The polar regions are extremely sensitive to global climate change, yet in many regards they are among the most poorly studied systems in the world. While Chapter 3 identified critical observational data sets that could be used to provide better detection and early warning of impending global change, this chapter assesses whether NASA's current polar observational programs and data sets provide the information that is most needed. The assessment, when related to the original scientific questions, will allow judgement of the adequacy of current data-collection efforts and help to formulate advice on ways to improve NASA's overall polar program strategy. The committee's assessment strategy is to briefly review NASA's contributions to development, evaluation, and availability of the data sets for each variable, to identify how NASA could enhance its current contributions, and to identify gaps and what NASA can do to fill the gaps. After the variable-by-variable review, the committee summarizes NASA contributions and identifies key gaps in the current generation of polar geophysical data sets.
This summary provides the basis for recommendations presented in Chapter 5. The variables assessed here include:
surface heat fluxes
land surface characteristics
terrestrial CO2 and CH4 flux
ice sheet elevation
ice sheet dynamics
sea ice concentration
sea ice thickness
sea ice velocity
ocean surface temperature and salinity
sea surface height
ocean productivity and CO2 flux
wildlife habitat and migration
ASSESSMENT OF DATA SET AVAILABILITY BY VARIABLE
Atmospheric profiles depict the vertical variation of quantities such as air temperature, humidity, and wind. Current conventional meteorological data sets (e.g., collections available at the National Snow and Ice Data Center [NSIDC]) generally do not provide adequate spatial and temporal coverage, especially over ocean and sea ice portions of the polar regions. Even so, some first-order analyses of energy and water transports into the Arctic (e.g., Peixoto and Oort, 1992) and Antarctic (e.g., Bromwich et al., 1998; Giovinetto et al., 1992) have been attempted. Satellite-based temperature and humidity profiles are available from NOAA satellites, but there are substantial uncertainties associated with these products. NASA makes little contribution to this area, except in supplying NOAA data to European centers; the major activities are conducted by NOAA's National Center for Environmental Prediction (NCEP), The Laboratoire Meteorologic Dynamique (LMD), and the European Center for Medium-range Weather Forcasting (ECMWF). NASA has also sponsored a re-analysis of the NOAA data sets in the Arctic, which has improved the operational analysis being done by LMD. NASA should ensure that coverage is extended to the south polar regions. Although the Atmospheric Infrared Sounder/Advanced Microwave Sounding Unit (AIRS/AMSU) may be of some use, NASA's potential contributions are limited unless it sponsors work on advanced, multi-instrument analysis techniques in preparation for Net Primary Production (NPP) and the National Polar-orbiting Operational Environmental Satellite System (NPOESS) and its forerunner, the NPOESS Preparatory Project (NPP).
These advanced techniques could also be applied retrospectively to the Advanced Very High-Resolution Radiometer (AVHRR)/High-Resolution Infrared Sounder/Microwave Sounding Unit data sets. Given the challenges associated with surface-based sensors in the extreme and isolated polar regions, flight of a wind-profiling mission could be very useful, but the persistent cloudiness of polar regions could be a constraint.
Existing satellite data sets (two global, two Arctic only) provide appropriate coverage, but their accuracy in the difficult polar conditions is still unknown (e.g., Curry et al., 1996; Rossow and Schiffer, 1999). The properties determined are areal cover, cloud top temperature, optical thickness or emissivity, and mean particle size. The Arctic data sets are held at the NSIDC, one global data set is held at Langley Distributed Active Archive Center (DAAC), and one global data set is partially archived at the Goddard DAAC. NASA supports the production of all four of these data sets; however, NASA's contribution would be significantly improved if it supported the studies needed to evaluate the accuracy of these data sets. These studies could be an extension of NASA's First ISCCP Regional Experiment—Arctic Cloud Experiment (FIRE-ACE) Project based on Surface Heat Budget of the Arctic project (SHEBA) and Atmospheric Radiation Measurement (ARM)—Barrow surface radar and lidar data sets and also on more effective use of conventional surface weather observations, which are available from the Oak Ridge DAAC. There is also a need for extended modeling studies of the polar atmospheric boundary layer and its associated cloudiness using recent field data sets.
In the near future, NASA's CloudSat and Picasso missions should add crucial missing information about cloud vertical structure needed for a more accurate assessment of the radiative effects of polar clouds. There will be a need for targeted analyses of CloudSat, IceSat, and Picasso data to characterize the vertical distribution of clouds in polar regions.
There are no extensive aerosol data sets for the polar regions. Existing satellite retrieval methods have not been applied to this region and are unlikely to be successful over snow- and ice-covered surfaces. Little surface-based data are available. New instruments (MISR, MODIS) are unlikely to provide the needed improvement in this difficult situation, although GLAS (the Geoscience Laser Altimeter System to be carried on ICESat, scheduled for launch in mid-2001) has the potential to provide some useful information on the physical properties and distribution of
polar aerosols. Because it is unclear how critical this issue is for understanding current polar processes, model experiments incorporating field measurements could provide a basis for the identification of key processes and measurement needs.
Existing satellite data sets (two global, two Arctic only) provide appropriate coverage except for the fact that the infrared sensors (e.g., TOVS Path-P) cannot provide surface temperatures when cloud fractions exceed 90%, which leads to a bias and to an incomplete description of energy exchanges at the surface. Although there are four relevant data sets (the two Arctic-only data sets are held at the NSIDC, one global data set is held at the Langley DAAC, and the remaining global data set is partly archived at the Goddard DAAC), little is known of their accuracy. Although NASA has not made a focused contribution to the development of the needed all-sky product, it has supported most of the analyses that have produced this quantity as a by-product; however, this support is waning just when research interest is growing. There are unexploited (microwave) data sets that could improve the information if they were analyzed in combination with infrared measurements; so NASA could make a significant contribution by supporting advanced analyses of combined measurements (this could be a side benefit of careful sea ice product intercomparisons).
Surface Heat Fluxes
Routine meteorological data at the surface (e.g., surface air temperature, winds, and humidity) are available at the NSIDC, but coverage is incomplete, especially over ocean and sea ice and over Antarctica. NASA is not a major contributor, although it could contribute significantly by sponsoring work on advanced multi-instrument analysis techniques in preparation for NPP and NPOESS.
NASA supports two major efforts to determine surface radiative fluxes for the whole globe, including the polar regions. These are state of the art; evaluation under the difficult polar conditions can be done using data from SHEBA and ARM-Barrow. This is a key contribution to polar research that should continue, since such data are needed for characterizing the surface energy budget of all polar surface types. Currently, there is no NASA contribution to the determination of surface turbulent fluxes, except for support for FIRE-ACE, which is ending. Extending studies to modeling of the surface boundary layer and its associated clouds would provide some advancement on this topic: Progress could come from fur-
ther analysis of the SHEBA and ARM-Barrow data sets, together with modeling studies, but may also require some additional field studies.
The onset of melt on ice sheet surfaces is relatively straightforward to detect. However, measurement of melt rate either directly or through measurement of individual components of the surface energy balance is problematic. Direct measurement of spatial melt extent requires frequent sampling in time; this type of coverage is only available at the coarse resolution of passive microwave sensors and scatterometers. Satellite sensing of surface conditions is further complicated by clouds and difficulties in cloud masking in the visible and near-infrared (IR). Moreover, remotely sensed estimates of melt extent have yet to be quantitatively translated into ablation rates. Measurement uncertainties in several components of the surface energy balance make the combined errors large, thereby limiting the accuracy of estimated changes of surface elevation due to melt, as well as estimation of runoff from snow packs on land. Integrated analysis of available and future satellite measures of albedo and surface temperature, in combination with sensible and latent fluxes, may provide a solution; however, representative in situ studies and instrumentation are required. For ice sheet margins these include the continuation of measurements in Greenland (funded by NASA and the National Science Foundation [NSF]) and in Antarctica (funded by NSF). Re-analyses incorporating these measurements need to be evaluated and most likely improved. The optimum approach to improved estimates of melt rates will likely involve a combination of forecast or re-analysis data sets (at improving resolutions), mesoscale models forced by these observations, automated weather station (AWS) data, and thermal-IR and microwave emission data sets to produce internally consistent pictures of surface processes.
Both accumulation and surface energy balance require an improved characterization of meteorological conditions at the ice surface; storm tracks, drainage winds (also called katabatic winds), and boundary-layer stability all play a role. Current AWS instrumentation in Greenland is adequate for characterizing many of these variables locally; the continuation of this time series is essential to efforts to understand processes and integrate them into climate models. Similar data are available in Antarctica through NSF efforts, although the region is larger and station density is low; future continuity on either ice sheet is not guaranteed. Better integration of the measurements with other research efforts is required.
Two existing satellite data sets (one global, one Arctic only) provide appropriate coverage (the Arctic data set is held at the NSIDC and the global data set is held at the Langley DAAC), but these data sets are limited to the visible portion of the spectrum. For sea ice and deep snow, these may be sufficient, since the reflection spectrum is known fairly well, but for shallower snow and vegetation mixtures on land, where the spectrum is not well characterized, these data sets may not be adequate for many applications. Moreover, time resolution may be inadequate for defining melt or thaw onsets (this may be insurmountable because of persistent cloud cover). MODIS data provide complete spectral coverage for the first time, so if clear sky can be successfully identified, analysis of these data will provide the best measure of surface albedo; thus, thorough analysis of the entire time record of MODIS is an important contribution. In particular, there is a need for targeted analysis of MODIS data to obtain better surface albedos for mixtures of snow and vegetation.
Although there are many conventional data sets for precipitation, they do not provide adequate coverage of the polar regions. The one extensive snow cover data set does not contain any information about snow cover on ice. Global precipitation data sets combining satellite and ground-based data are available from other agencies such as the WMO's Global Precipitation Climatology Center and the Global Precipitation Climatology Project. For Arctic applications, the NSIDC maintains historical archives of land-station precipitation from the Former Soviet Union and Canada.
A variety of spatially extensive (in situ and satellite) precipitation (P) data are available as global products that would encompass the polar regions. The WMO-GPCC Global Precipitation Climate Center (GPCC located in Offenbach Germany), serves as official custodian of precipitation data for the World Meteorological Organization (WMO). The Center maintains major compilations of station records, error-checked and merged to yield the most comprehensive station-based P archive (and derived products) for the globe. New fields and time series are developed on an operational (2-month delay) basis, ( www.dwd.de/research/gpcc). The standard “monitoring product” starts in 1986, at 1° and 2.5° resolution.
The Global Precipitation Climatology Project (GPCP) is part of the Global Energy and Water Cycle Experiment (GEWEX) of the World Climate Research Program. Currently available GPCP products (Version 1c and 2.x) combine precipitation from microwave (Special Sensor Micro-
wave/Imager, SSM/I) and infrared sensors at 2.5°. Version 2.x products incorporate the TIROS Operational Vertical Sounder (TOVS) and OLR Precipitation Index (OPI) for time periods when SSM/I was not available. The Version 1.c products cover 1987 through the present and Version 2.x covers 1979 to present.
In addition, precipitation data sets are available from individual research teams. For example, monthly climate records and time series (1901-95) of various climatic variables from the Climate Research Unit (CRU) at the University of East Anglia are available for land areas. These include precipitation station data from formal and informal sources, and gridded fields using a variety of interpolation methods.
There are many difficulties in using the available station records to develop spatially varying fields of precipitation. Gauge undercatch is well documented, especially for solid precipitation. The use of station monitoring data sets in a vast remote area that has low station density creates biases in interpolated precipitation amounts. Gaps in station records, re-location of stations, differences in gauge technology and wind correction factors used for snow all add to the difficulties in obtaining a clear picture. Use of un-manned, automated stations is a problem because of equipment failure in the harsh environment. In addition, the closure of stations due to the collapse of monitoring networks, especially in the Russian Arctic, adds to the uncertainty.
The use of coherent, spatially contiguous remote-sensing measurements could therefore make, in principle, a valuable contribution to the monitoring of high latitude precipitation. Despite the potential opportunity, NASA makes only a limited contribution to the direct measurement of precipitation, primarily in support of microwave analyses of snow cover (water content) on land. NASA has supported one recent precipitation-relevant data set, which is due to be released in early 2001; this data set will include horizontal vapor fluxes and moisture convergence or net precipitation (P-E). Net precipitation will not achieve all that is required. Given the importance of hydrological and energy/water balance questions in climate research, an effort to develop a satellite-based determination of precipitation, particularly ice phase, in the polar regions would seem to be of paramount importance and would eliminate a key deficiency of polar observations and data availability for NASA's Earth Science Enterprise science.
While NASA has not taken on a unique role in providing information on polar precipitation, it has supported science and data collection efforts, notably with respect to the special sensor microwave/imager (SSM/I) data products from Defense Meteorological Satellite Program (DMSP) satellites. NSIDC has also been active in identifying and securing ground-based measurements of precipitation and other meteorological variables
that can serve as validation for the satellite products. Information from these satellites has been incorporated into the global precipitation data sets noted above. Advanced Microwave Scanning Radiometer E (AMSR-E) data sets for rainfall should become available at the NSIDC from EOS and AQUA (NASA's first and second Earth Observing Systems), and assessments of the polar subsets of these data should be a high priority.
Permafrost-related research is not a prominent feature of NASA's high-latitude science program. The most comprehensive data related to distribution of permafrost is from digitized maps and expert opinion from the International Permafrost Association, but the resolution of the mapping is coarse. Remote-sensing approaches are not capable of directly mapping permafrost extent, but progress in mapping permafrost extent may be possible through approaches that combine remote-sensing technology, modeling of the soil thermal profile, and in situ measurements. In particular, data sets that describe the timing of freeze and thaw with adequate spatial coverage and spatial resolution may overcome some constraints. The NSIDC provides a land-based freezing/thawing/degree-day product based on climatology station records, but it does not provide adequate resolution or coverage of the polar regions, particularly in mountainous terrain. Analyses based on data from NASA scatterometers (NSCAT, SeaWinds) show some promise for monitoring the freeze-thaw status of the land mass, but the algorithms and technology need further evaluation and development as the scatterometers are not formally optimized for land-based cryospheric applications. It may also be possible to detect onset of surface melt from changes in surface microwave emissivity, which would go from relatively high emissivities when the surface is frozen to low values when the surface has standing water. Besides freeze and thaw data sets, other satellite-derived products that can contribute to monitoring or diagnosing permafrost dynamics include land-surface temperature, snow characteristics (cover, depth, water equivalent, thermal properties), vegetation characteristics, disturbance characteristics, and wetlands extent. Also, there are several remote-sensing technologies that could be brought to bear on monitoring thermokarst topography, which would aid in understanding permafrost dynamics.
Land Surface Characteristics
Important variables of land surface characteristics include canopy characteristics (leaf area, canopy density, albedo), structural composition (e.g., proportion of trees, shrubs, and tundra), land-cover type (classifica-
tion), wetland dynamics, and disturbance characteristics (timing, location, extent, severity). There are numerous NASA and non-NASA sensors that contribute to these data sets. Several European and Japanese sensors, similar to those supported by NASA, can also be used in estimating land cover. NASA's EOS, NASA's contribution to LANDSAT-7, NOAA/NASA SSM/I Pathfinder products (at the NSIDC), and the planned Next Millenium Program ALI sensor are important in this context. Through NASA's Alaska Synthetic Aperture Radar (SAR) Facility, NASA provides an important service to the U.S. and global research community by archiving and distributing European Earth Remote Sensing Satellite (ERS), Canadian Synthetic Aperture Radar Satellite (RADARSAT), and Japanese Earth Remote-Sensing Satellite (JERS) data sets; however, it must be stressed that due to foreign commercial ownership of these data, access to new data acquisitions is highly competitive and as such this service is limited for U.S. investigators, serves a small “NASA approved” user community, and places severe copyright restrictions on the redistribution of this information.
Products to measure land surface characteristics are not adequate for applications in high latitudes. For example, satellite-based technology is not able to monitor temporal changes in land surface characteristics of high-latitude regions, such as gradual changes in canopy characteristics associated with gradual vegetation changes and abrupt changes in land surface characteristics associated with disturbances, particularly fire. Changes in wetland extent are especially important to monitor in high latitudes. Wetland extent can be monitored with analyses of both optical data (multispectral) and microwave data (multitemporal), and a combination of multispectral and multitemporal data has been shown to improve the ability to delineate wetlands. Research in the Boreal EcosystemAtmosphere Study region has shown that wetlands were substantially underestimated with analyses based on 30-m TM resolution imagery, and were not identified at all in analyses based on 1-km AVHRR imagery. Thus, the estimation of wetland extent may require high-resolution satellite imagery (e.g., IKONOS) or aircraft imagery (e.g., CASI). With respect to fire disturbance, current satellite technology is capable of determining the timing, location, and extent of disturbance, but it is less clear how well disturbance severity can be determined. The estimation of disturbance severity, which includes effects on vegetation biomass and ground-layer carbon storage, may require analysis of a variety of remote-sensing data sets, including thermal anomalies at the time of fire, data sets on flaming and smoldering ratio, and land surface characteristics after the fire.
The simplest so-called reference surface techniques for measuring evapotranspiration use mean daily temperature as the measured characteristic. Complex, physically-based functions, typically employed in soil-vegetation-atmosphere-transfer schemes and process-based studies, require surface temperature (sometimes called skin temperature), humidity, net radiation, aerodynamic roughness, albedo, leaf area index, and wind speed. Several of these required input variables or parameters measured routinely by satellites. NASA supports the collection of such information either directly through its sensors or indirectly through data support. NASA efforts result in the provision of clear-sky surface temperature, surface albedo, radiation and cloud properties, and clear-sky surface air temperature. Vegetative properties such as land-atmosphere interactions and land cover, discussed earlier, are also essential. These data sets are not unique to the Arctic. Sub-setting and re-projection for Arctic applications is in several cases carried out by the NSIDC. While not space-based observations, variables such as wind, temperature, humidity, and pressure data over polar land areas are obtained from rawinsonde observational networks and are archived at the NSIDC. Important comparisons can be made by utilizing the NCEP and the ECMWF operational and re-analysis activities. Their products provide many of the same surface data sets as listed above, but are derived as blended model and assimilated observational products. Validation over grid-cell areas is needed in order to provide a basis for improved parameterizations of evapotranspiration in future re-analyses.
Passive microwave radiometers have been used to detect soil moisture. Systems such as SSM/I and AMSR have the advantage of acquiring this information on a routine basis (few days re-visit time) from space with a resolution on the order of 50 km or larger; however, interference from vegetation limits what can usefully be obtained from these sensors at this time. The Global Soil Wetness Project, as part of the NASA-funded International Satellite Land Surface Climatology Project, uses AVHRR-derived vegetation information and modeling to produce global soil moisture fields. These are not otherwise archived at the NSIDC or the ASF, although they are archived elsewhere within the DAAC system, this fact is not well known. A proposed NASA post-2002 mission to monitor soil moisture at 10-km resolution is under consideration. Higher resolution SAR data have also been shown to provide some information about soil
moisture, thereby pointing to a potentially important use of the ERS, JERS, and RADARSAT data sets archived at the ASF.
Terrestrial CO2 and CH4 Flux
Data sets required to model terrestrial CO2 and CH4 flux include absorption of photosynthetically active radiation, biomass, leaf area index, soil moisture, land surface temperature, precipitation, evapotranspiration, and land surface characteristics. Measurements relevant to these variables are provided by a variety of NASA and non-NASA sensors; however, many of the satellite-derived data sets required for estimating terrestrial CO2 and CH4 flux are not yet available to the scientific community. NASA could improve this situation by enhancing the availability of these data sets to the broader community. These data sets can be used to drive models for estimating the timing and location of trace gases exchanged with the atmosphere. Simple algorithms that use remote-sensing inputs have been developed to estimate components of CO2 exchange at 1-km resolution at regional and global scales (e.g., correlations between sensor signatures and field-measured fluxes, as well as production efficiency models) for mapping the spatial distribution of CO2 and trace gas exchanges across the landscape. Process-based models driven by input variables derived from remote-sensing technologies represent another possible approach to estimating the temporal and spatial dynamics of CO2 and CH4 in high-latitude regions. Although some of the controls over CO2 and CH4 fluxes in high-latitude regions are being measured with remote-sensing technologies, these data sets need to be evaluated to determine whether they are adequate for high-latitude applications.
Testing the temporal and spatial dynamics of models at different scales remains a major challenge, and data sets derived from remote-sensing technologies have an important role to play in testing aspects of forward-modeling approaches. NASA makes a valuable contribution in terms of ground-based flask and tower monitoring for carbon fluxes through FLUXNET, as part of a NASA EOS calibration and validation activity. Canopy chemistry, indicating vegetative stress and potential rates of nutrient recycling in ecosystems, can be derived from hyperspectral sensor data. The development of data sets on canopy chemistry, which is responsive to NH4 and NOx deposition rates, is important because NH4 and NOx deposition rates do not appear to be directly amenable to remote sensing.
River discharge on all but the largest rivers must typically be obtained from ground-based station data available through several non-NASA archives. Major sub-sets of these holdings have been organized into pan-Arctic data collection. One such archive, developed at the University of New Hampshire and funded by the NSF Arctic System Science, resides at the NSIDC. A post-2002 mission using some combination of radar altimetry, interferometric SAR or lidar may be able to detect morphological changes in river channels to infer discharge. Lake morphometry and stage variations, from which volumetric changes can be inferred, are also possible to determine using such sensors. A small archive of “ice-out” data for one river in Alaska is maintained by the NSIDC.
Variations in topography, slope, and aspect are important for defining snow storage and melt, sediment flux, stream channel velocities, potential floodplain inundation, and the presence of sheetflow. A global 1-km data set developed by the USGS (GTOPO30) is already in existence and widely used by the community < http://www.edcdaac/usgs.gov/gtopo3o/gtopo30.html>. Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) on TERRA (NASA's Earth Observing System first multi-sensor platform) and Geoscience Laser Altimeter System on ICESAT will provide NASA-sponsored data sets for surface topography. ASTER, in particular, will be able to supply very-high-resolution data with 30-m horizontal resolution.
Ice Sheet Elevation
In spite of poor temporal coverage, existing data sets from radar altimetry (SEASAT, GEOSAT, ERS) provide a picture of change in the high interior of ice sheets over the last few decades; this data stream continues but is of no use near the ice margins, where the most rapid variations occur. NASA airborne laser measurements have identified significant changes in the Greenland ice sheet over the last half of the 1990s. This result has provided the first definitive measurement of one ice sheet's contributions to the present rate of sea level rise. ICESAT will provide systematic coverage in Greenland and Antarctica and, with the aid of aircraft results, will allow the first comprehensive assessment of the present contribution of both large ice sheets to sea level rise. Note that extended time series of these measurements are necessary to understand the background variability of the processes that determine surface elevation and provide the opportunity to assess the causes of change; this is required to develop a predictive understanding of ice sheet variations.
Also required are data on accumulation rates and variations, melt rates, and changes in ice flow.
Surface elevation changes may be driven in part by changes in accumulation rate. Understanding the implications of these changes requires knowledge of the background (long-term) accumulation rate and its variability. NASA has funded efforts under the Program for Arctic Regional Climate Assessment Program to provide this information from shallow ice cores in Greenland. There are limited efforts of a similar nature in West Antarctica funded by the NSF. ICESAT interpretation will require data from all sectors of Antarctica. Broad-band airborne sounding radar can reveal the spatial variations of accumulation and NASA has funded a pilot project. The development of an operational version of this radar will greatly enhance our knowledge of spatial variations of accumulation over ice sheets and allow point measurements from ice cores to be placed in a regional context. An ice-penetrating radar that measures both accumulation variations and ice thickness would provide a unique opportunity for completing the ice thickness map of Antarctica. Such maps would provide major contributions to ice sheet mass budget studies and to ice sheet modeling.
Ice Sheet Dynamics
Surface elevation changes may be produced by changes in ice flow, especially in rapidly flowing outlet glaciers and ice streams. Interferometry and feature tracking in visible-band imagery makes measurement of ice flow from space a routine process if suitable data are available. Interferometric measurements can also reveal the position of the grounding line (the boundary between grounded and floating ice), which can shift location in response to changes in ice thickness. Changing rates of ice flow have been identified by a combination of field and satellite studies in both Greenland and Antarctica. Detection of changing flow rates and grounding-line positions is pursued on a glacier-by-glacier basis, depending on availability of data suitable for interferometry. Limited studies to date have shown that variation is common. Comprehensive coverage is necessary. Modified Antarctic Mapping Mission 2 should provide the first nearly comprehensive interferometric coverage of coastal Antarctica. Continuation of suitable SAR acquisition is important, both to cover areas not yet mapped and to follow areas that are changing. NASA should assure access to data for these studies. There would be significant glaciological return from access to data from a sensor optimized for change detection (e.g., one with more frequent revisits and longer wavelength [L band]). Understanding of ICESAT-detected elevation change will require measurement of ice flow variations over longer periods than are repre-
sented in the SAR data archive. The present acquisition strategies for LANDSAT-7 and ASTER are providing annual data sets that earlier high-resolution sensors did not provide. These data complement interferometric data, providing greater detail for change detection and helping to ensure continuity of coverage.
Predictability of ice sheet responses requires improvements to available ice sheet models, including defining the domain (through aircraft-based ice-penetrating radar measurements of ice thickness) and improving knowledge of basal conditions and constraining models (using improved measures of ice velocity and the age-depth relationships that come from internal layer distributions and ice-penetrating radar). Present ice flow models incorporate ice streams and outlet glaciers at a rudimentary level; improvements in resolution and incorporation of processes related to outlet glacier flow variation are needed to match newly available measurements of ice flow and variability. Model development is funded in a limited way by NASA and as part of field data interpretation efforts by NSF. Significant efforts are being made by a number of European countries. Efforts are needed to combine models and satellite- and aircraft-derived data to promote the evolution of model designs. In addition to constraining models, ice thickness data are also required to determine discharge rates in combination with ice velocity measurements.
The spatial distribution of snow cover and its characteristics, including snow water equivalent, are key variables, and attempts have been made to quantify snow cover and snow cover characteristics from space. NASA supports either directly or through science and data archive activities MODIS, AVHRR/GOES, and LANDSAT-7 snow products that can be used to map snow extent. NASA also supports passive microwave data set activities (for SMMR, SSM/I) and will fly its AMSR (on ADEOS-II) and AMSR-E (on AQUA) sensors. Of primary interest in the Arctic is the capability of passive microwaves to characterize snow depth and water equivalent; while this is being applied for snow water equivalent on land, snow data on ice and on ice sheets is not available. Again, NSIDC activities to collect and distribute land-based data sets for precipitation and, in this context, snow cover provide important auxiliary information for satellite snow retrievals.
Sea Ice Concentration
Sea ice concentration and extent fields are perhaps the most frequently used of all polar geophysical data sets. They provide high quality data
(daily 25-km resolution) except in the summer, when atmospheric and ice and snow surface moisture creates ambiguous returns. Passive microwave sensors have provided a time series starting in the 1970s that has been used to document climate change (e.g., Cavalieri et al., 1997) and validate numerical models. A confusing variety of these data sets are available to the public at the NSIDC Web site. Some of these have been generated by different algorithms, some pertain to only one sensor, some bridge several sensors, and some cover only certain space and time subsets. Some are provided as near-real-time data in a file separate from identical fields generated with a (presumably) more accurate but slower algorithm. NASA should help clear out this clutter by funding independent studies that clearly document the strengths and weaknesses of differing algorithms and that encourage long time series (i.e., merging of individual sensor data sets). Further work is also recommended on the determination of summer ice concentration.
Sea Ice Thickness
The amount of sea ice in a given region is determined by measuring the ice thickness distribution, which in general includes a category for very thin ice and open water. Our survey revealed much interest in this variable. Sea ice thickness is directly observable by in situ buoys, submarine sonars, or direct ice coring, all of which suffer from gaps in space and/or time. However, several new satellite methods have been introduced that show some promise. The most direct method has been developed by British researchers using ERS altimetric data to estimate thick (greater than ~0.5-1.0 m) ice thickness over a 10-km footprint. Preliminary studies have produced intriguing fields of mean thickness and interannual variance. Another satellite method uses AVHRR surface temperature observations to deduce thin (less than ~0.8 m) ice thickness, although this has been applied only to small space and time regions. Finally, RADARSAT SAR data have been processed through the RADARSAT Geophysical Processing System (RGPS), at ASF and JPL, to produce Lagrangian grids of thin (less than ~2.0 m) ice thickness. Changes in open water area are very accurately observed with this method, which then uses a simple thermodynamic model to calculate ice growth in divergent regions.
Recognition of the possibility of routinely measuring sea ice freeboard (and inferring thickness) using a satellite-based altimeter represents a significant step forward. NASA needs to investigate the utility of ICESAT in this regard, and if it is not suitable, to pursue other techniques and instruments. With regard to other methods, both AVHRR and RGPS “observations” rely heavily on simplifying thermodynamic assumptions,
although the resulting errors may in fact be relatively small. These methods (or some combination with or without numerical models) should be strongly encouraged by NASA. Validation and error analysis needs to be emphasized in these still-exploratory studies.
Sea Ice Velocity
Until recently sea ice motion fields were produced solely by tracking a relatively sparse network of drifting buoys. This has changed dramatically, to the point that there now exists a variety of ice motion data sets derived from satellites (SAR, AVHRR, passive microwave sensors, scatterometers) and from combinations of these with buoy data. This information has been used to track interannual-to-decadal changes in the polar regions, although much more could be done. Blending of data types and assimilation into numerical models seem to reduce errors relative to buoys, but more work is necessary to ascertain the best methods. All satellite-based methods suffer from decreased performance in the summer, owing to feature degradation and/or cloud issues. In addition, clouds obscure passive sensor algorithms during much of the year in Antarctic seas. Some algorithms also have problems near the ice edge.
NASA should continue to fund processing of passive microwave sea-ice-motion products (e.g., from the AMSR) since this represents a time series that started in the 1970s. New algorithm development should also be funded if it specifically targets the summer problem.
There is a need for assessments of the relative merits and potential user base of these various data sets. For example, SAR-derived ice motion data sets have lower error and higher spatial resolution relative to passive microwave. However, it is under-used by the observational and modeling communities, probably in part because it is new, but also perhaps because of its large data streams and Lagrangian gridding. NASA should encourage more communication and collaboration between SAR and modeling researchers (e.g., by funding joint workshops). It could prove useful to provide some form of RGPS data on an Eulerian grid such as that used by numerical modelers.
Ocean Surface Temperature and Salinity
In perennially ice-covered seas, sea surface temperature (SST) stays close to the freezing point and is thus not a particularly important variable. This is not the case, however, for areas with seasonal or no ice cover in the polar and sub-polar regions. These regions have significant seasonal and interannual SST variability that may in fact influence water mass properties at higher latitudes (e.g., the Nordic seas origin of recent
Atlantic water layer warming in the Arctic Ocean). Unfortunately, the calibration of global SST fields using AVHRR and other visible/IR satellite sensors is biased toward data-rich conditions (i.e., the tropics). This can lead to significant errors at high latitudes (Emery et al., in press). NASA can improve this situation by supporting measurements of in situ SST (and particularly, surface temperature) at high latitudes. Because one key issue at higher latitudes is changes of SST by storms, a combined IR-microwave method is needed. Instrument calibrations in high latitudes should also be encouraged as part of the global product. This has been proposed for high-latitude Arctic waters as part of the AVHRR polar Pathfinder Project, but has not yet been implemented. This Pathfinder Project has also created a preliminary integrated surface temperature data set that includes ocean, ice, and land, but only for the polar regions. This type of activity should be strongly encouraged, and on a global basis.
Sea surface salinity (SSS) in high latitudes is a crucial parameter, as it controls the ocean stratification and traces the circulation pathways of freshwater. To date it has been observed only by in situ means; however, this is changing as the use of low frequency passive microwave remote sensing to measure SSS has been demonstrated. A European mission known as SMOS (Soil Moisture and Ocean Salinity) will be launched in 2005. Accuracies will be several tenths of a practical salinity unit (psu) for a 200 km footprint. Unfortunately, accuracy decreases with decreasing sea surface temperature and with decreasing SSS, to perhaps 1 psu for typical polar conditions. This spatial resolution is probably inadequate for the polar regions, where land and ice boundaries make large open water regions relatively rare. Until the next generation of sensors, a promising avenue could be the use of aircraft SSS observations such as those performed by scientists at the Naval Research Laboratory (Miller, 2000). These obviously have much better spatial resolution and, if flown in the summer over Arctic coastal regions, could easily detect river water plumes even with 1-psu accuracy. Much of the justification for the SMOS mission centers on the role of the high latitudes in the global freshwater balance; NASA should strive to improve SSS sensors so that the high latitudes are not neglected in future measurement programs.
Sea Surface Height
Sea surface height (SSH) is used to determine the ocean surface geostrophic current. It is obtained from satellite altimeters with an accuracy of a few centimeters. The most precise observations are from the joint NASA-ERS Topex/Poseidon (T/P) satellite, which covers latitudes between 66° S and 66° N. Most work has focused on tropical (e.g., ENSO) and mid-latitude (e.g., Gulf Stream) phenomena, where SSH anomalies
can be as large as 100 cm; however, T/P has also been used to study the sub-polar gyres and the Antarctic Circumpolar Current. Altimeters on the ERS-1 and ERS-2 satellites are not quite as accurate, but cover a greater range of latitudes, up to 82° N/S. These data have been applied to some high-latitude problems, although with some difficulty, given the relatively weak SSH gradients found in polar waters; however, much more could be done with very high-latitude SSH data that could contribute to understanding of interannual and interdecadal climate change. At present, NASA has no plans for a polar radar altimeter beyond the latitudinal range of T/P, even as European efforts continue in this field with ERS and CRYOSAT. NASA's participation in future high-latitude radar altimetry programs is a possibility deserving of consideration.
Ocean Productivity and CO2 Flux
The measurements of ocean productivity and CO2 flux present different challenges depending on whether the area under consideration is open ocean or is influenced by sea ice. For open oceans, satellite-derived measurements of ocean color provide a means of quantifying phytoplankton chlorophyll. High resolution mapping (~1 km) of ocean color has become possible with Sea-Viewing Wide-Field-of-View Sensor. The continued availability of these measurements must be ensured to monitor ocean color at seasonal, interannual, and decadal time scales. In addition to ocean color, estimating ocean productivity and CO2 flux of open oceans requires other measurements, including photosynthetically active radiation, clouds, sea ice, sea surface temperature, surface albedo, and surface winds. Because sea ice is a bright target that saturates highly sensitive marine sensors, it is difficult to accurately determine ocean color in the vicinity of sea ice. As the transition area between ice and open water supports more complex food webs than the open ocean, the estimation of productivity and CO2 fluxes from satellite-based measurements is problematic in these areas. Also, photosynthetically active radiation, clouds, snow, and ice cover are necessary to predict light transmission accurately in these areas. Refinements to snow and ice cover observations are particularly critical as the vertical distribution (< 5 cm) and horizontal distribution (~1 m) of snow and ice thickness would improve estimates of ice algal productivity.
Wildlife Habitat and Migration
Changes in sea ice coverage, sea ice characteristics (e.g., thickness, lead and ridge distributions), permafrost extent and active layer characteristics can have important effects on wildlife (e.g., caribou herds, bow-
head whales, seals, walrus, polar bears). Satellite remote sensing may prove useful in monitoring changes in wildlife habitat, migration routes and migration patterns in the Arctic. Such changes will be important to native residents and other Arctic stakeholders. While satellite tracking of wildlife has proven to be feasible and has been implemented on a limited basis, coordinated activity of this kind has yet to be undertaken by the biospheric research community. A related need is the monitoring of sub-surface conditions, especially permafrost and active layer characteristics over Arctic terrestrial regions that may serve as wildlife habitats and/or migration routes.
SUMMARY OF MAJOR NASA CONTRIBUTIONS
NASA is contributing to the development, evaluation, and availability of many significant data sets of importance in addressing the science-driving questions. For instance, NASA has done well in supporting demonstration projects and field campaigns that have led to identifying the potential of applying remote-sensing technologies for the development of data sets depicting the spatial and temporal variability of key land-surface quantities: terrestrial CO2 and CH4 flux, freeze and thaw dynamics, snow cover, evapotranspiration, soil moisture, and other land surface characteristics. NASA also makes a valuable contribution in terms of ground-based flask and tower monitoring for carbon fluxes through FLUXNET, as part of a NASA EOS calibration and validation activity.
NASA has also supported analyses of satellite data collected by other agencies to determine atmospheric temperature and humidity profiles enhanced for polar conditions; surface temperature and visible albedo; cloud properties including polar-specific products; and surface radiative fluxes. NASA's contributions to the SHEBA/ARM/FIRE field programs have helped produce a valuable suite of in situ measurements for the calibration and validation of satellite-derived products for the Arctic.
A key NASA contribution has been to define and initiate a program of measurement in Greenland that has provided the first useful measure of mass change of an ice sheet. This program has addressed the effects of most variables that influence the measurements, providing context for understanding the implications of measured changes. It is clear from this work that a range of variables must be addressed to understand the implications of measured surface height changes. NASA is also contributing to the understanding of ice sheet surface energy balance, local meteorology, and the patterns of accumulation by supporting a network of AWS stations and several projects measuring accumulation variability and local height change on the Greenland ice sheet. This work uses surface and remotely sensed data.
NASA is contributing to the understanding of ice flow in Antarctica by supporting interferometric studies from RADARSAT and ERS SAR data in the ASF archive, and through limited acquisition of new data. The RADARSAT Antarctic Mapping Project mosaic has provided an excellent base map for change detection. The Modified Antarctic Mapping Mission 2 interferometric map of ice motion, when combined with other NASA-funded interferometric studies and field and remote-sensing work in Antarctica funded by NSF, will give the most comprehensive picture of ice flow to date; it will provide a reference for the detection of changing flow patterns in future interferometric studies.
NASA has also supported the processing and analysis of sea ice data collected by other agencies' satellites (e.g., the Department of Defense's DMSP/SSMI). The documentation and interpretation of variations of satellite-derived sea ice coverage has been a highly visible NASA contribution.
NASA contributes to the high-resolution mapping (~1 km) of ocean color through SEAWIFS. The continued availability of these measurements will permit the monitoring of ocean color at seasonal, interannual, and decadal time scales, thereby stimulating research in high-latitude ocean biogeochemistry. This high-resolution color mapping may also have terrestrial applications for the summer growth period in the Arctic.
BETTER MANAGEMENT OF NASA'S CURRENT GENERATION OF POLAR GEOPHYSICAL DATA SETS
Many of the variables identified in this chapter are already estimated in re-analyses, including atmospheric profiles, cloud properties, surface temperature, surface heat fluxes, surface albedo, precipitation, evapotranspiration and soil moisture. While the reliability of many of these estimates remains to be established, re-analyses and other data assimilation techniques are powerful vehicles for assessing the impacts of data, especially because re-analyses can accept observations whenever and wherever they are made. For example, AVHRR estimates of surface temperature are potentially valuable inputs to constrain the surface energy budget over data-sparse areas in re-analyses. Similarly, SAR and passive microwave observations can enhance land surface descriptions in re-analyses. Atmospheric retrievals of temperature and moisture amount are challenging problems over sea ice, yet reliable retrievals could provide important constraints on the reanalyzed atmospheric circulation in data-sparse areas such as the Arctic Ocean and the interior of Antarctica. A vigorous research program to optimize the uses of polar satellite-derived products in re-analyses will do much to eliminate concerns that polar remote sensing observations suffer from isolation and lack integra-
tion into the global context. Such an activity will also enhance the relevance of the measurements to the broader ESE program and hence to societal issues.
In addition to data assimilation activities, there are a number of other ways in which the committee believes that NASA can improve its current contributions. First, NASA could enhance the use of its products by polar users by sponsoring evaluation and intercomparison activities directed specifically at polar sub-sets of its global products. Candidates are products depicting clouds, surface temperature, surface winds, sea ice, and land surface characteristics. These should be assessed from the standpoint of providing consistent quantification of key biogeophysical variables that are bound by mass and energy balance constraints.
The development and application of integrated multi-instrument (infrared, passive microwave, radar) analysis methods would enhance the value of products depicting such variables as surface temperature, surface albedo, wetland extent, and indices of melt occurrence from older satellite data sets and in preparation for NPP and NPOESS missions.
If high-latitude terrestrial products are successfully validated, NASA can greatly facilitate studies of global polar hydrologic interactions by developing global data sets of such land surface variables as vegetation distribution, soil moisture, evapotranspiration, and precipitation.
Many measurements are needed to model high-latitude CO2 and CH4 fluxes and to assess the forces that affect these variables. The needed quantities include land surface temperature, precipitation, evapotranspiration, photosynthetically active radiation, soil moisture, biomass, leaf area index, and land surface characteristics. Although some of the variables are already being measured by remote-sensing techniques, these data sets need to be evaluated to determine whether they are adequate for high-latitude applications. Many of the satellite-derived data sets required for estimating terrestrial CO2 and CH4 fluxes are still not available in the scientific community. It is recognized that the data stream activity related to the Earth Observing Systems TERRA and AQUA missions has been designed to provide data sets in a timely manner to the scientific community; however, as these missions have a global focus, NASA needs to support evaluations of the adequacy of these products for high-latitude applications.
A particular challenge surrounds the validation of spaceborne estimates of precipitation, when the accurate measurement of precipitation in polar regions using traditional gound-based monitoring remains so problematical. Gage undercatch and interpolation bias are well-known problems in many parts of the globe, but especially in high latitudes. A dedicated effort by NASA to help reconcile the land and space-based estimates
would constitute an important enhancement to its current activities in this arena.
The committee was struck by the multitude of sea ice data sets, especially for ice concentration and, to a lesser extent, ice motion. In the case of sea ice concentration, NASA should help to clear up the confusion about similar data sets by funding independent studies to (1) document the strengths and weaknesses of different algorithms and (2) facilitate the construction of longer time series by the merger of individual sensor data sets. Further work is also recommended on the determination of summer ice concentration. Comparative assessments of sea ice motion data sets would also be prudent as soon as the record lengths of the various data sets (e.g., from the AMSR) are sufficient to permit meaningful comparisons.
Surface temperatures over terrestrial and ocean regions are needed for a variety of applications, including turbulent flux determinations, freeze and thaw mapping, and controls on trace gas fluxes. NASA can take steps to enhance surface temperature data sets by supporting programs that measure in situ SST (particularly surface temperature) in high latitudes. Regional calibrations in high latitudes should also be encouraged as part of global temperature products. This has been proposed but not implemented for Arctic waters as part of the AVHRR polar Pathfinder Project. This Pathfinder Project has also created a preliminary integrated surface temperature data set that includes ocean, ice, and land, but only for the polar regions. This type of activity should be strongly encouraged on a global basis.
The terrestrial hydrology, ocean modeling, and ice sheet research communities have all expressed the need for high-latitude precipitation data sets. Polar rainfall and snowfall, which are not measured with adequate spatial and temporal coverage, represent a major missing piece in the climate puzzle. NASA would make a major contribution if it could find a way to remotely sense this quantity in polar regions.
Another major contribution would be the measurement of ocean surface salinity at high latitudes. This quantity links the ocean, air, and land surfaces via the hydrologic cycle. Variations in this quantity are a key indicator of climate variability and ocean circulation. Much of the justification for upcoming international satellite missions to measure ocean salinity concerns high latitude processes. Unfortunately, the measurement precision by satellites in these regions is expected to be inadequate for climate studies. Interim aircraft-based studies might help in sensor and algorithm development.
The five-year changes measured by aircraft laser altimeter in Greenland show extensive thinning at the ice sheet margin; more work on ice flow and melt processes is necessary to understand the origins and implications of this signal. NASA has demonstrated capabilities in
Greenland for measuring ice sheet surface elevation and thickness. This ability, if combined with present efforts at detecting the distribution of shallow layers to understand the spatial variations in accumulation, could produce an unmatched tool for completing the ice thickness map of Antarctica. Tying measurements from an accumulation radar to shallow core sites where background rates and variability have been determined in both Greenland and Antarctica would greatly enhance efforts to determine appropriate forcing for ice sheet models.
Altimetry offers the most promising approach to sea ice thickness mapping for change detection; this approach is the subject of an upcoming European Space Agency mission. The utility of ICESAT data for this application needs to be investigated, along with integrated approaches that use altimetric and ice motion time series, such as AVHRR and RGPS methods in the still exploratory stages. Validation and error analysis for these products should be pursued in an integrated manner as well.
NASA has no plans for a polar radar altimeter to measure ocean surface elevation beyond the latitudinal range of Topex-Poseidon. NASA should consider participation in future high-latitude radar altimetry programs, perhaps in conjunction with the satellite agencies of other nations.
Finally, the Earth Science Enterprise research activities of the polar community will be limited by the lack of appropriate SAR coverage, which may require a U.S. SAR. Many of the recent NASA-funded advances in cryospheric research have come from interferometric analysis of time series of SAR data. The processes that are studied are time-variable and so require relatively frequent repeat observations. Both sea ice and ice sheet research will suffer without access to SAR data acquired at appropriate intervals. The absence of a U.S. SAR satellite and/or MOUs with entities flying SARs is of great concern to the polar ocean and ice sheet communities.