Weddell ice pack
Sensitivity studies on a model of the Variables examined In model sensitivity studies and values used Weddell ice pack Input variable Value in standard case Sensitivity-study range 0.95-1.00 0.97 Ice emissivity 0.95-1.00 Snow emissivity 0.99 0.95-1.00 Water emissivity 0.97 Solar radiation formulation of Zillman 0.90 x standard-1.10 (1972); varies spa- x standard tially and temporally Drag coefficient 0.0024 0.001-0.004
CLAIRE PARKINSON
Goddard Laboratory for Atmospheric Sciences National Aeronautics and Space Administration Goddard Space Flight Center Greenbelt, Maryland 20771
A sequence of sensitivity studies has been run for a numerical model of the growth and decay of sea ice in the Weddell Sea and surrounding regions. The model formulations follow Parkinson and Washington (1979), and the model grid is a subset of the Parkinson and Washington grid (see figure 1). Each model run simulates 14 months, beginning on January 1. Outputs include (1) contour maps of ice thickness and concentration at the midpoint of each month, and (2) plots of the following variables as a function of time: ice thickness and ice concentration at grid points (5,7) and (9,4); ice thickness and concentration averaged over the grid squares within the Weddell Sea (figure 1) that contain some ice (labeled "ice-laden"); the area of ice-laden waters; and the area of actual ice coverage. For many of the model inputs, a sufficiently large change in the input value causes a significant change in the model results. Of more relevance is whether a significant change in results can be produced by a "reasonable" change in the input. For the present studies, a change is considered "reasonable" if it falls within the variable's current range of uncertainty. For other studies, "reasonable" changes might be much greater than those used here (for instance, if the purpose of the study were to calculate the sea-ice response to a predicted increase in atmospheric carbon dioxide, or if it were to determine sea-ice extents under such changed boundary conditions as might have existed during a past major glaciation). The table lists the input variables examined in the current studies and the range
340'E
17
O'E 20'E
320°E 13
300°E
40'E
1
5
9
13
17
21 23
Figure 1. Grid structure of the model and geographical location. The dashed lines at 200 E and 3000 E indicate the boundaries of the region defined for the analysis as the Weddell Sea. 94
0.40-0.60 0.50 Ice albedo 0.05-0.15 0.10 Water albedo 0.60-0.90 0.75 Snow albedo Turbulent exchange 0.00175 0.001-0.003 coefficients Snowfall rate 0.3 cm per month Feb- 0-30 cm per month ruary—November Air temperatures mean monthly climato- standard ± 5 Kelvin logical fields, from Taljaard et at. (1969) Winds mean monthly climato- a variety of alternate logical fields, from fields Taljaard et at. (1969) Ocean heat flux 25 watts/sq meter 0-40 watts/sq meter
used for each. For each set of model runs, only the variable in question was altered; the other variables were assigned "standard case" values. The sensitivity studies establish that the model is almost entirely insensitive to "reasonable" changes in long-wave emissivities of ice, water, and snow. The effect of increasing emissivities from 0.95 to 1.00 was imperceptible until month 12, when it increased the ice amount slightly. This increase suggests that increased absorption of long-wave radiation from the atmosphere, due to higher emissivities, is outweighed by increased emission from the surface to the atmosphere. The model is somewhat more sensitive to changes in solar flux, short-wave albedos for ice and water, and drag coefficient. Increases in solar flux and/or decreases in ice albedo naturally result in a thinner ice cover, whereas decreases in water albedo have a stronger effect on lowering ice concentrations. The latter effect derives from increased absorption of solar radiation in the ocean and the consequent delay in ice formation and increase in lateral ice melt. The influence of ice albedo is strong early in the simulation, leading to central ice thicknesses in mid-February that are 30 centimeters lower for an albedo of 0.4 than for an albedo of 0.6. However, this effect is much reduced after March, when ice begins to acquire a snow cover. Since the drag coefficient (CD enters the model calculations only through ice velocities, the effect of changing its value is to redistribute the ice; hence, its effect is not uniform over the grid. The model is moderately sensitive to changes in snow albedo, turbulent heat exchange coefficients, and amount of snowfall. Model sensitivity to snow albedo is nonexistent in the first 2 months due to the absence of a snow cover, weak during the fall and winter when there is not much incident solar radiation, but strong during the subsequent melt season. In mid-September, ice extents are close to identical for snow albedos of 0.6 and 0. 9, but by the middle of the following January total ice coverage in the 0.6-albedo case is only half the ice coverage in the 0.9 case. ANTARCTIC JOURNAL
(A) ICE THICKNESS AT (9, 4) 3.0 2.5 E Cl) Ch LU 2
01.5 I Iuj 1.0 0 0.5
0.0
(B)
J FMAMJ JASON DJ F
ICE CONCENTRATION AT (9, 4) 100
0 NO z75 0
Iz50 Ui
0 2
0 0 25 LU
nesses at grid point (5,7) are about half their values for turbulent coefficients of 0.003. The additional insulation of the ice from the cold winter atmosphere provided by a thicker snow cover means that higher snowfall rates will decrease ice thicknesses. Numerically, the thicker snow cover increases the calculated snow-ice interface temperature, in turn reducing the temperature contrast between the bottom and top of the ice and thereby reducing the upward conductive flux through the ice. The decreased flux results in a lower growth rate of ice in winter and a higher decay rate in spring, as there is more energy available for melt at the ice/water interface. The model proved highly sensitive to air temperatures, winds, and ocean heat flux (figure 2). Thus, it is especially important to insert proper values for these variables. The large effect produced by altering the wind fields is of particular concern since these fields can change radically over short timespans. For two of the model runs, the mean climatological winds were replaced by 1974 winds from the Australian Bureau of Meteorology data sets. For one run, the winds were inserted as monthly averages and interpolated for each time step; for the other, they were inserted with 12-hour temporal resolution. The twice-daily winds produced a much stronger summer/winter contrast in the simulated ice cover, with almost no ice remaining in March but with at least a slight ice cover in almost every grid square by June. The effect of a variety of wind fields on the modeled Weddell polynya will be described elsewhere (Parkinson in preparation), as will the effect of atmospheric temperature increases on the entire southern ocean (Parkhurst and Bindschadler in preparation). The model's sensitivity to ocean heat flux has been detailed by Parkinson and Good (1982). This work was supported by NASA's Oceanic Processes Branch in the Environmental Observations Division. I acknowledge with thanks the assistance of Michael R. Good of Computer Sciences Corporation.
0
0
J F M A M J J A SON D J F MONTH
References Parkinson, C. L. In preparation. On the development and cause of the Weddell polynya in a sea ice simulation.
Parkinson, C. L., and Bindschadler, R. A. In preparation. Response of antarctic sea ice to uniform atmospheric temperature increases. Figure 2. Sample results illustrating the sensitivity of the time sequences of Ice thicknesses (A) and ice concentrations (B) at gridsquare (9,4) to values of the ocean heat flux ranging from 0 to 40 watts per square meter.
Since the sensible and latent heats are calculated as positive upward, the effect of decreasing the turbulent exchange coefficients is to decrease the transfer of heat from the surface to the atmosphere and thereby to decrease ice thicknesses and concentrations. With turbulent heat coefficients of 0.001, ice thick-
1982 REVIEW
Parkinson, C. L., and Good, M. R. 1982. Sensitivity of a climatologicallydriven sea ice model to the ocean heat flux (NASA Technical Memorandum 83877). Greenbelt, Md.: Goddard Space Flight Center. Parkinson, C. L., and Washington, W. M. 1979. A large-scale numerical model of sea ice. Journal of Geophysical Research, 84, 311-337. Taljaard, J . J. , van Loon, H., Crutcher, H. L., and Jenne, R. L. 1969.
Climate of the upper air, I, Southern Hemisphere, Vol. 1, Temperatures, dew points and heights at selected pressure levels (NAvAIR Report 50-1C-55.
Washington, D.C.: U.S. Naval Weather Service Command.
Ziliman, J. W. 1972. A study of some aspects of the radiation and heat budgets of the Southern Hemisphere oceans (Meteorological studies, Vol. 26). Canber-
ra, Australia: Bureau of Meteorology, Department of the Interior.
95
CLAIRE PARKINSON
Goddard Laboratory for Atmospheric Sciences National Aeronautics and Space Administration Goddard Space Flight Center Greenbelt, Maryland 20771
A sequence of sensitivity studies has been run for a numerical model of the growth and decay of sea ice in the Weddell Sea and surrounding regions. The model formulations follow Parkinson and Washington (1979), and the model grid is a subset of the Parkinson and Washington grid (see figure 1). Each model run simulates 14 months, beginning on January 1. Outputs include (1) contour maps of ice thickness and concentration at the midpoint of each month, and (2) plots of the following variables as a function of time: ice thickness and ice concentration at grid points (5,7) and (9,4); ice thickness and concentration averaged over the grid squares within the Weddell Sea (figure 1) that contain some ice (labeled "ice-laden"); the area of ice-laden waters; and the area of actual ice coverage. For many of the model inputs, a sufficiently large change in the input value causes a significant change in the model results. Of more relevance is whether a significant change in results can be produced by a "reasonable" change in the input. For the present studies, a change is considered "reasonable" if it falls within the variable's current range of uncertainty. For other studies, "reasonable" changes might be much greater than those used here (for instance, if the purpose of the study were to calculate the sea-ice response to a predicted increase in atmospheric carbon dioxide, or if it were to determine sea-ice extents under such changed boundary conditions as might have existed during a past major glaciation). The table lists the input variables examined in the current studies and the range
340'E
17
O'E 20'E
320°E 13
300°E
40'E
1
5
9
13
17
21 23
Figure 1. Grid structure of the model and geographical location. The dashed lines at 200 E and 3000 E indicate the boundaries of the region defined for the analysis as the Weddell Sea. 94
0.40-0.60 0.50 Ice albedo 0.05-0.15 0.10 Water albedo 0.60-0.90 0.75 Snow albedo Turbulent exchange 0.00175 0.001-0.003 coefficients Snowfall rate 0.3 cm per month Feb- 0-30 cm per month ruary—November Air temperatures mean monthly climato- standard ± 5 Kelvin logical fields, from Taljaard et at. (1969) Winds mean monthly climato- a variety of alternate logical fields, from fields Taljaard et at. (1969) Ocean heat flux 25 watts/sq meter 0-40 watts/sq meter
used for each. For each set of model runs, only the variable in question was altered; the other variables were assigned "standard case" values. The sensitivity studies establish that the model is almost entirely insensitive to "reasonable" changes in long-wave emissivities of ice, water, and snow. The effect of increasing emissivities from 0.95 to 1.00 was imperceptible until month 12, when it increased the ice amount slightly. This increase suggests that increased absorption of long-wave radiation from the atmosphere, due to higher emissivities, is outweighed by increased emission from the surface to the atmosphere. The model is somewhat more sensitive to changes in solar flux, short-wave albedos for ice and water, and drag coefficient. Increases in solar flux and/or decreases in ice albedo naturally result in a thinner ice cover, whereas decreases in water albedo have a stronger effect on lowering ice concentrations. The latter effect derives from increased absorption of solar radiation in the ocean and the consequent delay in ice formation and increase in lateral ice melt. The influence of ice albedo is strong early in the simulation, leading to central ice thicknesses in mid-February that are 30 centimeters lower for an albedo of 0.4 than for an albedo of 0.6. However, this effect is much reduced after March, when ice begins to acquire a snow cover. Since the drag coefficient (CD enters the model calculations only through ice velocities, the effect of changing its value is to redistribute the ice; hence, its effect is not uniform over the grid. The model is moderately sensitive to changes in snow albedo, turbulent heat exchange coefficients, and amount of snowfall. Model sensitivity to snow albedo is nonexistent in the first 2 months due to the absence of a snow cover, weak during the fall and winter when there is not much incident solar radiation, but strong during the subsequent melt season. In mid-September, ice extents are close to identical for snow albedos of 0.6 and 0. 9, but by the middle of the following January total ice coverage in the 0.6-albedo case is only half the ice coverage in the 0.9 case. ANTARCTIC JOURNAL
(A) ICE THICKNESS AT (9, 4) 3.0 2.5 E Cl) Ch LU 2
01.5 I Iuj 1.0 0 0.5
0.0
(B)
J FMAMJ JASON DJ F
ICE CONCENTRATION AT (9, 4) 100
0 NO z75 0
Iz50 Ui
0 2
0 0 25 LU
nesses at grid point (5,7) are about half their values for turbulent coefficients of 0.003. The additional insulation of the ice from the cold winter atmosphere provided by a thicker snow cover means that higher snowfall rates will decrease ice thicknesses. Numerically, the thicker snow cover increases the calculated snow-ice interface temperature, in turn reducing the temperature contrast between the bottom and top of the ice and thereby reducing the upward conductive flux through the ice. The decreased flux results in a lower growth rate of ice in winter and a higher decay rate in spring, as there is more energy available for melt at the ice/water interface. The model proved highly sensitive to air temperatures, winds, and ocean heat flux (figure 2). Thus, it is especially important to insert proper values for these variables. The large effect produced by altering the wind fields is of particular concern since these fields can change radically over short timespans. For two of the model runs, the mean climatological winds were replaced by 1974 winds from the Australian Bureau of Meteorology data sets. For one run, the winds were inserted as monthly averages and interpolated for each time step; for the other, they were inserted with 12-hour temporal resolution. The twice-daily winds produced a much stronger summer/winter contrast in the simulated ice cover, with almost no ice remaining in March but with at least a slight ice cover in almost every grid square by June. The effect of a variety of wind fields on the modeled Weddell polynya will be described elsewhere (Parkinson in preparation), as will the effect of atmospheric temperature increases on the entire southern ocean (Parkhurst and Bindschadler in preparation). The model's sensitivity to ocean heat flux has been detailed by Parkinson and Good (1982). This work was supported by NASA's Oceanic Processes Branch in the Environmental Observations Division. I acknowledge with thanks the assistance of Michael R. Good of Computer Sciences Corporation.
0
0
J F M A M J J A SON D J F MONTH
References Parkinson, C. L. In preparation. On the development and cause of the Weddell polynya in a sea ice simulation.
Parkinson, C. L., and Bindschadler, R. A. In preparation. Response of antarctic sea ice to uniform atmospheric temperature increases. Figure 2. Sample results illustrating the sensitivity of the time sequences of Ice thicknesses (A) and ice concentrations (B) at gridsquare (9,4) to values of the ocean heat flux ranging from 0 to 40 watts per square meter.
Since the sensible and latent heats are calculated as positive upward, the effect of decreasing the turbulent exchange coefficients is to decrease the transfer of heat from the surface to the atmosphere and thereby to decrease ice thicknesses and concentrations. With turbulent heat coefficients of 0.001, ice thick-
1982 REVIEW
Parkinson, C. L., and Good, M. R. 1982. Sensitivity of a climatologicallydriven sea ice model to the ocean heat flux (NASA Technical Memorandum 83877). Greenbelt, Md.: Goddard Space Flight Center. Parkinson, C. L., and Washington, W. M. 1979. A large-scale numerical model of sea ice. Journal of Geophysical Research, 84, 311-337. Taljaard, J . J. , van Loon, H., Crutcher, H. L., and Jenne, R. L. 1969.
Climate of the upper air, I, Southern Hemisphere, Vol. 1, Temperatures, dew points and heights at selected pressure levels (NAvAIR Report 50-1C-55.
Washington, D.C.: U.S. Naval Weather Service Command.
Ziliman, J. W. 1972. A study of some aspects of the radiation and heat budgets of the Southern Hemisphere oceans (Meteorological studies, Vol. 26). Canber-
ra, Australia: Bureau of Meteorology, Department of the Interior.
95
