Search

Remarkable correlation exists between warm water, poleward winds over the ocean, and low sea ice concentrations and extents over the winter sea ice pack around Antarctica, even to the point of continuing paterns across the ice-sea boundary. Since the wind stress associated with poleward wind is expected to compact the ice during sea ice edge retraction, we conclude that the memory of the Antarctic circumpolar wave in the sea ice pack is carried from one austral winter to the next by the neighboring water temperatures since the sea ice pack retracts nearly to Antarctica in austral summer.

We extend earlier analyses of a 8.8-year sea ice data set that described the local seasonal variations and trends in each of the hemispheric sea ice covers to the recently merged 18.2-year sea ice record from four satellite instruments. The seasonal cycle characteristics remain essentially the same as for the shorter time series, but the local trends are markedly different, in some cases reversing sign. The sign reversal reflects the lack of a consistent long-term trend and could be the result of localized long-term oscillations in the hemispheric sea ice covers. By combining the separate hemispheric sea ice records into a global one, we have shown that there are statistically significant net decreases in the sea ice coverage on a global scale. The change in the global sea ice extent is −0.01 ± 0.003 × 106 km2 per decade. The decrease in the areal coverage of the sea ice is only slightly smaller, so that the difference in the two, the ice-free areas within the packs, has no statistically significant change.

Satellite passive-microwave data for November 1978 through December 1996 reveal marked seasonal, regional, and interannual variabilities, with an overall decreasing trend of −34,300±3700 km2/yr (−2.8%/decade) in Arctic sea ice extents over the 18.2-year period. Decreases occur in all seasons and on a yearly average basis, although they are largest in spring and smallest in autumn. Regionally, the Kara and Barents Seas have the largest decreases, at −15,200±1900 km2/yr (−10.5%/decade), followed by the Seas of Okhotsk and Japan, the Arctic Ocean, Greenland Sea, Hudson Bay, and Canadian Archipelago. The yearly average trends for the total, the Kara and Barents Seas, and the Seas of Okhotsk and Japan all have high statistical significance, with the null hypothesis of a 0 slope being rejected at a 99% confidence level. Regions showing increasing yearly average ice extents are Baffin Bay/Labrador Sea, the Gulf of St. Lawrence, and the Bering Sea, with only the increases in the Gulf of St. Lawrence being statistically significant at the 99% level. Hemispheric results for sea ice areas exhibit the same −2.8%/decade decrease as for ice extents and hence a lower absolute decrease (−29,500±3800 km2/yr), with the ice-free area within the ice pack correspondingly decreasing at −4800±1600 km2/yr. Confidence levels for the trends in ice areas and ice-free water areas exceed 99% and 95%, respectively. Nonetheless, interannual variability is high, and, for instance, the Arctic Ocean ice extents have a positive trend 1990–1996, in spite of their negative trend for the time period as a whole.

We have generated consistent sea ice extent and area data records spanning 18.2 years from passive-microwave radiances obtained with the Nimbus 7 scanning multichannel microwave radiometer and with the Defense Meteorological Satellite Program F8, F11, and F13 special sensor microwave/imagers. The goal in the creation of these data was to produce a long-term, consistent set of sea ice extents and areas that provides the means for reliably determining sea ice variability over the 18.2-year period and also serves as a baseline for future measurements. We describe the method used to match the sea ice extents and areas from these four multichannel sensors and summarize the problems encountered when working with radiances from sensors having different frequencies, different footprint sizes, different visit times, and different calibrations. A major obstacle to adjusting for these differences is the lack of a complete year of overlapping data from sequential sensors. Nonetheless, our procedure reduced ice extent differences during periods of sensor overlap to less than 0.05% and ice area differences to 0.6% or less.

The “Odden” is a large sea ice feature that forms in the east Greenland Sea that may protrude eastward to 5 °E from the main sea ice pack (at about 8 °W) between 73° and 77 °N. It generally forms at the beginning of the winter season and can cover 300,000 km2. Throughout the winter the outer edge of the Odden may advance and retreat by several hundred kilometers on timescales of a few days to weeks. Satellite passive microwave observations from 1978 through 1995 provide a continuous record of the spatial and temporal variations of this extremely dynamic phenomenon. Aircraft synthetic aperture radar, satellite passive microwave, and ship observations in the Odden show that the Odden consists of new ice types, rather than older ice types advected eastward from the main pack. The 17-year record shows both strong interannual and intra-annual variations in Odden extent and temporal behavior. For example, in 1983 the Odden was weak, in 1984 the Odden did not occur, and in 1985 the Odden returned late in the season. An analysis of the ice area and extent time series derived from the satellite passive microwave observations along with meteorological data from the International Arctic Buoy Program (IABP) determined the meteorological forcing associated with Odden growth, maintenance, and decay. The key meteorological parameters that are related to the rapid ice formation and decay associated with the Odden are, in order of importance, air temperature, wind speed, and wind direction. Oceanographic parameters must play an important role in controlling Odden formation, but it is not yet possible to quantify this role because of a lack of long-term oceanographic observations.

We have done a frequency analysis of the 9-year record of sea ice concentrations obtained with the NASA Nimbus 7 scanning multichannel microwave radiometer (SMMR). Because of the periodic revisit time of the SMMR, we are able to examine recurrence of shorter-period oscillations, as aliased by the revisit frequency, and longer-period oscillations per se. A number of findings result from a sequence of two analysis schemes, as follows. We have first determined, then removed, the mean, linear trend, and seasonal cycle (the latter composed of five harmonics of the annual cycle) from the time series for each pixel of gridded SMMR ice concentration data in the Arctic, using multiple linear regression. We find that the 9-year trends are most positive in the Beaufort and East Greenland Seas, whereas the most negative areas are in the Barents Sea and the sea ice surrounding Svalbard ; this is similar to trend patterns in the lengths of the sea ice season reported earlier by Parkinson [1992]. In the residual data the standard deviation ranges from 12% to 16% in the central Arctic basin, considerably above the estimated wintertime ice concentration accuracies of 5% in that region ; this is in part due to fluctuations in the ice concentration rather than noise. Secondly, we have used Fourier analysis and multiple-taper filtering techniques to identify tentatively some of these fluctuations as tidal components or multiyear components similar to those found in the El Nino-Southern Oscillation Index, and we show their spatial distribution. However, there are some high-amplitude, periodic components that we have not yet been able to identify ; we show the spatial distribution of one of these components.

Understanding the passive microwave emissions of a snowpack, as observed by satellite sensors, requires knowledge of the snowpack properties : water equivalent, grain size, density, and stratigraphy. For the snowpack in the Upper Colorado River Basin, measurements of snow depth and water equivalent are routinely available from the U.S. Department of Agriculture, but extremely limited information is available for the other properties. To provide this information, a field program from 1984 to 1995 obtained profiles of snowpack grain size, density, and temperature near the time of maximum snow accumulation, at sites distributed across the basin. A synoptic basin-wide sampling program in 1985 showed that the snowpack exhibits consistent properties across large regions. Typically, the snowpack in the Wyoming region contains large amounts of depth hoar, with grain sizes up to 5 mm, while the snowpack in Colorado and Utah is dominated by rounded snow grains less than 2 mm in diameter. In the Wyoming region, large depth hoar crystals in shallow snowpacks yield the lowest emissivities or coldest brightness temperatures observed across the entire basin. Yearly differences in the average grain sizes result primarily from variations in the relative amount of depth hoar within the snowpack. The average grain size for the Colorado and Utah regions shows much less variation than do the grain sizes from the Wyoming region. Furthermore, the greatest amounts of depth hoar occur in the Wyoming region during 1987 and 1992, years with strong El Nino Southern Oscillation, but the Colorado and Utah regions do not show this behavior.

Selected data obtained during the first year of operation of the scanning multichannel microwave radiometer (SMMR) on board the Nimbus 7 satellite (launched in late October 1978) have been used to calculate, on a global basis, various geophysical parameters over open oceans, polar regions, and terrain. Over open oceans these calculations have provided values for sea surface temperatures, near-surface winds, atmospheric water vapor in a column, and rainfall rates. In polar regions, sea ice concentration, multiyear ice fraction, and radiating temperatures have been obtained. Finally, the extent and water equivalence of snow cover over terrain have been calculated. These parameters have been compared with in situ measurements of the same geophysical parameters, where available, and the results of these comparisons are described. The self-consistency of the global displays of all the parameters is discussed along with the plans for archiving them for subsequent research purposes. A description of the SMMR calibration and data processing scheme is also given.

Aircraft observations of the microwave emission from the wind-driven foam-covered Bering Sea substantiate earlier results and show that the combination of surface roughness and white water yields a significant microwave brightness temperature dependence on wind speed over a wide range of microwave wavelengths, with a decreasing dependence for wavelengths above 6 cm. The spectral characteristic of brightness temperature as a function of wind speed is consistent with a foam model in which the bubbles give rise to a cusped surface between the foam and the sea. In the fetch-limited situation the contribution of the wave structure at the surface appears to increase as the foam coverage decreases. Although the data show that the thin streaks are the most important part of the white water signature, there is some evidence for the contribution of whitecaps.

Microwave and infrared data were obtained from a research aircraft over the Beaufort Sea ice from the shoreline of Harrison Bay northward to a latitude of almost 81 deg N. The data acquired were compared with microwave data obtained on the surface at an approximate position of 75 deg N, 150 deg W. Over this north-south transect of the polar ice canopy it was discovered that the sea ice could be divided into five distinct zones. The shorefast sea ice was found to consist uniformly of first-year sea ice. The second zone was found to be a mixture of first-year sea ice, medium size multiyear floes, and many recently refrozen leads, polynyas, and open water; considerable shearing activity was evident in this zone. The third zone was a mixture of first-year and multiyear sea ice which had a uniform microwave signature. The fourth zone was found to be a mixture of first-year sea ice and medium-to-large size multiyear floes which was similar in composition to the second zone. The fifth zone was almost exclusively multiyear ice extending to the North Pole.

A method of determining sea ice parameters using Nimbus 7 polarized multispectral radiance data obtained with the Nimbus 7 Scanning Multichannel Microwave Radiometer (SMMR) is presented. Observed radiances from selected areas in the Arctic region for the period February 3-7, 1979 were used in computing algorithm coefficients. Polar maps of sea ice concentration, multiyear fraction, and ice temperature are illustrated for this period. The variation of the mean and standard deviation of ice concentration and multiyear ice fraction for a region of perennial ice cover over the first 11 months of SMMR operation is also presented. Comparisons are made between the calculated sea ice parameters and information obtained from previous studies using aircraft, submarine and surface observations. The absolute accuracy of the SMMR parameters remains uncertain.

Microwave radiometry has been used for the remote sensing of soil moisture in a series of aircraft flights over an agricultural test area in the vicinity of Phoenix, Arizona, The radiometers covered the wavelength range 0.8–21 cm. Ground truth in the form of gravimetric measurements of the soil moisture in the top 15 cm were obtained for 200 fields at this site. The results indicate that it is possible to monitor soil moisture variations with airborne radiometers. The emission is a function of the radiometer wavelength and the distribution of the moisture in the soil. At a wavelength of 1.55 cm there is little or no variation in the emission for soil moisture values below 10 or 15% moisture content by weight. Above this value there is a linear decrease in the emission with a slope of approximately 3°K for each percentage point increase in soil moisture. On the other hand, the emission was observed to be a linear function of soil moisture over the range 0–35% at a wavelength of 21 cm. Results of calculations using a layered model are presented that indicate the importance of the surface moisture in determining the emission from the soil.

A combination of remote sensing from an aircraft and simultaneous surface measurements have confirmed the feasibility of identifying old and new sea ice according to its emission of thermal radiation at wavelengths between 0.3 and 3 cm. Emissivity of first-year thick ice with a surface temperature of about 260 K is 0.95 or greater for wavelengths between 0.81 and 11 cm; the emissivity of multiyear ice is 0.8 at 0.81 cm and 0.95 at 11 cm, increasing monotonically in this wavelength interval. The ease with which multiyear ice can be distinguished from first-year ice using a passive microwave radiometer is demonstrated by comparing mosaics prepared both from photographs and images of 1.55-cm radiation.

An algorithm has been developed for estimating total and multiyear sea ice concentration from passive microwave and surface air temperature measurements. The algorithm was made for use with Nimbus 7 scanning multichannel microwave radiometer (SMMR) data. It is based on radiation physics and may thus easily be modified to suit other passive microwave instruments. A comparison between Nimbus 7 SMMR and aircraft microwave measurements indicates that estimates of total ice concentration are accurate to ±3% and those of multiyear ice concentration to ±10%. These accuracies are not valid during the melt season when the snow on the ice is wet. For the first time such a comparison of independent estimates has been performed to validate the capability of measuring sea ice coverage from space. The ability of the SMMR to follow moving patches of multiyear ice and of rain/wet snow areas has been demonstrated. From the concentration estimates the sharpness of the ice edge can be estimated. The need for accurate concentration estimates for reliable heat budget estimates in the Arctic is also discussed.

This paper compares satellite data on the marginal ice zone obtained during the Marginal Ice Zone Experiment in 1984 by Nimbus 7 with simultaneous mesoscale aircraft (in particular, the NASA CV-990 airborne laboratory) and surface observations. Total and multiyear sea ice concentrations calculated from the airborne multichannel microwave radiometer were found to agree well with similar calculations using the Nimbus SMMR data. The temperature dependence of the determination of multiyear sea-ice concentration near the melting point was found to be the same for both airborne and satellite data. It was found that low total ice concentrations and open-water storm effects near the ice edge could be reliably distinguished by means of spectral gradient ratio, using data from the 0.33-cm and the 1.55-cm radiometers.

The extent and area of sea ice and the amount of open water within the ice boundaries have been computed for the Arctic, Antarctic, and global ice covers every 2 days from November 1978 to August 1987 utilizing the Nimbus 7 scanning multichannel microwave radiometer (SMMR) observations. During this 9-year period, no significant trends were found in the ice areas of the Arctic and Antarctic packs or in their sum, the global ice area. Also, no significant trends were found in the maxima and minima of the annual ice extents for either polar pack. However, during these 9 years, significant negative trends were found in the annual global ice extent maxima (5%) and in the annual maxima of the global open water area within the sea ice boundaries (7–14%). Significant negative trends were also found in the annual open water maxima in the Antarctic (5–10%) and the annual open water minima in the Arctic (13–23%). The open water within the ice boundaries averaged approximately 14% in the Arctic and 18% in the Antarctic, at the time of their maximum extents. This negative trend in the global ice extent maxima during 1978–1987 also occurred during 1973–1976, found by an analysis of the monochromatic microwave observations obtained by the Nimbus 5 satellite. The combined 15-year negative trend is 6%. A significant change between the 1973–1976 and the 1978–1987 periods in the Antarctic ice pack is that the very large polynya that formed and persisted in the Weddell Sea during three out of four of the austral winters in the earlier period did not form in any of the 9 years of the latter period. This study, in conjunction with recent studies of global temperature change, lends support to the thesis that climatic changes in the global average temperature might be detectable by observing variations in global sea ice extent.

Sea ice concentration contour maps derived from the 0.81 cm wavelength channels on the Nimbus 7 scanning multichannel microwave radiometer (SMMR) are compared with surface and aircraft data and with sea ice charts derived from the TIROS-N satellite imagery. The regional ice and weather data and the derivation of the TIROS ice charts are discussed, and the Nimbus 7 SMMR data and the physical basis of the algorithm used to retrieve ice concentrations are described. A detailed comparison of the SMMR imagery with all available surface, satellite, and meteorological data for two days in March, 1979 is made. The results show that the ice concentration maps derived from the 0.81 cm wavelength SMMR radiances provide accurate information on both the ice edge position and the location of low and high ice concentration regions within the Bering Sea pack ice.

Aircraft remote sensing data collected during the 1984 summer Marginal Ice Zone Experiment in the Fram Strait are used to compare ice concentration estimates derived from synthetic aperture radar (SAR) imagery, passive microwave imagery at several frequencies, aerial photography, and spectral photometer data. The comparison is carried out not only to evaluate SAR performance against more established techniques but also to investigate how ice surface conditions, imaging geometry, and choice of algorithm parameters affect estimates made by each sensor.Active and passive microwave sensor estimates of ice concentration derived using similar algorithms show an rms difference of 13 percent. Agreement between each microwave sensor and near-simultaneous aerial photography is approximately the same (14 percent). The availability of high-resolution microwave imagery makes it possible to ascribe the discrepancies in the concentration estimates to variations in ice surface signatures in the scene.

The most thorough field study of the Bering Sea Marginal Ice Zone (MIZ) attempted to date was conducted February 5–27, 1983. This study, MIZEX West, was part of a larger program addressing processes which control interactions among the atmosphere, ice, and oceans in the northern hemisphere MIZ's [Muench, 1983a]. The other part of this overall program, MIZEX East, addresses processes in the Greenland Sea MIZ [McPhee, 1983a].

The data acquired during the summer 1984 Marginal Ice Zone Experiment in the Fram Strait-Greenland Sea marginal ice zone, using airborne active and passive microwave sensors and the Nimbus 7 SMMR, were analyzed to compile a sequential description of the mesoscale and large-scale ice morphology variations during the period of June 6 - July 16, 1984. Throughout the experiment, the long ice edge between northwest Svalbard and central Greenland meandered; eddies were repeatedly formed, moved, and disappeared but the ice edge remained within a 100-km-wide zone. The ice pack behind this alternately diffuse and compact edge underwent rapid and pronounced variations in ice concentration over a 200-km-wide zone. The high-resolution ice concentration distributions obtained in the aircraft images agree well with the low-resolution distributions of SMMR images.

A technique is presented which improves existing methods of calculating sea ice concentrations from microwave radiances by reducing weather-related effects over open ocean areas and in the vicinity of marginal sea ice zones. Winds, atmospheric water vapor, cloud liquid water, and rain increase the microwave emission over these regions and thus result in erroneous values of computed sea ice concentration. The method described is based on the microwave spectral properties of sea ice and ice-free ocean and utilizes ratios of the polarized radiances at the 0.81-cm (37 GHz) and 1.7-cm (18 GHz) wavelengths. Following a discussion of the physical basis for this technique, examples are provided which demonstrate its utility. While the technique was developed for use with the Nimbus 7 scanning multichannel microwave radiometer data, it is applicable also to data from other microwave radiometers operating in a similar wavelength range.