SciELO - Scientific Electronic Library Online

Home Pagelista alfabética de revistas  

Servicios Personalizados




Links relacionados


Gayana (Concepción)

versión impresa ISSN 0717-652Xversión On-line ISSN 0717-6538

Gayana (Concepc.) v.68 n.2 supl.TIProc Concepción  2004 


Gayana 68(2) supl. t.I. Proc. : 311-316, 2004 ISSN 0717-652X



Erica L. Key & Peter J. Minnett

University of Miami, Rosenstiel School for Marine and Atmospheric Science Meteorology and Physical Oceanography Division 4600 Rickenbacker Causeway Miami, Florida, USA 33149


Global cloudiness distributions, though an important component in radiative and hydrological budgets, are neither adequately known nor easily retrieved by the spatial and spectral resolutions afforded by current satellite instrumentation. At high latitudes, cold, high albedo surfaces present a particular challenge to cloud retrieval, offering little or no thermal or visible contrast for cloud-ice discrimination. It is in these frequently cloudy and climate-sensitive regions that changing cloud amounts and optical parameters enact the greatest influence, enhancing or suppressing melt through cloud base emission of longwave radiation or scattering of incident shortwave radiation. Polynyas and leads, seasonally ice-free areas characterized by intense air-sea fluxes of heat and moisture, are useful features for exploring the relationships between cloud cover and the underlying surface. Using polar-optimized CASPR (Cloud and Surface Parameter Retrieval) algorithms to process multi-channel AVHRR radiances, cloud amounts, microphysics, and surface forcing are evaluated and validated against in situ measurements collected in several polynyas and leads across the Western Arctic during the years 1992-2000.



Cloud cover is an essential component of the Earth's energy budget; yet, it is poorly represented in global climate simulations and over-simply and inaccurately parameterized. These limitations in cloud rendering have added significance in irregularly or sparsely sampled areas, which experience variable cloud conditions and where radiative uncertainties translate into large surface heat budget errors and questionable climatic responses. The Arctic typifies one of these low data volume areas with a high incidence of cloud. It is further characterized by a rapidly-changing high-albedo surface, which modifies cloud feedbacks, and thus, variations in boundary layer air temperature, moisture, and radiative profiles. Recent surface observations of warming (e.g. Overland et al., 2002) and sea ice retreat in this region (e.g., Serreze et al., 2003) have renewed interest in the role of Arctic clouds in the coupled air-ice-ocean system and their capacity for moderating deep water formation, freshwater export, weather teleconnections, and, ultimately, global climate change.

Clouds are known to alter the radiative profile through both scattering and absorption of incident solar radiation as well as emission of longwave radiation from cloud base. When the scattering efficiency of the cloud outweighs its longwave contribution, the surface below the cloud is cooled, prompting ice re-freeze, a negative forcing. In the opposing scenario, the longwave emission prevails over scattering effects, warming the surface and encouraging melt, a positive forcing. These radiative interactions constitute the surface cloud radiative forcing over a given area, a factor used to determine the impact of clouds on insolation. This differs from the net cloud forcing, which takes into account the response of the surface to this incident radiation, requiring precise measurement of the surface albedo as well as the longwave surface emission. Inevitably, however, some percentage of this upwelling radiation does enter into the calculation of surface cloud radiative forcing. Due to the high albedo of ice and snow, a significant portion of the energy which reaches the surface, is reflected skyward where it may again interact with clouds. Such multiple reflections add a positive bias to radiation measured at the surface, but the magnitude of this bias is generally unknown.

If the summation of the downwelling forcing components is positive, and favors ice melt, then the exposed low-albedo ocean surface would absorb more incoming radiation, supporting additional lateral melt. The open water area becomes a source of heat, moisture, and convective energy, which promotes new cloud formation and possible further melt. Alternatively, negatively-forcing clouds decrease surface insolation, allowing ice to re-form, reducing the heat and moisture exchange across the air-sea interface, and the energy necessary for new local cloud formation. Through these various interactions, freshwater and brine production, heat and moisture fluxes, ocean and ice area are affected, giving rise to higher order feedbacks on regional scales, such as the impact on deepwater formation and freshwater export. In turn, these synoptic and regional variations cascade to longer time scales, affecting seasonal, interannual, and climatic patterns, both hemispherically and globally (Deser et al., 2000; Cavalieri et al., 2003).

Measurement of Arctic clouds at any scale has proved challenging to both in situ and satellite instruments. This difficulty is due in part to the seasonal variations in daylight over polar regions, which provides no illumination for detection of wintertime cloud by visible means (see also Hahn et al., 1995). Accordingly, most in situ cloud measurements are collected in summer or transition months when some hours of daylight are available for visual observation of cloud amount and type, either by human observers or automated surface-based sensors. Satellite retrievals, on the other hand, capitalize on the frequent sampling rate of polar orbiting sensors and multi-channel radiances to detect cloud year-round at high temporal resolution. Measurements, such as those from the 23-year record of the 5-channel AVHRRs (Advanced Very High Resolution Radiometers) aboard the NOAA (National Oceanic and Atmospheric Administration) satellites, include information in the visible (VIS), near infrared (NIR) and infrared (IR) wavelengths at 1km resolution. Classification using the latter of these wavebands, however, may not distinguish cloud from the underlying snow and ice surface due to a lack of thermal contrast between the two. In order to capitalize on the long time series of AVHRR data while overcoming these spectral and spatial limitations, it becomes necessary to blend the product with modeled fields and use polar-optimized calibrations and algorithms to capture the cloud variability and assess its interaction with the underlying surface.

Using natural models is yet another way to approach variability, sampling, and connectivity issues in the Arctic. Polynyas, ice-free areas in otherwise ice-covered waters, provide a range of surface albedo, meteorological forcing, and radiative conditions with which to test the sensitivity of downwelling radiation to changing cloud cover. Although bound by a common definition, each polynya is formed, maintained, and forced uniquely, interacting with local topography, ice flow, subsurface water masses, and the overlying atmosphere. Their more linear counterpart, flaw leads, exhibit a strong sensitivity to mechanical forcing; though, through lateral melting, they may occupy a surface area equivalent to polynyas (o103-105 km2). Although the formation and maintenance of these ice-free areas is achieved through a variety of physical forcing mechanisms, some atmospheric and others oceanic, the combined open water area produced is a small fraction of the total Arctic area. However, it is through these "windows in the ice" that the majority of Arctic heat exchange occurs (Maykut, 1978; 1982), modifying the boundary layer and providing necessary heat and moisture for new cloud formation.

Clouds forming over open water areas do not limit their surface forcing to polynya or lead boundaries, but extend their influence into remote areas. Hence, this analysis uses an amalgam of data from a variety of sources, including in situ measurements ­ shipboard, coastal station, and specialized land sites ­ remote sensing retrievals, and modeled cloud parameters to determine the variability in Arctic cloud cover and its relationship with the underlying sea-ice-snow surface. The combined data set totals eleven months of measurements collected during an 8-year time span, 1992-2000 (Figure 1, Table 1). All voyages were made between April and October, permitting the use of polar daylight for best possible cloud illumination.

Figure 1: Bathymetric map and satellite ice cover imagery for each research site. From east to west, they are the Northeast Water (NEW), North Water (NOW), Barrow Flaw Lead (BFL), and St. Lawrence Island Polynya (SLIP).


Cloud amount, type, and distribution among three vertical levels are analyzed to WMO standards from all-sky images reflected in a hemispheric dome mirror mounted on top of the ship's bridge, away from visual obstructions. The 2p sky images, recorded every 17 seconds by an RGB camera and time-lapse video recorder, are reviewed by a meteorologist who observes cloud movement at low, middle, and high levels to assess cloud cover at 10-minute intervals. In contrast to instantaneous measurements, in which lower clouds may obscure upper levels, leading to a surface observer bias and an underestimation of total cloud cover, the 10-minutes' of imagery provides an ample sampling window for estimating upper level cloud through breaks in the lower and middle cloud layers. Key et al. (2004), Hanafin and Minnett (2001), and Minnett (1995) further discuss the advantages of all-sky measurements and detail polynya cloud cover as measured in each of the six study regions.

Analogous hemispheric radiation measurements are collected at 1-minute intervals with Eppley pyranometers and pyrgeometers, also sited away from obstructing or shadowing ship superstructure. Both sensors are calibrated, well-ventilated units, mounted on gimbals to minimize the effects of ship motion on measured downwelling short- and longwave radiation. Meteorological information is recorded by a Coastal Environmental Systems' Weatherpak and includes minute averages of air temperature, wind speed and direction, relative humidity, and barometric pressure. Additional air temperature measurements are derived from infrared spectra in the 14.3 to15.4 µm range observed by the Marine-Atmospheric Emitted Radiance Interferometer (M-AERI), at 14-minute intervals.

These cloud, radiation, and meteorological data are used to quantitatively estimate the surface cloud radiative forcing, which is defined by equations developed by Ramanathan et al. (1989):


Thus, the calculation of cloud effect is summarized as the difference between clear-sky (0) and all-sky (c) downward radiative fluxes observed at or near the surface. Measurements of clear-sky radiation under all-sky conditions are achieved through parameterization, using equations which are optimized for polar conditions and functions of routinely-measured variables such as surface air temperature. The sign of the forcing determines whether the clouds are preferentially scattering shortwave radiation (negative forcing) or absorbing and re-emitting radiation in the longwave (positive forcing).

In order to analyze cloud forcing over the full polynya scale and determine the influence of open water on downwind areas, Cloud and Surface Parameter Retrieval (CASPR) software was implemented (Key, 2002). CASPR retrieval algorithms were applied to twice-daily 25 km AVHRR APP-x (Advanced Polar Pathfinder-x) pixels augmented with collocated NCEP-Reanalysis gridded surface and pressure level data by Dr. Jeff Key of the University of Wisconsin-Madison and made available for analysis. Final output files, including cloud microphysical fields (e.g., cloud particle effective radius, cloud optical depth, cloud top temperatures and pressures), top-of-atmosphere and surface radiation, and a net cloud forcing, were provided. Each variable is valid for a domain 263 x 263 pixels, extending from 47.11° to 89.87°N latitude and 179.73°E to 179.82°W longitude, an area that covers all six polynyas.

While the polynya environment encompasses a broad scope of variability with which to exercise CASPR, not all variables necessary for accurate validation of polynya conditions were measured at all times. Some statistical ranges based upon data collected in the same month of a different year or in close proximity to the polynya were often used as proxies for in situ conditions. Cloud microphysical measurements, which were not sampled during any of the field seasons, were gathered from aircraft survey literature, which was focused primarily north of Barrow over the Beaufort Sea. From these cloud-specific historical data, an assumed characterization, including cloud effective radius, liquid water content, particle shape and phase, for each cloud species was contrived. Variations in these cloud parameters as well as surface type and coverage over a representative range of solar zenith angles were then used to evaluate CASPR microphysical retrievals.


Calibrated, geo-located AVHRR radiances at a 25 km resolution were collocated with NCEP-Reanalysis profile and surface data for processing by CASPR algorithms. The results include gridded cloud and radiation information not routinely gathered at Arctic weather stations or on ships of opportunity, and provide this information over four surface types (snow, ice, land, and ocean) detailed in the AVHRR landmask. Each cloud and radiation parameter was spatially averaged twice once over polynya ice and ocean pixels, and again as a large-scale regional average over all ice and ocean surfaces north of 47.11°. Data for 2000 were not available, excluding BFL00 from analysis.

Effective cloud radii derived over the remaining polynyas ranged between 2 and 42 µm, with larger mean particle sizes occurring over all polynyas except NOW99. Cloud optical depths were more evenly matched between the two spatial averages, generally measuring > 10 with occasional optically thick clouds over the polynyas; though, again, NOW99 optical depths were particularly low. This low bias persisted throughout the analysis for NOW99, which also registered a near zero incident cloud forcing over the length of the cruise. The other four polynyas received more insolation and downwelling longwave radiation despite higher, warmer cloud tops. A seasonal cycle in albedo over the length of NOW98 reveals similar polynya and hemispheric albedos in early spring, followed by a rapid decrease in albedo over the hemisphere while polynya values remain relatively high (0.5). As ice continues to melt in summer, the polynya albedo decreases until it again converges with the hemispheric albedo near zero in mid-summer. Other late summer cruises in the NEW reported low albedos, though these values were still greater than the hemispheric average, despite the large percentage of open water. Early spring albedo results, such as SLIP99, show a number of fluctuations, perhaps related to ice reformation and removal under high wind conditions conducive to polynya development.

Cloud forcing, calculated in much the same way as the Ramanathan et al. (1989) equation, corroborates daily-averaged values derived from the polynya data. At both hemispheric and polynya averages, clouds negatively force the underlying surface, by almost ­150 W·m-2 in all cases except NOW99. Diurnal signals in this forcing exist for all cruises, though a seasonal shift from positive to negative forcing in NOW98 also appears near year day 140, corresponding approximately to the time of failure of the ice bridge in Smith Sound.

This multi-dimensional analysis of cloud forcing over Arctic polynyas reinforces the importance of these ice-free areas to the Arctic climate system. The abundance of heat and moisture within and downwind of the polynya boundaries encourages cloud formation, resulting in loci of extremely high cloud occurrence within an already cloudy environment. Despite the relatively dry Arctic atmosphere, the large cloud effective radii, persistence, and ubiquity of cloud are sufficient to affect the state of the underlying cryosphere, alternatively amplifying and retarding melt over the daylight season.


Although the CASPR software is not a model in the strictest sense, but is instead a set of algorithms and radiative transfer codes with which calibrated satellite data and modeled fields may be analyzed, it provides a number of parameters, which could not otherwise be deduced from simple raw radiances over a number of polar surfaces. Comparisons of CASPR cloud and radiative products over a combination of ocean and ice were conducted over hemispheric and polynya areas at a 12-hour, 25-km resolution.

Increased heat and moisture over each polynya positively affected the cloud effective radii, cloud top temperature, cloud top pressure, and, on occasion, cloud optical depth. The majority of this effect was registered over ice, which is usually in agreement with the polynya average. Hemispheric averages, which despite covering the central Arctic, more closely followed the oceanic component and tended towards clouds with finer particles at lower concentrations. Due to the low cloud top temperature and cloud top pressure of these clouds, the analysis would suggest that mid-to-high level clouds populate the hemisphere with low cloud predominantly forming over the convective polynya areas. While these results could be affected by satellite perspective and algorithm inadequacies over bright surfaces, theories of low-cloud formation in the Arctic (Herman and Goody, 1976) would agree with the tendency toward low cloud over polynyas and leads.


The authors wish to acknowledge funding for this work from NASA NAG56577, NASA NAS5-31361, and NSF OPP 9708045, and ship support from the USCGC Polar Sea, USCGC Polar Star, and CCGS Pierre Radisson. All-sky camera cloud analyses were performed by Robert A. Jones from the University of Miami and APP-x data provided by Dr. Jeff Key of the Advanced Satellite Products Team, National Oceanic and Atmospheric Administration, National Environmental Satellite, Data, and Information Service, Madison Wisconsin.


Cavalieri, D.J., C.L. Parkinson, & K.Y. Vinnikov, 2003, 30-Year satellite record reveals contrasting Arctic and Antarctic decadal sea ice variability, Geophysical Research Letters, 30, doi:10.1029/2003GL018031. [ 1]

Deser, C., J.E. Walsh, & M.S. Timlin, 2000, Arctic sea ice variability in the context of recent atmospheric circulation trends, Journal of Climate, 13, 617-633. [ 2]

Hahn, C.J., S.G. Warren, & J. London, 1995, The effect of moonlight on observation of cloud cover at night, and application to cloud climatology, Journal of Climate, 8, 1429-1446. [ 3]

Hanafin, J.A. & P.J. Minnett, 2001, Cloud forcing of surface radiation in the North Water Polynya during NOW '98, Atmosphere-Ocean, 39, 239-255. [ 4]

Herman, G. & R. Goody, 1976, Formation and persistence of summertime Arctic stratus clouds, Journal of Atmospheric Science, 33, 1537-1553. [ 5]

Key, E.L., P.J. Minnett, & R.A. Jones, 2004, Cloud distributions over the coastal Arctic Ocean: surface-based and satellite observations. Atmospheric Research, in press. [ 6]

Key, J. 2002. The Cloud and Surface Parameter Retrieval (CASPR) system for Polar AVHRR User's Guide, Cooperative Institute for Meteorological Satellite Studies, University of Wisconsin, Madison, WI 53562, 61 pp. [ 7]

Maykut, G.A., 1978, Energy exchange over young sea ice in the central Arctic, Journal of Geophysical Research, 83, 3646-3658. [ 8]

Maykut, G.A., 1982, Large-scale heat exchange and ice production in the central Arctic, Journal of Geophysical Research, 80, 7971-7984. [ 9]

Minnett, P.J., 1995, Measurements of the summer surface heat budget of the Northeast Water Polynya in 1992, Journal of Geophysical Research, 100, 4309-4322. [ 10]

Overland, J.E., M. Wang, & N.A. Bond, 2002, Recent temperature changes in the Western Arctic during Spring, Journal of Climate, 15, 1702-1716. [ 11]

Ramanathan, V., R.D. Cess, E.F. Harrison, P. Minnis, B.R. Barkstrom, E. Ahmad, & D. Hartmann, 1989, Cloud-radiative forcing and climate: results from the Earth Radiation Budget Experiment, Science, 243, 57-63. [ 12]

Serreze, M.C., J.A. Maslanik, T.A. Scambos, F. Fetterer, J. Stroeve, K. Knowles, C. Fowler, S. Drobot, R.G. Barry, & T.M. Haran, 2003, A record minimum arctic sea ice extent and area in 2003, Geophysical Research Letters, 30, doi:10.1029/2002GL016406. [ 13]


Creative Commons License Todo el contenido de esta revista, excepto dónde está identificado, está bajo una Licencia Creative Commons