Dayton Street, Madison, WI 53706. Email: yinghuil@ssec.wisc.edu

1. Introduction

The Arctic is warming at a rate that is approximately twice that of the global average (Solomon et al. 2007; Serreze and Francis 2006). There has been a dramatic reduction in the observed sea ice extent in the Arctic Ocean (Serreze et al. 2007). Model projections suggest a continuation of the warming trend and a decrease in sea ice extent through this century (Holland and Bitz 2003; Zhang and Walsh 2006). These observed and projected changes result from various feedback processes, including the ice–albedo feedback and cloud feedbacks (Serreze and Francis 2006; Curry et al. 1996).

The properties of Arctic clouds are extremely important to the region’s climate system and for predicting Arctic climate change. Clouds have a strong radiative influence on the energy budget at the surface, which controls sea ice growth and melt (Intrieri et al. 2002; Liu et al. 2008; Tjernstrom et al. 2008; Curry et al. 1996). The sensitivity of regional climate to cloud processes is a major uncertainty in projecting changes in the Arctic (Solomon et al. 2007). Various satellite-derived products have documented seasonal changes in Arctic cloud amount in recent decades that are associated with measurable impacts on surface energy fluxes (Wang and Key 2003, 2005b; Schweiger 2004; Liu et al. 2007).

Polar cloud amount is not well represented in either global or regional climate models (Walsh et al. 2002, 2005; Vavrus 2004; Inoue et al. 2006; Birch et al. 2009). These misrepresentations are due, in part, to a lack of observations and the complexity of cloud formation and dissipation mechanisms (Vavrus and Waliser 2008; Beesley and Moritz 1999). Errors in observing Arctic cloud amount are among the factors contributing to the large uncertainties in projecting future Arctic climate change.

It is inherently difficult to detect clouds in the Arctic using passive visible, reflected infrared, and thermal (emitted) infrared portions of the electromagnetic spectrum. Cloud detection is challenging over surface ice and snow because of poor thermal and visible contrast between clouds and the underlying surface, small radiances from cold polar atmosphere, and temperature inversions in the lower troposphere (Lubin and Morrow 1998). Much effort has been put into the development and improvement of cloud detection in the Arctic using passive infrared and visible satellite remote sensing (Key and Barry 1989; Lubin and Morrow 1998; Rossow and Schiffer 1999; Gao et al. 1998; Liu et al. 2004b; Ackerman et al. 2008; Frey et al. 2008). The Moderate Resolution Imaging Spectroradiometer (MODIS) onboard the National Aeronautic and Space Administration (NASA) Terra and Aqua satellites measures radiances at 36 wavelengths, including infrared and solar bands, with spatial resolutions of 250 m to 1 km. Such a robust set of measurements provides the potential for improving cloud detection in the Arctic (Ackerman et al. 1998; King et al. 2003; Liu et al. 2004b; Frey et al. 2008; Ackerman et al. 2008). While improvements to MODIS cloud detection have been made (Liu et al. 2004b; Frey et al. 2008), there are still larger errors in nighttime Arctic cloud detection than for most other regions on earth (e.g., Holz et al. 2008).

The recent availability of observations from CloudSat and Cloud–Aerosol lidar and Infrared Pathfinder Satellite Observation (CALIPSO) provide an unprecedented opportunity to give a complete picture of cloud cover in the Arctic (Stephens et al. 2002; Winker et al. 2003). The cloud profiling radar (CPR) onboard CloudSat can penetrate deep within almost all nonprecipitating clouds, although it has limited sensitivity to thin cirrus, especially with small particle sizes. The CALIPSO cloud–aerosol lidar with orthogonal polarization (CALIOP) is sensitive to optically thin cloud layers. Combining information from these two instruments allows for observations over a wide spectrum of Arctic clouds.

This study focuses on evaluating MODIS cloud amount over sea ice and open water using cloud observations from CloudSat–CALIPSO in both daytime and nighttime. Focusing on cloud amount evaluation is a first step toward understanding the quality of current passive satellite cloud property retrievals and its influence on assessing changes in Arctic cloud properties. The next section extends the work of Holz et al. (2008) by assessing the capabilities of the MODIS cloud mask algorithm using collocated CALIOP observations. Section 2 compares MODIS regional cloud amount with the combined CloudSat–CALIPSO cloud product. Differences in cloud amount between the active and passive sensors are explored as a function of surface type. The impact of these differences on radiative fluxes is discussed in section 3. The appendix compares two daytime mean cloud amounts in the level-3 MODIS atmosphere daily global product, and evaluates both daily means with cloud amount from CloudSat–CALIPSO.

2. MODIS and CloudSat–CALIPSO cloud amount over sea ice

Detection of clouds over snow–ice surfaces from passive infrared and solar satellite observations is more difficult than over water surfaces because of poor thermal and visible contrast between clouds and snow–ice, and temperature inversions in the lower troposphere. MODIS cloud detection capabilities are better over water than over snow–ice (Holz et al. 2008), because of this higher contrast. Linear trends in Arctic sea ice extent from 1979 to 2006 are negative in every month of the year, indicating a larger proportion of open (unfrozen) water (Serreze et al. 2007). Recent sea ice concentration (SIC) changes in the Arctic, negative over most of the Arctic Ocean, are related to the changes in atmospheric circulation (Deser and Teng 2008; Liu et al. 2004a). These changes can potentially influence the overall quality of MODIS cloud amount in the Arctic and thus impact inferences about the relationship between changing surface conditions and changing cloud amount.

Cloud detection with MODIS can be compared at a pixel-to-pixel level with radar/lidar cloud amount (Liu et al. 2004b; Jin et al. 2007; Holz et al. 2008). This direct comparison provides valuable information of the performance of each individual MODIS cloud detection test, and for developing new detection methods (Liu et al. 2004b). This paper further explores the errors in cloud detection over the Arctic through a direct comparison of MODIS and CALIOP cloud products. While this pixel-to-pixel comparison identifies areas of disagreement, it is not sufficient for studying the impact of changing SIC on MODIS cloud detection capabilities. This is accomplished by gridding the passive and active remote sensing datasets, as discussed in the next section.

a. Pixel-to-pixel analysis

This section compares results from MODIS cloud detection algorithm with the collocated CALIOP cloud detection method. The collocation procedure is described in Holz et al. (2008). Figure 1 plots the overall agreement between collection 5 cloud mask from MODIS onboard Aqua (MYD35) and CALIOP during the polar night cases for July 2007 through June 2008. There are a total of 13–16 million collocated MODIS 1-km pixels per month. The overall agreement is expressed as the hit rate:
i1520-0442-23-7-1894-eq1
where Ncld is the number of cloud pixels in agreement, Nclr is the number of clear pixels in agreement, and N is the total number of collocated pixels. The agreement is also expressed by the Hanssen–Kuiper Skill Score (KSS; Hanssen and Kuipers 1965). The score has a range of −1 to +1, with 0 representing no skill. The KSS expresses the hit rate relative to the false alarm rate, and will remain positive as long as the hit rate is greater than the false alarm rate. The KSS is a useful metric when analyzing phenomena that are not normally distributed, such as cloud cover.

The Arctic nighttime cloud detection agreement between CALIOP and MODIS (Fig. 1) is characterized as a function of surface type (ice surface or water surface) as determined by a combination of near-real-time ice and snow extent (NISE) and National Centers for Environmental Prediction (NCEP) sea ice concentration data. For the open water scenes, the hit rate is generally greater than 85% and the KSS is generally 0.6 or better. During this yearlong period, the hit rate was best in the month of September 2007, with a broad minimum in the late winter and spring. The hit rate for ice surfaces is significantly less, with nearly all values less that 70%. This is consistent with the two-month comparisons of Holz et al. (2008). Surprisingly, the KSS hit rates are lowest during the late summer–early fall time period over the ice surfaces. The generally larger difference between hit rates and KSSs for ice surfaces compared to water surfaces indicates a higher false cloud detection rate over ice, particularly in the summer months when the region is typically very cloudy. This comparison also indicates that cloud detection over sea ice in winter is more difficult than over open water during the same time period. The next section explores how this difference might translate into misinterpretation of cloud amount changes.

b. Level-3 gridded analysis

The geographic area of interest of this study is the Arctic Ocean between 0°–240° longitude, and 70°–85° latitude. This region is largely land free and has a relatively higher temporal sampling by polar orbiting satellites. For the level-3 gridded analysis, the area is divided into 120 10° longitude by 3° latitude cells. In each grid cell, the cloud amount from Aqua MODIS and CloudSat–CALIPSO, and SIC from the Aqua Advanced Microwave Scanning Radiometer–Earth Observing System (AMSR-E) are averaged daily from July 2006 to December 2008.

The level-3 MODIS atmosphere daily global product contains the daily mean, daytime daily mean, and nighttime daily mean cloud amount for each 1° × 1° cell on an equal-angle grid. These daily mean values originate from the MODIS cloud mask level-2 product (collection 5), called MYD35_L2. The MODIS daily mean, daytime daily mean, and nighttime daily mean cloud amounts in each 10° longitude by 3° latitude cell are calculated as the means of all 1° × 1° daily mean, daytime daily mean, and nighttime daily mean cloud amounts, respectively, inside that cell. The combined CPR and CALIOP level-2 product 2B-GEOPROF-lidar (Mace 2007; Mace et al. 2009) provides cloud information in a 2.5 km (along-track) by 1.4 km (cross-track) footprint at nadir. CloudSat–CALIPSO daily mean cloud amount in each 10° longitude by 3° latitude cell is calculated as the percentage of footprints with cloud detected to all the footprints falling inside this cell boundary. The AMSR-E product provides daily SIC at a spatial resolution of 25 km by 25 km in polar stereographic projection (Cavalieri et al. 2004). The AMSR-E daily mean SIC in each 10° longitude by 3° latitude cell is calculated as the mean of all 25 km by 25 km daily mean SICs with the center of the 25 by 25 km box inside that cell. It should be noted that MODIS and CloudSat–CALIPSO have different viewing angles and different spatial resolution, which might cause disagreement in the cloud amount comparisons. The cloud detection difference that results from nadir versus scanning spatial sampling was discussed by Ackerman et al. (2008).

Frequency distributions of MODIS daily mean cloud amount, CloudSat–CALIPSO daily mean cloud mount, and the difference of these two daily means, with regard to daily mean SIC are shown in Fig. 2. Each pair of MODIS daily mean cloud amount and SIC, CloudSat–CALIPSO daily mean cloud amount and SIC, and the difference of the two daily mean cloud amounts and SIC in a cell of a day during the period 1 July 2006 to 31 December 2008 is counted as one sample in calculating these frequency distributions. Samples are most common in the Arctic with SIC less than 10% and greater than 90%. To better illustrate the frequency distribution from low to high SIC, those samples with SIC less than 10% and greater than 90% are excluded in the following discussion; however, the conclusions are the same with those cases included.

Daily mean MODIS cloud amounts tend to be smaller with higher sea ice concentration (Fig. 2a). Over surfaces with an SIC less than 20%, the MODIS daily mean cloud amounts are mostly larger than 90%. Over surfaces with SIC larger than 80%, the MODIS daily mean cloud amounts are mostly larger than 60%, with cloud amounts between 60% and 90% more frequent than those over surfaces with SIC less than 20%. The MODIS cloud amount median values change from 92.8% for SIC between 10% and 20% to 80% for SIC between 80% and 90%. The decrease of the daily mean MODIS cloud amount with an increase in SIC shown in Fig. 2a could be attributed to: 1) the annual cycle of cloud amount, 2) the effect of more open water causing more cloud amount, 3) better MODIS cloud detection capability over open water relative to over sea ice, and 4) better MODIS cloud detection capability during daytime than nighttime because of more frequent open water conditions and more information from solar bands during daytime. Mean Arctic cloud amount has a pronounced annual cycle, with smaller cloud amounts in winter and early spring, when SIC peaks, and a maximum cloud amount in summer and early fall, when SIC is lower (Wang and Key 2005a; Intrieri et al. 2002). The annual mean cloud amount is greater than 80%. Based on the aircraft data collected on research flights in the Beaufort Arctic Storms Experiment from September to October 1994, Paluch et al. (1997) demonstrated that over ice, vertical moisture fluxes tend to be weak. The boundary layer is stably stratified and cloud formation is suppressed. Conversely, over open water, vertical moisture fluxes are stronger, the boundary layer is less stable, and more clouds form, which is consistent with the findings in Evan et al. (2008). Kato et al. (2006) found that the peak cloud fraction in the Arctic is coincident with minimum sea ice coverage. From the MODIS mean cloud distribution in the Arctic from 2002 to 2006, a higher cloud fraction is almost always observed over open water than over sea ice. These findings suggest that newly open water in the Arctic might be related to the larger cloud amount associated with SIC changes. MODIS cloud detection capabilities are better over water than over snow–ice, because of larger thermal and spectral reflectance contrasts between clouds and open water. MODIS cloud detection capabilities are also expected to be better during the day, when visible and near-infrared spectral information is available (Liu et al. 2004b).

On the other hand, CloudSat–CALIPSO daily mean cloud amounts are generally over 90% for surfaces with sea concentrations between 10% and 90%, with median cloud amounts change from 96.5% to 92.5%. There is a slight decrease in cloud amount with increasing SIC (Fig. 2b). This decrease in cloud amount with increasing SIC for CloudSat–CALIPSO is weaker than that for MODIS. If we assume that CloudSat–CALIPSO cloud detection method is equally accurate over open water and ice, then this decrease can be attributed to real differences in cloud amount over the two surfaces.

To reduce the impact of the annual cycles of Arctic cloud amount and open water on our interpretation, we difference MODIS and CloudSat–CALIPSO amounts. Differences between the daily mean cloud amount from MODIS and CloudSat–CALIPSO are negative for surfaces with SIC between 10% and 90%; that is, MODIS detects less cloud over ice (Fig. 2c). This negative difference can be attributed to overall better cloud detection capabilities of the radar–lidar combination, especially for very thin cloud, or differences in sampling between nadir observations and those from a scanner (Ackerman et al. 2008). The differences are larger for higher SIC. There are two possible explanations for this dependence on SIC. First, MODIS has better cloud detection capabilities during daytime than nighttime. The daylight months have lower SICs, while nighttime (winter) is accompanied by higher SICs. Second, MODIS has better cloud detection capabilities over surfaces with lower SIC than over surfaces with higher SIC.

The frequency distribution of MODIS daily mean cloud amount, CloudSat/CALIPSO daily mean cloud amount, and the difference between them as a function of daily mean SIC is shown in Fig. 3 for daytime conditions and in Fig. 4 for nighttime. Both MODIS daily mean cloud amounts decrease with increasing SIC, with larger negative values for the higher sea ice concentrations. The relationship is more evident for nighttime conditions than in daytime. This is consistent with the pixel-to-pixel comparison of Fig. 1.

The linear relationships between the MODIS–CloudSat–CALIPSO cloud amount difference and SIC are calculated for the daily mean, daytime daily mean, and nighttime daily mean data. The values, in units of percent cloud amount per percent SIC, are −0.10, −0.08, and −0.21, respectively. For an increase in SIC from 10% to 90%, the daily mean cloud amount difference between MODIS and CloudSat–CALIPSO decreases 8.0% for the daily mean, 6.4% for the daytime daily mean, and 16.8% for the nighttime daily mean. We attribute these differences to poorer MODIS cloud detection capabilities over ice. These differences between MODIS and CloudSat–CALIPSO are nearly the same for the daily, daytime, and nighttime means when cases with SIC of less than 10% and larger than 90% are included in the computation.

Comparisons of the daily mean MODIS–CloudSat–CALIPSO cloud amount difference for varying SIC were also made over different subregions of the Arctic Ocean in spring (March, April, May), summer (June, July, August), autumn (September, October, November), and winter (December, January, February). The numerical results were similar, with larger decreasing difference magnitudes in spring, autumn, and winter than those in summer.

3. Implications

Satellite data revealed an unprecedented minimum Arctic sea ice extent from August to October 2007 (Comiso et al. 2008; Lindsay et al. 2009). Much of the sea ice decline occurred over the Chukchi Sea, the central Arctic, the Kara Sea, and part of the Beaufort Sea, with the maximum SIC anomalies (over −60%) over the central Arctic and the Chukchi Sea (Fig. 5d). Mean cloud amounts are greater over the Arctic Ocean in September and October as shown in Fig. 5a. The largely ice-free regions in September and October 2007 correspond to large positive cloud amount anomalies from MODIS data (Fig. 5b), in agreement with findings by Evan et al. (2008). There are also positive cloud amount anomalies in 2007 on the Kara Sea, and on the east side of Greenland. From CloudSat–CALIPSO, September and October mean cloud amounts are approximately 10% higher in 2007 than 2006 over the central Arctic and the Chukchi Sea (Fig. 5c). The difference between MODIS mean cloud amount in 2007 and 2006 over the same region is larger than 20%. This 10% difference between MODIS and CloudSat–CALIPSO can be attributed to the MODIS cloud detection capability differences over changing sea ice conditions, or the lower sampling of the CloudSat–CALIPSO (Ackerman et al. 2008). Given the relationship between the MODIS–CloudSat–CALIPSO cloud amount difference and SIC calculated in the previous section (−0.10%, −0.08%, and −0.21% percent cloud amount per percent SIC for the daily mean, the daytime daily mean, and the nighttime daily mean), and given SIC anomalies of more than −60% in 2007, the MODIS cloud amount difference between 2007 and 2006 can be between 4.8% to 12.6% because of differences in MODIS cloud detection capabilities over open water and ice.

Changes in SIC from 1982 to 2004 differ seasonally in terms of sign, magnitude, and spatial distribution (Fig. 6). The pattern of winter and spring SIC trends exhibits slightly positive values over the Canada Basin and most of the central Arctic Ocean, with magnitudes less than 1.5% per decade on average. Moderate negative values are near the ice edges in Nansen Basin, Barents and Kara seas, with magnitudes less than 2.0% per decade. Strong negative trends in the Baffin Bay are over −7.0% decade−1. Summer and autumn SIC trends are negative over the Chukchi and Beaufort seas, the Canada archipelago, Baffin Bay, and the ice edges on the Atlantic side of the Arctic. The largest SIC decrease is over the Chukchi and Beaufort seas in summer and autumn, with an average trend of −6.1% per decade in summer and −9.8% per decade in autumn, exceeding −15% per decade over a portion of the Chukchi Sea in autumn. The sign, magnitude, and spatial pattern of SIC trends shown here are very similar to SIC trends calculated using the same dataset from 1979 to 2006 (Deser and Teng 2008).

Based on the previous analysis, part of the cloud trends derived from satellite passive imager products may be caused by trends in SIC as a result of differences in satellite cloud detection capabilities for different surface types, as demonstrated above. This part of the cloud trends, calculated as the product of the SIC trends and the linear slopes of the relationship between the MODIS–CloudSat–CALIPSO cloud amount difference with SIC, are shown in Fig. 7 as the anomalous trends in cloud amount for the time period corresponding to Fig. 6. In winter and spring, anomalous trends in cloud amount are not significant over most regions, with the magnitude less than 0.1% per decade, since there are only slight SIC changes. Over Baffin Bay, these trends in cloud amount are as high as 0.9% per decade in winter, and 0.3% per decade in spring. In summer, these anomalous trends are mainly positive and evenly distributed over most of the Arctic Ocean, with mean trends of 0.4% per decade averaged over the Arctic Ocean. The most noticeable trends resulting from cloud detection are over the Chukchi and Beaufort seas in autumn, with 1.7% and 2.7% per decade respectively. The mean trend over the entire Arctic Ocean is 0.8% per decade in autumn.

Errant cloud amount trends associated with satellite cloud detection capabilities in the presence of sea ice concentration changes will affect the calculation of the trends in other surface properties. Surface radiative fluxes and cloud forcing (radiative effect) are good examples, as they are significant components of the surface energy budget. In the Arctic, especially over the Arctic Ocean, clouds warm the surface at all times of the year except during a portion of the summer (Schweiger and Key 1994; Intrieri et al. 2002). The positive cloud amount trends in summer and autumn would have a cooling effect on the surface in summer and a warming effect in autumn. Schweiger and Key (1994) found the net cloud forcing at the surface to be approximately 66.0, 26.7, −23.7, and 47.3 W m−2 for winter, spring, summer, and autumn, respectively. By applying these cloud forcing values to the anomalous cloud amount trends shown in Fig. 7, the cloud forcing distribution in four seasons is computed (Fig. 8). The annual cloud forcing trend from 1982 to 2004 averaged over the Arctic Ocean is 0.43 W m−2 decade−1; it is 0.27 W m−2 decade−1 from 1982 to 1999. These trends should be adjusted in the calculation of cloud forcing trend from satellite products because of changes in MODIS’ ability to detect clouds over ice and open water. It should be noted that the cloud forcing trends given here (0.43 and 0.27 W m−2 year−1) are approximate and for illustrative purposes only. More accurate values require detailed calculations with information of other cloud macrophysical (e.g., cloud-top altitude) and microphysical characteristics (e.g., cloud optical depth and effective radius), and other physical parameters, including surface albedo. However, these estimates suggest that clouds might have cooled the surface even more strongly than previously estimated. Wang and Key (2003) estimated the decadal net cloud forcing from 1982 to 1999 to be −3.17 W m−2. This study suggests that the cooling effect of Arctic clouds on the surface temperature would be stronger by about 8.5%, or −0.27 W m−2.

4. Discussion and conclusions

An accurate cloud detection capability in the Arctic is crucial to understanding Arctic cloud formation and dissipation mechanisms and complex cloud feedbacks, and to better predict Arctic climate change. Satellites provide complete coverage of the Arctic at a relatively high spatial resolution, but the unique Arctic environment makes cloud detection using passive solar reflected and thermal infrared observations from spacecraft difficult.

In this study, cloud observations from CloudSat and CALIPSO were used to evaluate MODIS cloud amounts with a particular focus on differences in cloud detection over sea ice and open (unfrozen) water. The daily means were based on the MODIS cloud mask level-2 product (collection 5), MYD35_L2. Over surfaces with sea ice concentration (SIC) from 10% to 90%, daily mean cloud amounts from MODIS and CloudSat–CALIPSO decrease with increasing SIC (more open water is associated with increased cloud amount). The decreasing relationship is much stronger for MODIS than for CloudSat–CALIPSO combined product. The daily mean cloud amount difference between MODIS and CloudSat–CALIPSO is negative for SIC between 10% and 90%, meaning that MODIS detects less cloud than CloudSat–CALIPSO over ice. This can be attributed to the overall better cloud detection capabilities of the radar–lidar observations. However, the large cloud amount from CloudSat–CALIPSO during both daytime and nighttime (Figs. 3b and 4b) could in part be a result of dense aerosols misidentified as clouds. In addition, the appendix discusses errors in using the wrong MODIS cloud amount product.

Pixel-to-pixel comparisons of collocated MODIS level-2 cloud mask and CALIOP cloud data from July 2007 to June 2008 show a higher hit rate and skill score over water surface than over ice surface, which indicates that MODIS cloud mask products perform better over water than over ice. The differences between MODIS and CloudSat–CALIPSO daily mean, daytime and nighttime cloud amounts all increase with increasing SIC, with MODIS detecting less cloud over ice. This appears to be due to poorer MODIS cloud detection over ice both in daytime and nighttime. The magnitudes of the differences between the two sensors (MODIS minus CloudSat–CALIPSO cloud amounts) are −0.10%, −0.08%, and −0.21% cloud amount per percent SIC for daily mean, daytime, and nighttime daily mean. The relationships are also found in all four seasons, with larger decreasing magnitudes in spring, autumn, and winter than those in summer. The daily mean MODIS cloud amount from the MODIS cloud product level-2 product, called MYD06_L2, is also evaluated. Details are available in the appendix.

The differences in MODIS cloud amounts over ice and open water, as determined assuming CloudSat–CALIPSO as “truth,” are important for Arctic climate studies. By considering the computed difference to adjust MODIS cloud amounts, the disparity between the cloud amount anomalies in 2007 from MODIS and CloudSat–CALIPSO over the newly open water in the central Arctic, and the Chukchi Sea can be explained. Additionally, the quantified ice–water difference in MODIS cloud detection can be used to adjust estimated trends in cloud amount in the presence of changing sea ice extent. It was found that cloud amount trends might be in error by up to 2.7% per decade. The impact of these errors on the net cloud radiative effect (forcing) at the surface can be significant, as high as 8.5%.

The 2B-GEOPROF-lidar product from CloudSat and CALIPSO merges cloud information from the CPR and CALIOP. Threshold algorithms are applied to radar reflectivity and to successive iterations of a multiscale spatial averaging routine to derive cloud information for CPR and CALIOP (Mace et al. 2009). Eloranta et al. (2008) showed that observed fractional cloud cover from ground-based lidar are strongly dependent on the thresholds used to define the presence of cloud. Such dependence will likely exist in the 2B-GEOPROF-lidar products, which might lead to uncertainty in the cloud amount because of the uncertainty in the thresholds chosen. Arctic haze, mainly a mixture of sulfate and particulate organic matter, has a distinct seasonal cycle, with a maximum in late winter and early spring because of intense meridional transport from the midlatitude and minimum removal process during that time period (Quinn et al. 2007). A thick Arctic haze layer can be identified as cloud by lidar due its high backscattering, which could also cause uncertainty in cloud amount from the 2B-GEOPROF-lidar products. These uncertainties merit further investigation.

Geophysical parameters derived from satellite retrievals may have condition-dependent biases. This study explores the dependence on cloud detection as a function of surface type, and suggests that research concerning the trend of these parameters need to be taken into account when defining trends and biases.

Acknowledgments

The authors thank the National Snow and Ice Data Center (NSIDC) for providing the AMSR-E sea ice concentration product, the CloudSat data processing center at Colorado State University for providing the CloudSat–CALIPSO product 2B-GEOPROF-LIDAR data, and Goddard Space Flight Center at NASA for providing the MODIS data. Zhenglong Li assisted with plotting the figures, Steve Dutcher and Robert Holz helped with the collocated MODIS and CALIPSO observations in the pixel-to-pixel analysis. This work was NOAA/NESDIS, NASA, and the NPOESS Integrated Program Office. The views, opinions, and findings contained in this report are those of the author(s) and should not be construed as an official National Oceanic and Atmospheric Administration or U.S. Government position, policy, or decision.

REFERENCES

  • Ackerman, S. A., , K. I. Strabala, , W. P. Menzel, , R. A. Frey, , C. C. Moeller, , and L. E. Gumley, 1998: Discriminating clear sky from clouds with MODIS. J. Geophys. Res., 103D , 3214132157.

    • Search Google Scholar
    • Export Citation
  • Ackerman, S. A., , R. E. Holz, , R. Frey, , E. W. Eloranta, , B. C. Maddux, , and M. McGill, 2008: Cloud detection with MODIS. Part II: Validation. J. Atmos. Oceanic Technol., 25 , 10731086.

    • Search Google Scholar
    • Export Citation
  • Beesley, J. A., , and R. E. Moritz, 1999: Toward an explanation of the annual cycle of cloudiness over the Arctic Ocean. J. Climate, 12 , 395415.

    • Search Google Scholar
    • Export Citation
  • Birch, C. E., , I. M. Brooks, , M. Tjernstrom, , S. F. Milton, , P. Earnshaw, , S. Soderberg, , and P. O. G. Persson, 2009: The performance of a global and mesoscale model over the central Arctic Ocean during late summer. J. Geophys. Res., 114 , D13104. doi:10.1029/2008JD010790.

    • Search Google Scholar
    • Export Citation
  • Cavalieri, D., , M. Thorsten, , and J. Comiso, cited. 2004: AMSR-E/Aqua daily L3 25-km brightness temperature and SIC polar grids V002, July 2006 to December 2008. National Snow and Ice Data Center. [Available online at http://nsidc.org/data/ae_si25.html].

    • Search Google Scholar
    • Export Citation
  • Comiso, J. C., , C. L. Parkinson, , R. Gersten, , and L. Stock, 2008: Accelerated decline in the Arctic sea ice cover. Geophys. Res. Lett., 35 , L01703. doi:10.1029/2007GL031972.

    • Search Google Scholar
    • Export Citation
  • Curry, J. A., , W. B. Rossow, , D. Randall, , and J. L. Schramm, 1996: Overview of Arctic cloud and radiation characteristics. J. Climate, 9 , 17311764.

    • Search Google Scholar
    • Export Citation
  • Deser, C., , and H. Teng, 2008: Evolution of Arctic sea ice concentration trends and the role of atmospheric circulation forcing, 1979-2007. Geophys. Res. Lett., 35 , L02504. doi:10.1029/2007GL032023.

    • Search Google Scholar
    • Export Citation
  • Eloranta, E. W., , J. P. Garcia, , I. A. Razenkov, , T. Uttal, , and M. Shupe, 2008: Cloud fraction statistics derived from 2-years of high spectral resolution lidar data acquired at Eureka, Canada. Proc. 24th Int. Laser Radar Conf., Boulder, CO, NCAR and NOAA, 555–558.

    • Search Google Scholar
    • Export Citation
  • Evan, A. T., , Y. Liu, , and B. Maddux, 2008: Global cloudiness. Bull. Amer. Meteor. Soc., 89 , (Suppl.). S23S26.

  • 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
  • Gao, B. C., , W. Han, , S. C. Tsay, , and N. F. Larsen, 1998: Cloud detection over the Arctic region using airborne imaging spectrometer data during the daytime. J. Appl. Meteor., 37 , 14211429.

    • Search Google Scholar
    • Export Citation
  • Hanssen, A. W., , and W. J. A. Kuipers, 1965: On the relationship between the frequency of rain and various meteorological parameters. Meded. Verh., 81 , 215.

    • Search Google Scholar
    • Export Citation
  • Holland, M. M., , and C. M. Bitz, 2003: Polar amplification of climate change in coupled models. Climate Dyn., 21 , 221232.

  • 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., 113D , D00A19. doi:10.1029/2008JD009837.

    • Search Google Scholar
    • Export Citation
  • Inoue, J., , J. Liu, , J. O. Pinto, , and J. A. Curry, 2006: Intercomparison of Arctic regional climate models: Modeling clouds and radiation for SHEBA in May 1998. J. Climate, 19 , 41674178.

    • Search Google Scholar
    • Export Citation
  • Intrieri, J. M., , C. W. Fairall, , M. D. Shupe, , P. O. G. Persson, , E. L. Andreas, , P. S. Guest, , and R. E. Moritz, 2002: An annual cycle of Arctic surface cloud forcing at SHEBA. J. Geophys. Res., 107 , 8039. doi:10.1029/2000JC000439.

    • Search Google Scholar
    • Export Citation
  • Jin, X., , J. M. Hanesiak, , and D. G. Barber, 2007: Time series of daily averaged cloud fractions over landfast first-year sea ice from multiple data sources. J. Appl. Meteor. Climatol., 46 , 18181827.

    • Search Google Scholar
    • Export Citation
  • Kato, S., , N. G. Loeb, , P. Minnis, , J. A. Francis, , T. P. Charlock, , D. A. Rutan, , E. E. Clothiaux, , and S. Sun-Mack, 2006: Seasonal and interannual variations of top-of-atmosphere irradiance and cloud cover over polar regions derived from the CERES data set. Geophys. Res. Lett., 33 , L19804. doi:10.1029/2006GL026685.

    • Search Google Scholar
    • Export Citation
  • Key, J., , and R. G. Barry, 1989: Cloud cover analysis with Arctic AVHRR data. 1. Cloud detection. J. Geophys. Res., 94D , 1852118535.

  • King, M. D., and Coauthors, 2003: Cloud and aerosol properties, precipitable water, and profiles of temperature and water vapor from MODIS. IEEE Trans. Geosci. Remote Sens., 41 , 442458.

    • Search Google Scholar
    • Export Citation
  • Lindsay, R. W., , J. Zhang, , A. Schweiger, , M. Steele, , and H. Stern, 2009: Arctic sea ice retreat in 2007 follows thinning trend. J. Climate, 22 , 165176.

    • Search Google Scholar
    • Export Citation
  • Liu, J. P., , J. A. Curry, , and Y. Y. Hu, 2004a: Recent Arctic sea ice variability: Connections to the Arctic Oscillation and the ENSO. Geophys. Res. Lett., 31 , L09211. doi:10.1029/2004GL019858.

    • Search Google Scholar
    • Export Citation
  • Liu, Y., , J. R. Key, , R. A. Frey, , S. A. Ackerman, , and W. P. Menzel, 2004b: Nighttime polar cloud detection with MODIS. Remote Sens. Environ., 92 , 181194.

    • Search Google Scholar
    • Export Citation
  • Liu, Y., , J. R. Key, , J. A. Francis, , and X. Wang, 2007: Possible causes of decreasing cloud cover in the Arctic winter, 1982-2000. Geophys. Res. Lett., 34 , L14705. doi:10.1029/2007GL030042.

    • Search Google Scholar
    • Export Citation
  • Liu, Y., , J. R. Key, , and X. Wang, 2008: The influence of changes in cloud cover on recent surface temperature trends in the Arctic. J. Climate, 21 , 705715.

    • Search Google Scholar
    • Export Citation
  • Lubin, D., , and E. Morrow, 1998: Evaluation of an AVHRR cloud detection and classification method over the central Arctic Ocean. J. Appl. Meteor., 37 , 166183.

    • Search Google Scholar
    • Export Citation
  • Mace, G. G., cited. 2007: Level-2 radar-lidar GEOPROF product version 1.0, process description and interface control document. Jet Propulsion Laboratory. [Available online at http://www.CloudSat.cira.colostate.edu/dataICDlist.php?go=list&path=/2B-GEOPROF-LIDAR].

    • Search Google Scholar
    • Export Citation
  • 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
  • Paluch, I. R., , D. H. Lenschow, , and Q. Wang, 1997: Arctic boundary layer in the fall season over open and frozen sea. J. Geophys. Res., 102D , 2595525971.

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

    • Search Google Scholar
    • Export Citation
  • Quinn, P. K., , G. Shaw, , E. Andrews, , E. G. Dutton, , T. Ruoho-Airola, , and S. L. Gong, 2007: Arctic haze: Current trends and knowledge gaps. Tellus, 59B , 99114.

    • Search Google Scholar
    • Export Citation
  • Rossow, W. B., , and R. A. Schiffer, 1999: Advances in understanding clouds from ISCCP. Bull. Amer. Meteor. Soc., 80 , 22612287.

  • Schweiger, A. J., 2004: Changes in seasonal cloud cover over the Arctic seas from satellite and surface observations. Geophys. Res. Lett., 31 , L12207. doi:10.1029/2004GL020067.

    • Search Google Scholar
    • Export Citation
  • Schweiger, A. J., , and J. R. Key, 1994: Arctic Ocean radiative fluxes and cloud forcing estimated from the ISCCP C2 cloud dataset, 1983–1990. J. Appl. Meteor., 33 , 948963.

    • Search Google Scholar
    • Export Citation
  • Serreze, M. C., , and J. A. Francis, 2006: The Arctic amplification debate. Climatic Change, 76 , 241264.

  • Serreze, M. C., , M. M. Holland, , and J. Stroeve, 2007: Perspectives on the Arctic’s shrinking sea-ice cover. Science, 315 , 15331536.

  • Solomon, S., , D. Qin, , M. Manning, , Z. Chen, , M. Marquis, , K. B. Averyt, , M. Tignor, , and H. L. Miller, 2007: Summary for policymakers. Climate Change 2007: The Physical Science Basis, S. Solomon et al., Eds., Cambridge University Press, 1–18.

    • Search Google Scholar
    • Export Citation
  • Stephens, G. L., and Coauthors, 2002: The CloudSat mission and the a-train—A new dimension of space-based observations of clouds and precipitation. Bull. Amer. Meteor. Soc., 83 , 17711790.

    • Search Google Scholar
    • Export Citation
  • Tjernstrom, M., , J. Sedlar, , and M. D. Shupe, 2008: How well do regional climate models reproduce radiation and clouds in the Arctic? An evaluation of ARCMIP simulations. J. Appl. Meteor. Climatol., 47 , 24052422.

    • Search Google Scholar
    • Export Citation
  • Vavrus, S., 2004: The impact of cloud feedbacks on Arctic climate under greenhouse forcing. J. Climate, 17 , 603615.

  • Vavrus, S., , and D. Waliser, 2008: An improved parameterization for simulating Arctic cloud amount in the CCSM3 climate model. J. Climate, 21 , 56735687.

    • Search Google Scholar
    • Export Citation
  • Walsh, J. E., , V. M. Kattsov, , W. L. Chapman, , V. Govorkova, , and T. Pavlova, 2002: Comparison of Arctic climate simulations by uncoupled and coupled global models. J. Climate, 15 , 14291446.

    • Search Google Scholar
    • Export Citation
  • Walsh, J. E., , S. J. Vavrus, , and W. L. Chapman, 2005: Workshop on modeling of the Arctic atmosphere. Bull. Amer. Meteor. Soc., 86 , 845852.

    • Search Google Scholar
    • Export Citation
  • Wang, X. J., , and J. R. Key, 2003: Recent trends in Arctic surface, cloud, and radiation properties from space. Science, 299 , 17251728.

    • Search Google Scholar
    • Export Citation
  • Wang, X. J., , and J. R. Key, 2005a: Arctic surface, cloud, and radiation properties based on the AVHRR Polar Pathfinder dataset. Part I: Spatial and temporal characteristics. J. Climate, 18 , 25582574.

    • Search Google Scholar
    • Export Citation
  • Wang, X. J., , and J. R. Key, 2005b: Arctic surface, cloud, and radiation properties based on the AVHRR Polar Pathfinder dataset. Part II: Recent trends. J. Climate, 18 , 25752593.

    • Search Google Scholar
    • Export Citation
  • Winker, D. M., , J. Pelon, , and M. P. McCormick, 2003: The CALIPSO mission: Spaceborne lidar for observation of aerosols and clouds. Lidar Remote Sensing for Industry and Environment Monitoring III, U. N. Singh, T. Itabe, and Z. Liu, Eds., International Society for Optical Engineering (SPIE Proceedings, Vol. 4893), 1–11.

    • Search Google Scholar
    • Export Citation
  • Zhang, X., , and J. E. Walsh, 2006: Toward a seasonally ice-covered Arctic Ocean: Scenarios from the IPCC AR4 model simulations. J. Climate, 19 , 17301747.

    • Search Google Scholar
    • Export Citation

APPENDIX

MOD35 and MOD06 Cloud Amount Comparison

There are four daily mean cloud amounts in level-3 MODIS atmosphere daily global product: Cloud_Fraction_Mean, Cloud_Fraction_Day_Mean, Cloud_Fraction_Night_Mean, and Cloud_Fraction_Combined_FMean. As the names suggest, the first three represent the daily mean cloud amounts during an entire day, during daytime only, and during nighttime only, respectively. These daily means originate from the MODIS cloud mask level-2 product, MYD35_L2. The fourth mean represents the daily mean fraction of pixels with high-quality assurance indices for the cloud microphysical characteristics retrievals during the daytime, which excludes the pixels that do not return a successful retrieval of cloud optical properties. This daily mean originates from the MODIS cloud product level-2 product, MYD06_L2.

This appendix compares the Cloud_Fraction_Day_Mean from MYD35_L2 (hereafter, CF) and Cloud_Fraction_Combined_FMean (hereafter, CFC) from MYD06_L2 in order to determine how these two daytime daily means differ when comparing with CloudSat–CALIPSO daily means as a function of SIC (Figure A1). The CFC represents only those pixels with a high confidence in retrieved cloud microphysical properties. The CFC is always a lower value than the MYD35 cloud mask product because it is a subset of the MYD35 dataset.

Both the CF and CFC have smaller values than those from CloudSat–CALIPSO for SIC between 10% and 90%. The magnitude of the negative differences between CF and CloudSat–CALIPSO daily mean is much smaller than that from CFC, particularly with SIC over 60%. Both negative differences have decreasing trends with increasing SICs, with −0.08 for CF and −0.18 for CFC. The differences in the decreasing trends suggest that cloud amount from CF is less affected by the surface SIC changes than cloud amount from CFC. For climate studies related to cloud amount changes in the Arctic, care is needed when using CF and CFC, especially in applying CFC to cloud amount studies.

The CFC, being a cloud microphysical retrieval fraction, would be less likely to be retrieved over ice–snow than over open water (Platnick et al. 2003). Therefore, if over a given region the ice field retreated from one year to the next and was replaced by open water, we would expect to see an increase in the CFC relative to previous years (Fig. A2, top right). Thus, in September 2007 the CFC and CF exhibited similar anomalies with respect to their own time series but the magnitude was significantly less in the CF dataset (Fig. A2, bottom left). Also, as would be expected, the difference between the CFC and CF over water was a minimum (Fig. A2, bottom right). As can be seen in Fig. A2, it would be relatively easy to erroneously argue that the cloud amount anomaly over regions of newly open water is greater than 20% in September 2007 in the CFC dataset; however, this cloud amount anomaly is the true cloud amount anomaly plus the anomaly due to increased cloud microphysical retrievals.

Fig. 1.
Fig. 1.

(left) The hit rate (solid lines) and (right) skill score (dotted lines) of the MODIS cloud detection algorithm in comparison to CALIOP cloud detection for a year period for open water (red) and sea ice (blue) surfaces in polar night conditions.

Citation: Journal of Climate 23, 7; 10.1175/2009JCLI3386.1

Fig. 2.
Fig. 2.

Frequency distributions of (a) MODIS daily mean cloud amount (%), (b) GEOPROF-lidar daily mean cloud amount (%), and (c) daily mean cloud amount difference from MODIS and GEOPROF-lidar (%) vs AMSR-E SIC (%). Median values are overlaid as thick black line.

Citation: Journal of Climate 23, 7; 10.1175/2009JCLI3386.1

Fig. 3.
Fig. 3.

As in Fig. 2, but for daily mean daytime MODIS cloud amount.

Citation: Journal of Climate 23, 7; 10.1175/2009JCLI3386.1

Fig. 4.
Fig. 4.

As in Fig. 2, but for daily mean nighttime MODIS cloud amount.

Citation: Journal of Climate 23, 7; 10.1175/2009JCLI3386.1

Fig. 5.
Fig. 5.

(a) Mean cloud amount September to October 2007, (b) difference between September to October 2007 and 2006 mean cloud amount from MODIS, (c) CloudSat–CALIPSO, and (d) AMSR-E SIC anomalies for 2007 vs the 2002–07 time period. White areas for CloudSat–CALIPSO signify an insufficient number of observations, with less than 1800 observations.

Citation: Journal of Climate 23, 7; 10.1175/2009JCLI3386.1

Fig. 6.
Fig. 6.

Sea ice concentration decadal trends in winter, spring, summer, and autumn from 1982 to 2004.

Citation: Journal of Climate 23, 7; 10.1175/2009JCLI3386.1

Fig. 7.
Fig. 7.

Anomalous cloud amount decadal trends caused by sea ice concentration changes and leading to changes in satellite detection capabilities. The decadal trends are for winter, spring, summer, and autumn from 1982 to 2004.

Citation: Journal of Climate 23, 7; 10.1175/2009JCLI3386.1

Fig. 8.
Fig. 8.

Anomalous cloud radiative forcing decadal trend caused by anomalous cloud amount trend (Fig. 7) associated with trends in SIC, that lead to low bias in cloud amount over ice surface. The decadal trends are for winter, spring, summer, and autumn from 1982 to 2004.

Citation: Journal of Climate 23, 7; 10.1175/2009JCLI3386.1

i1520-0442-23-7-1894-fa01

Fig. A1. Frequency distribution of cloud amount difference from (left) MODIS CF (%), (right) MODIS CFC (%), and GEOPROF-lidar (%) with regard to AMSR-E SIC (%). Medians are overlaid as thick black line.

Citation: Journal of Climate 23, 7; 10.1175/2009JCLI3386.1

i1520-0442-23-7-1894-fa02

Fig. A2. Daily mean cloud amount September 2007 anomaly using (top) (left) MODIS CF and (right) MODIS CFC; (bottom) (left) the difference of the September 2007 anomalies of CF and CFC (CFC minus CF) and (bottom) the difference between CF and CFC (CFC minus CF).

Citation: Journal of Climate 23, 7; 10.1175/2009JCLI3386.1

Save