next up previous
Next: Sea-Level Rise Up: 1 vs. 2 x CO Previous: Accumulation

Runoff

The total runoff originating from the Greenland and Antarctic ice sheets for both current and doubled carbon dioxide conditions are sumarized in Table 3.


 
Table: Runoff from the Greenland and Antarctic ice sheets for 1 and 2 x CO2 conditions. Units are 1012 kg a-1. Source for the observed is $ \dag$: Houghton et al. (1996).
  MIT ECHAM Observations
  Snowpack PDD Linear Snowpack PDD Linear  
Greenland 1 x CO2 162 172 299 122 353 568 237$\scriptstyle \dag$
Greenland 2 x CO2 306 291 448 295 515 832  
Greenland - Change 148 119 149 173 162 264  
Antarctica 1 x CO2 0 63 620 0 18 122  
Antarctica 2 x CO2 29 146 1029 0 10 112  
Antarctica - Change 29 83 409 0 -8 -10  

The estimates produced by the MIT/snowpack and the MIT/degree-day models for Greenland for the current climate are similar and 25-30% lower than the value of 237 . 1012 kg a-1 derived by Reeh (Houghton et al. (1996)) from measurements. The MIT/snowpack model estimates total melting at 176 . 1012 kg a-1, $ \sim$ 20% of the total melt- and rainwater ( 35 . 1012 kg a-1) is predicted to refreeze in-situ. The ablation at individual stations and the extent of the melt zone predicted by these model combinations are generally in excellent agreement with observations in the Southern two-thirds of the ice sheet, the extent and intensity of melting along the Northern coast is however underestimated. The linear model overestimates the source area of runoff. The three models do however predict similar increases in runoff over the 21st century, +119 - +149 . 1012 kg a-1. The ablation region is shown in Fig.1 for both single and double CO2 conditions and for the MIT/snowpack model combination. An increase in both the intensity and extent of the source area of runoff can be observed.


  
Figure: Runoff in m year-1 predicted by the MIT/snowpack model combination. Left column: 1 x CO2, Right column: 2 x CO2. Dotted lines are the 1000 m. topographic height contours.
\begin{figure}
\begin{center}
\epsfig{file=/u/u0/vero/SNOW/DOCS/FIGS/G1cND_run.e...
...CS/FIGS/G2cND_run.eps, width = 6 cm, height = 6 cm}\\
\end{center}
\end{figure}

The discrepancy bewteen the results obtained with the three melt models is much larger for the estimates produced with the ECHAM model input for Greenland. The snowpack model underestimates melting in the Southern half of the ice sheet because the climate model's summer temperatures are lower than observed, and the temperature dependence of the albedo parameterization does not allow the energy balance to become sufficiently positive to generate large amounts of meltwater. It does compensate by capturing melting and runoff in the Northern third of the ice sheet. The amount of refreezing taking place is however larger than was the case for the MIT model, 49 . 1012 kg a-1 or $ \sim$ 40% or the melt- and rainwater input and the aggregate estimate of runoff is, at 122 . 1012 kg a-1, lower than the MIT model's. The Degree-Day and in particular the linear model produce very intense melting in the Northern half of the ice sheet. The difference between 1 and 2 x CO2 conditions is shown in Fig.2 for the ECHAM/snowpack combination. The larger increase in runoff associated with the ECHAM model is due to a more rapid increase in summertime temperatures than in the MIT model, up to 5oC over the central portion of the ice sheet and $ \sim$ 1.7oC in the coastal areas source of runoff. The MIT model predicts a warming of the average summer temperature of 1.2oC between the current climate and the time of CO2 doubling with the largest warming $ \sim$ 1.5oC taking place in the Southern third of the ice sheet where most of the melting takes place. This distribution of the temperature change in that model is closely linked to modifications in the sea-ice distribution and associated albedo changes.


  
Figure: Runoff in m year-1 predicted by the ECHAM/snowpack model combination. Left column: 1 x CO2, Right column: 2 x CO2. Dotted lines are the 1000 m. topographic height contours.
\begin{figure}
\begin{center}
\epsfig{file=/u/u0/vero/SNOW/DOCS/FIGS/G1cID_run.e...
...CS/FIGS/G2cID_run.eps, width = 6 cm, height = 6 cm}\\
\end{center}
\end{figure}

The substantial increase in runoff which is predicted by the melt models for rather small changes in summer temperatures (+1.5 - 1.7oC) are a clear illustration of the high sensitivity of the mass balance of Greenland to changes in climate.

The differences in response of the melt models can in part be traced back to the runoff parameterizations. The runoff predicted by the linear model is a linear function of temperature. This is also the case for the degree-day model after the initial fraction of meltwater is refrozen. The albedo parameterization built into the snowpack model will however lead to a non-linear response to changes in air temperature: Once temperatures pass the melting point, the albedo drops very rapidly, as a cubic function of temperature. After the snow is melted away however, the albedo stabilizes at the constant value chosen for the reflectivity of ice and the amount of melting and runoff taking place depends entirely on the net surface energy balance.

The snowpack model, whether forced with the MIT or the ECHAM climate data, does not predict any runoff originating from Antarctica for the current climate and only minimal runoff at the time of CO2 doubling (Table 3). The input of liquid water in the form of rain ( $ \sim$ 250 . 1012 and $ \sim$ 35 . 1012 kg a-1 for the MIT and ECHAM models respectively) or meltwater ( $ \sim$ 25 . 1012 resp. $ \sim$ 2.5 . 1012 kg a-1) refreezes entirely in-situ. The temperature based methods predict small to moderate amounts of runoff, all of which is taking place on the Antarctic Peninsula. Although the source area of runoff is not inconsistent with the observed extent of the melt zone derived from satellite microwave remote sensing by Zwally and Fiegles (1994), the linear model's prediction of 620 . 1012 kg a-1 of ablation for the current climate would have led to a rapid depletion of ice in that region. The warming of air temperatures predicted by the ECHAM model is entirely concentrated in the fall, winter and spring seasons, leaving summertime temperatures, and therefore melting, unchanged between the current climate and the time of CO2 doubling.


next up previous
Next: Sea-Level Rise Up: 1 vs. 2 x CO Previous: Accumulation
Veronique Bugnion
1999-10-19