Apr 8, 2009
Monitoring Greenland's melting
The ten warmest years since 1880 have all taken place within the 12-year period of 1997–2008, according to the NASA Goddard Institute for Space Studies (GISS) surface temperature analysis. The Arctic has been subject to exceptionally warm conditions and is showing an extraordinary response to increasing temperatures. The changes in polar ice have the potential to profoundly affect Earth's climate; in 2007, sea-ice extent reached a historical minimum, as a consequence of warm and clear sky conditions. And during the 2007 and 2008 melting seasons new records for melt extent and duration since 1979 were recorded on the Greenland ice sheet.
Greenland plays a key role in the Arctic system. Climatologically, it is a centre of cooling because it reflects most of the incoming solar radiation and because its great height – more than 3000 m – acts as a topographic barrier and affects Arctic and sub-Arctic atmospheric circulation. Hydrologically, Greenland represents a large source of freshwater. If the whole ice sheet were to melt it would contribute up to about 7 m to sea-level rise. Recent studies report thinning at the margins of the Greenland ice sheet, an increased discharge from outlet glaciers and a significant increase in surface melt.
So what's the best way to measure these phenomena? Ground-based continuous monitoring is limited in Greenland because of the harsh weather conditions and its large territory – around 2.2 million sq. km, of which about 1.7 million sq. km is covered by the ice sheet. Weather stations and scientific camps are very few, and are mostly located along the coast or at strategic locations. So to gain a comprehensive view of phenomena occurring on the ice sheet, satellites must be used. These offer a powerful tool for monitoring geophysical processes over large areas and with frequent measurements.
Most people are familiar with visible satellite data, which provide images that are often suggestive and visually powerful. But sensors collecting data in the visible region are limited by cloud cover and by their requirement for illumination by the Sun. Plus, visible data can only provide information on near-surface processes.
Alternatively, microwave data are affected by sub-surface processes. So it's capable of detecting phenomena such as melting below the top layer of the snowpack. What's more, collection of microwave data does not depend on the presence of solar illumination, which is crucial for northern areas where the Sun does not shine for long periods of the year or is very low on the horizon. In addition, microwave data are only weakly affected by the presence of clouds and so can provide information under all weather conditions.
There are two categories of microwave sensors: active and passive. Active sensors provide their own source of microwave radiation to illuminate the target (e.g. RADAR), while passive sensors record the electromagnetic radiation naturally emitted by the targets (e.g. radiometers). The quantity measured by radiometers is called brightness temperature and is related to the thermometric temperature of an area through its emissivity. Microwave brightness temperatures have been measured for the past 30 years by sensors mounted on different satellites. These are: the Scanning Microwave Multichannel Radiometer (SMMR, 1978–1987); the Special Sensor Microwave/Imager (SSM/I, 1987 to date) flying on board the Defense Meteorological Satellite Program (DMSP) series of satellites; and the Advanced Microwave Scanning Radiometer for Earth (AMSR-E, 2002 to date) flying on the AQUA satellite. These data have been used to study melting over Greenland for the past 30 years.
The sensitivity of microwaves to liquid water is the key ingredient for detecting melting snow. Even if it's present only in small amounts, liquid water suddenly and abruptly increases the relatively low microwave brightness temperature of dry snow by increasing absorption and reducing scattering.
Melting in 2008
What have these measurements told us? My colleagues and I used space-borne passive microwave observations to determine the trends of standardised snowmelt extent – defined as the area subject to melting – and melting index (the number of melting days multiplied by the area subject to melting) for the period 1979–2008 (see Figure 1).
In terms of absolute values, the 2008 melt extent was slightly above the 1979–2007 average of ~ 0.4 x 106 sq. km, with a value of ~ 0.47 x 106 sq. km. The trend in melt extent for the period 1979–2008 is an increase of about 16,000 sq. km/year, slightly less than the total surface area of Spain. The melting index in 2008 was not exceptionally high over the whole Greenland ice sheet. Indeed it was much lower than in 2007, when the melting index was 2.7 times the standard deviation above the average (Figure 2).
Figure 1 also shows the 2008 anomaly of the number of melting days, defined as the number of melting days in 2008 minus the 1979–2007 average number of melting days. According to passive microwave data, extreme snowmelt occurred during summer 2008 over the northern part of the Greenland ice sheet, setting new records for the number of melting days observed in many northern areas. In 2008, melting in northern Greenland lasted up to 18 days longer than previous maximum values.
Areas where extreme snowmelt occurred also witnessed other noticeable events: a 29 sq. km area of the Petermann Glacier in northern Greenland (80°N, 60°W) broke away between 10 July and 24 July 2008 (Figure 3); and by July 2008 what had once been a massive ice fringe along the northern Ellesmere coast had been reduced to five isolated ice shelves. On 22 July 2008 a wave of ice shelf disintegration began and by late August these ice shelves had lost a total of 214 sq. km (Figure 4).
Positive snowmelt anomalies in 2008 were also detected along Greenland's west coast, with melting over this area lasting up to 5–10 days longer than average. The 2008 snowmelt anomaly map shows that snowmelt over the whole Greenland ice sheet in 2008 was not significant at high elevations, contrary to what happened in 2007 (Figure 2).
Adding in surface temperature
To complement satellite observations with ground measurements, surface data from automatic weather stations of the World Meteorological Organisation (WMO, http://www.ncdc.noaa.gov) were analysed. A total of 93 stations were originally selected for air temperature analysis. From these, nine stations were chosen as they had collected data between 1979 and 2008, overlapping with the satellite period. Names and locations of the selected stations are reported in the map in Figure 1. A more detailed analysis of the data is currently under way to confirm these results and to evaluate the consistency of the data set for the 1979–2008 period. In the meantime, here are the results from a first analysis.
For each station, the June-July-August (JJA) maximum temperature anomaly was examined. The values of the cumulative anomalies are plotted in Figure 5. The long-term trend (denoted as TR) for the period 1979–2008, the p-values (for the statistical significance, denoted as P) and the values of correlation coefficient (denoted as R) are reported in Table 1.
The analysis indicates that five of the nine selected stations (Pittufik, Nuuk, Narsarsuaq, Danmarkshavn and Prins Christian Sund) set a new air/surface JJA maximum temperature anomaly record in 2008, with values between 2.5 °C and 3 °C. For the remaining four stations, the 2008 maximum JJA averaged surface/air temperature ranked fourth highest since 1979 at Aasiaat and Qaqotorq, second at Tasiilaq and sixth at Kangerlussuaq. As shown in the map, most of the available stations are located along the coast while the melting anomalies occurred over internal areas. However, the maximum temperature anomaly trend derived from ground measurements for the period 1979–2008 can explain about 45% of the melt extent and index trends.
More than just temperature
What other factors might then explain the trends? Besides surface temperature, other mechanisms play a fundamental role in melting processes.
Among these, albedo-feedback plays a key role. When snow is fresh, its albedo is high and it reflects about 80% of incoming solar radiation. As snow gets older, its grain size increases and, consequently, albedo decreases. But time is not the only factor that ages snow. Melting can also accelerate its aging. As snow melts its grains are bound to each other with the liquid water acting as a "glue" between the particles. When the liquid water refreezes, the particles remain bound so that a major effect of melting-refreezing cycles is an increase in the average grain size. As mentioned earlier, this increases the amount of solar radiation absorbed which, in turn, accelerates the aging process and further increases the absorption of solar radiation. It's a vicious circle. During particularly warm summers snow would get old sooner than summers when temperatures are not high, and this would increase its capacity to absorb solar radiation, which in turn will further increase ageing and so on.
Because ice has a much lower albedo than snow (reflecting only about 40% of the incoming solar radiation), once the upper layers of the seasonal snow are melted away and bare ice is exposed, even more solar radiation will be absorbed. So another component that will affect melting and the production of liquid water is the amount of seasonal precipitation. Clouds are also driving factors as they affect both longwave and shortwave radiation.
To find out more, remote sensing data (a measurement tool) can and should be used in conjunction with outputs from regional surface energy balance models (modeling tools). That should significantly advance our ability to map, quantify, and assess melting over the Greenland ice sheet. Along these lines, there has been limited work to date in integrating remote sensing techniques and ground data (measurement tools) with models. Each of these components provides different information related to surface mass balance. Merging these disparate sources of information in a consistent way is currently under way by researchers at the City College of New York in collaboration with the University of Liege, Belgium. This forthcoming summer, fieldwork activities aimed at collecting ground information for validating satellite results will take place. During the first weeks of July 2009, the team will collect data using instruments similar to those mounted on the satellites together with snow and ice properties to improve satellite-based retrieval techniques.
About the author
Marco Tedesco is Assistant Professor at the City College of New York, City University of New York. He is also affiliated with the NASA Goddard Space Flight Center, US, and is a fellow of the University of Maryland, Baltimore County, US.