Polar snow and ice masses exert a profound influence on global climate and ocean circulation, and are in turn influenced by climate. Satellite remote sensing constitutes the only practical and systematic means of gaining long-term overviews of any change or variability that may be occurring in key snow and ice parameters. Current relevant satellite sensors are reviewed and future developments evaluated to determine how these may lead to improved retrievals of the key parameters. Sensors to be launched on satellite platforms planned for the end of the century and beyond include both improved versions of existing sensors (such as MODIS) and new classes of sensors (such as imaging spectrometers and laser rangers/altimeters) applied to polar remote sensing for the first time.

A survey of ice growth and decay processes on a selection of shallow and deep sub-Arctic and Arctic lakes was conducted using radiometrically calibrated ERS-1 SAR images. Time series of radar backscatter data were compiled for selected sites on the lakes during the period of ice cover (September to June) for the years 1991–92 and 1992–93. A variety of lake-ice processes could be observed, and significant changes in backscatter occurred from the time of initial ice formation in autumn until the onset of the spring thaw. Backscatter also varied according to the location and depth of the lakes. The spatial and temporal changes in backscatter were most constant and predictable at the shallow lakes on the North Slope of Alaska. As a consequence, they represent the most promising sites for long-term monitoring and the detection of changes related to global warming and its effects on the polar regions.

Trends in the annual minimum sea-ice extent, determined by three criteria (absolute annual minimum, minimum monthly mean, and the extent at the end of August), were investigated for the Barents and western Kara seas and adjacent parts of the Arctic Ocean during 1984–1993. Four definitions of ice extent were examined, based on thresholds of ice concentration: >90%, >70%, >40%, and >10% (El, E2, E3, and E4, respectively). Trends were studied using ice maps produced by the Russian Hydro-Meteorological Service, Kosmos and Okean satellite imagery, and data extracted from published literature. During 1984–1993, an increasing trend in the extent of minimum sea-ice cover was observed in the Barents, Kara, and combined Barents–Kara seas, for all ice-extent definitions. Root-mean-square differences between hydro-meteorological ice maps and satellite-image ice classifications for coincident areas and dates were 15.5%, 19.3%, 18.8%, and 11.5%, for ice extensions El–E4, respectively. The differences were subjected to Monte Carlo analyses to construct confidence intervals for the 10-year ice-map trends. With probability p = 0.8, the average 10-year increase in the minimum monthly mean sea-ice extent (followed in brackets by the average increase in the absolute annual minimum ice extent) was 12–46% [26–96%], 31–71% [55–140%], 30–69% [26–94%], and 48–94% [35–108%] in the Barents Sea; 20–60% [32–120%], 10–45% [20–92%], 2–36% [13–78%], and 10–47% [8–69%] in the Kara Sea; and 9–43% [26–59%], 9–41% [30–63%], 8–41% [22–52%] and 15–51% [21–51%] in the combined Barents–Kara seas, for ice concentrations El–E4, respectively. Including published data from 1966–1983, the trend in minimum monthly mean sea-ice extent for the combined 28-year period showed an average reduction of 8% in the Barents Sea and a 55% reduction in the western Kara Sea; ice extent at the end of August showed an average reduction of 33% in the Barents Sea.

This paper describes the development of a practical algorithm for the classification of sea-ice types from ERS-1 synthetic aperture radar (SAR) data. The algorithm was based on a combination of grey level and texture information in order to overcome ambiguous grey level values of different ice types. The problem of calculating texture parameters for windows containing more than one ice type was overcome by first segmenting the image so that only pixels from the same segment were included in the calculation of the texture measure. The segmentation procedure was based on the iterative application of a speckle noise reduction filter, and was thus crucially dependent on the ability of such a filter to smooth out noise without destroying edges and fine features. In order to achieve this, a modification to the sigma filter of Lee (1983b) was developed; it out-performed the sigma filter for a model problem. Two ERS-1 SAR scenes of the marginal ice zone east of Spitsbergen in March 1992 were analysed by calculating values of grey level and range for different ice types contained within raw data extracts. Although the grey levels of some of the ice types overlapped, most of the ambiguity was removed through the additional use of range. It was also necessary to test for the wave-like appearance of open water. The classification scheme was demonstrated to identify correctly most of the grease/new ice, first-year ice, multiyear ice, rough ice, pancake ice, and open water in the two SAR scenes, although there was some misclassification of open water as first-year ice.

A new emphasis on understanding natural systems at large spatial scales has led to an interest in deriving ecological variables from satellite reflectance images. The normalized difference vegetation index (NDVI) is a measure of canopy greenness that can be derived from reflectances at near-infrared and red wavelengths. For this study we investigated the relationships between NDVI and leaf-area index (LAI), intercepted photosynthetically active radiation (IPAR), and biomass in an Arctic tundra ecosystem. Reflectance spectra from a portable field spectrometer, LAI, IPAR, and biomass data were collected for 180 vegetation samples near Toolik Lake and Imnavait Creek, Alaska, during July and August 1993. NDVI values were calculated from red and near-infrared reflectances of the field spectrometer spectra. Strong linear relationships are seen between mean NDVI for major vegetation categories and mean LAI and biomass. The relationship between mean NDVI and mean IPAR for these categories is not significant. Average NDVI values for major vegetation categories calculated from a SPOT image of the study area were found to be highly linearly correlated to average field NDVI measurements for the same categories. This indicates that in this case it is appropriate to apply equations derived for field-based NDVI measurements to NDVI images. Using the regression equations for those relationships, biomass and LAI images were calculated from the SPOT NDVI image. The resulting images show expected trends in LAI and biomass across the landscape.

UK-based operations that range from ship routing and resource exploration to weather forecasting and glaciology have direct and growing interests in the oceans of the polar regions. Typically, information describing sea-ice conditions in localised regions is required on short time scales. To explore this market, the UK's Defence Research Agency, as part of a programme of the British National Space Centre, has commissioned the development of a prototype sea-ice workstation by a consortium led by Earth Observation Sciences Ltd.

The sea-ice workstation (SIWS) uses data from several current earth observation sensors, thereby combining the advantages of regional survey, all-weather capability, and high-resolution imagery. The workstation has been designed to run with a minimum of operator intervention in order to optimise speed of operation and ensure consistency of results. The geophysical processing chains generate charts of the ice edge, ice type, ice concentration, ice-motion vectors, and sea-surface temperatures.

Although taking full advantage of developments made elsewhere, the project has also made significant progress in research into the automated mapping of ice types. Existing ice-motion algorithms have been significantly enhanced as well. Considerable emphasis is being placed on the validation of the results from the system in order to assess their quality, this being one of the major concerns of potential users. The sea-ice workstation was completed in July 1994 and will form the basis for a series of evaluations that are intended to assess the value of the system for mapping and monitoring sea ice.

Systematic image-classification methods were applied to ALMAZ, ERS-1, and JERS-1 synthetic aperture radar (SAR) and Landsat-TM multispectral satellite images to evaluate the relative information content of the satellite data for discriminating wet tundra habitats in northern Alaska. Results suggest that SAR data can be used to concurrently detect a maximum of four or five landcover classes using the methods of this study. Combining two or more SAR images from different satellites improved the detection of some classes, particularly water bodies. Combining full-resolution SAR data with Landsat-TM did not improve the detection capabilities of Landsat-TM alone. Further research is needed to assess other image-classification and SAR data processing methods.

The patterns of the normalized difference vegetation index (NDVI) on three glacial surfaces of different ages in the vicinity of Toolik Lake, Alaska, were examined. NDVI was derived from SPOT multispectral digital data, and the images were stratified according to boundaries on glacial geology and vegetation maps. Ground-level measurements of NDVI from common vegetation types were also collected, using a portable spectrometer. Late Pleistocene glacial surfaces have lower image-NDVI than older Middle Pleistocene surfaces, and the mean NDVI is correlated with approximate time since deglaciation. The trends are related to differences in NDVI associated with vegetation growing on mineral vs peaty substrates. Nonacidic mineral substrates are more common on the younger landscapes, and acidic peaty soils are more common on the older surfaces. The field-NDVIs of acidic dry, moist, and wet tundra are consistently higher than those of corresponding nonacidic tundra types. These same trends are seen when the SPOT NDVI image is stratified according to vegetation boundaries appearing on two detailed vegetation maps in the region. Above-ground biomass of moist and wet acidic tundra is significantly greater than corresponding nonacidic types. Vegetation species composition was examined along two transects on the oldest and youngest glacial surfaces. Shrub cover is the most important factor affecting the spectral signatures and biomass. Older surfaces have greater cover of shrub-rich tussock tundra and shrub-filled water tracks, and the younger surfaces have more dry, well-drained sites with low biomass and relatively barren nonsorted circles and stripes. These trends are related to paludification and modification of the terrain by geomorphic and geochemical processes. Similar patterns of spectral reflectance have been noted in association with a variety of large-scale natural disturbances in northern Alaska. However, extrapolation of these results to much broader regions of the circumpolar Arctic will require the use of sensors covering larger areas, such as the AVHRR aboard the NOAA satellites.

Advanced Very High Resolution Radiometer (AVHRR) satellite data were acquired and composited into twice-a-month periods from 1 May 1991 to 15 October 1991 in order to map vegetation characteristics of the Alaskan landscape. Unique spatial and temporal qualities of the AVHRR data provide information that leads to a better understanding of regional biophysical characteristics of vegetation communities and patterns. These data provided synoptic views of the landscape and depicted phenological diversity, temporal vegetation phenology (green-up, peak of green, and senescence), photosynthetic activity, and regional landscape patterns. Products generated from the data included a phenological class map, phenological composite maps (onset, peak, and duration), and photosynthetic activity maps (mean and maximum greenness). The time-series data provide opportunities to study phenological processes at small landscape scales over time periods of weeks, months, and years. Regional patterns identified on some of the maps are unique to specific areas; others correspond to biophysical or ecoregional boundaries. The data provide new insights to landscape processes, ecology, and landscape physiognomy that allow scientists to look at landscapes in ways that were previously difficult to achieve.

In the future, data from the moderate resolution imaging spectroradiometer (MODIS) will be employed to map snow in an automated environment at a resolution of 250 m to 1 km. Using Landsat thematic mapper (TM) data, an algorithm, SNOMAP, has been developed to map snow-covered area. This algorithm will be used, with appropriate modification, with MODIS data following the launch of the first Earth Observing System (EOS) platform in 1998. SNOMAP has been shown to be successful in mapping snow in a variety of areas using TM data. However, significant errors may be present in mountainous areas due to effects of topography. To increase the accuracy of mapping global snow-covered area in the future using MODIS data, digital elevation model (DEM) data have been registered to TM data for parts of Glacier National Park, Montana, so that snow cover on mountain slopes can be mapped. This paper shows that the use of DEM data registered to TM data increases the accuracy of mapping snow-covered area. Using SNOMAP on a subscene within the 14 March 1991 TM scene of northwestern Montana, 215 km2 of snow is mapped when TM data are used alone to map the snow cover. We show that about 1062 km2 of snow are actually present as measured when the TM and DEM data are registered. Approximately five times more snow is present when the effects of topography are considered for this subscene, which is in a rugged area in Glacier National Park. A simple model has been developed to determine the relationship between terrain relief and the amount of correction that must be applied to map actual snow-covered area in Glacier National Park using satellite data alone.

The Earth's high-latitude regions are of critical importance in many climate-change scenarios, but a time continuous, spatially complete, and well-calibrated record of tropospheric temperatures is needed in order to assess past and future climate changes. Studies of recently compiled upper-air data sets show no evidence of CO2-induced warming, but the spatial pattern of tropospheric temperature variability in the Arctic has not been thoroughly examined. This study analyzes a 108-month segment of the data record from the TIROS Operational Vertical Sounder (TOVS) aboard NOAA polar-orbiting satellites to examine both the spatial and temporal variability of atmospheric temperature in the Arctic.

Temperature retrievals based on clear-column radiances archived at NOAA/NESDIS were done using algorithms tuned to Arctic conditions. The retrieved temperatures compared well with Arctic rawinsonde data, and include lowlevel inversions that are often problematic for satellite retrievals. The amplitude of the seasonal cycle in 500 mbar temperatures from the TOVS, NMC, and rawinsonde data generally agreed, whereas the phase comparisons produced mixed results. Principal component analyses of the TOVS and NMC temperatures revealed both monopole and dipole spatial patterns in the component loadings. Spatial patterns of the correlation between the upper-air data and the TOVS retrievals were similar to the principal component loading patterns. Whereas no significant trends were found in the station data for the same period as the TOVS record, a significant negative trend could be seen in the first principal component scores of the TOVS retrievals.

Four case studies are presented that focus on a joint analysis of AVHRR images with other independently obtained data sets to identify examples of both optically thick and thin polar stratospheric clouds (PSCs) over Antarctica. Optically thick PSCs are shown to be readily identifiable by the temperature signatures in AVHRR channel 5 images over the Antarctic Peninsula–Weddell Sea area. A time series of optically thick PSCs as seen in satellite images obtained at Palmer Station, Antarctica, is produced and discussed. Four events are identified and discussed where the areal coverage by cold temperatures indicative of PSCs expands to cover a significant portion of the area sampled. Plans for further research are outlined.

At the Bonanza Creek Experimental Forest (BCEF), past ecological research has been directed at forest successional processes on the floodplain of the Tanana River and adjacent uplands. Research at the Bonanza Creek site continues on the mosaic of forests, shrublands, and wetlands in a wide variety of successional stages on the Tanana floodplain. This paper reviews research since 1988 into the capabilities of Synthetic Aperture Radar (SAR) for monitoring, classification, and characterization of these forests using radar remote sensing and modelling techniques. Classifications of successional stages, obtained by use of different classifiers on multi-frequency and multi-polarimetric AIRSAR data, are contrasted; these classifications have been used to predict classification accuracies obtained with ERS-1 data, and to estimate the utility of an ERS-1 and RADARSAT combination for classification. Forest classifications, used in combination with ground-truth data for more than 50 forest stands, are used to summarize the distribution of biomass on the landscape. This will allow projections of future biomass. Monitoring of forest phenology, seasonality of flooding, and freeze–thaw transitions is ongoing. Also, direct monitoring of dominant tree species is demonstrating diurnal variation and interrelationships among environmental, physiological, and backscatter measurements.

The Finnish reindeer-herding area lies in the northern boreal forest zone. The herding area is divided into 57 reindeer management districts with a total area of 114,355 km2 (excluding water areas). The official number of reindeer in the year 1992 was 263,789, or 2.3 reindeer per square kilometre. In 1946 the density was only about one reindeer per square kilometre. The increased pressure on the mid-winter lichen pastures has reduced the volume of lichen. Also the decreased area of old coniferous forests with arboreal lichens has diminished the supply of food during the late-winter months. The lack of suitable winter pastures has led to the need for supplementary feeding with hay.

The intention of this study was to develop and test methods for cost-effective reindeer range land inventory and to assess the present state of pastures in seven districts. Use was made of Landsat 5 TM satellite imagery that was processed with remote sensing and GIS software. The analysis of well-known field sites proved the image classification to be between 80 and 90% accurate. By comparing field data and classification results, estimates can be made on the available fodder reserves in the different pasture types. The results show that mid-winter pastures are both scarce and degraded in six of the seven areas investigated. Only one area (Muotkatunturi) has mid-winter pastures in reasonably good condition. Late-winter pastures with arboreal lichens are more widespread than mid-winter pastures and determine the amount of additional winter feeding required. Summer pastures are abundant in all seven reindeer-herding districts and are therefore not a limiting factor. The method used is objective, inexpensive, and fast when compared to conventional methods. Information can be produced on the quantity and quality of reindeer pastures. This knowledge can be used to make better estimates of the natural grazing capacity of reindeer districts in Finland.

Carbon flux measurements made at an elevated point are associated with an effective upwind area or ‘footprint.’ Since Arctic tundra landscapes can exhibit substantial heterogeneity within the footprint of an eddy correlation tower, it may be necessary to determine the relative point source contributions to the observed flux if landscape properties are to be related to the flux. This study evaluates the potential importance of representing footprint source contributions in relationships that are developed between tower observations of net ecosystem exchange of carbon dioxide (NEE) and a remotely sensed spectral vegetation index. Satellite data collected over the foothills region of the North Slope of Alaska are used to determine spatial patterns of a spectral vegetation index in the calculated footprints of 30 randomly selected tower locations. A previously developed relationship between NEE and the vegetation index is used to calculate NEE at each tower location using two techniques, one that explicitly considers the footprint pattern of relative contributions to tower fluxes and another that ignores these patterns. The results indicate that if carbon fluxes measured at a tower are to be related to remotely sensed spectral vegetation indices, then it is necessary to consider the relative flux contributions from within the tower footprints for sites on the North Slope of Alaska.

Biogenic emissions of sulfur from the ocean surface are believed to be a significant contribution to the atmospheric aerosol burden, thus playing a significant role in climate. The possibility exists for using remotely sensed data to locate sources, map distributions, and estimate global-scale fluxes of marine sulfur flux. By definition, estimates of surface trace-gas flux from satellites are indirect. Empirical algorithms must be derived using direct surface flux measurements. This technology does not currently exist, and may be many years from coming to fruition. This paper discusses the possibilities of developing a satellite-based dimethyl sulphide (DMS) flux capability and describes a new technique that can be used to develop the necessary empirical relationships. It demonstrates the feasibility of using a sulfur chemiluminescence detector (SCD) for measuring surface sulfur-gas flux directly from the ground. It also estimates ocean surface sulfur-gas flux using two related, indirect methods, known as the variance method and the inertial dissipation method. These methods can be used in the Arctic, where the ocean-to-atmosphere flux may be a significant fraction of global biogenic sulfur emissions.

An area of Arctic vegetation on Brøggerhalvøya in the northwestern part of Spitsbergen, was studied. A classified Landsat-TM image with 57 land-cover classes (LCCs) was compared to an existing vegetation map with a comparison matrix. The 57 LCCs were grouped in seven different groups, where the seventh group contained 15 LCCs with vegetation. The other groups were regarded as non-vegetated. The correspondence between the existing vegetation map and the satellite-based vegetation map was not high, but the classes from the satellite-based land-cover map could be interpreted in botanical terms. A vegetation index (VI) was used to discriminate between vegetated and non-vegetated LCCs. The VI was correlated to plant biomass measurements, and the total plant biomass for Brøggerhalvøya was estimated. The estimated biomass coincided with other estimates from the same area.