Temperature measurements at various depths in Polar shelf ice, glaciers and ice caps need not be very intensive to be of value, but several precautions must be taken in their gathering, and certain allied data must be obtained concurrently, if the sub-surface temperature distribution is to be not only described but fully explained.
The following paragraphs have been written to amplify one point in the “Discussion on Glaciological Research in the Antarctic,” held by the Society on 14 May 1947. They are based on intensive study and attempts at interpretation of the original observations of sub-surface temperature made at “Little America III” by Wade,Reference Wade
1
,
Reference Fleming and Alton
2
and of reports of others who have measured sub-surface temperatures in snow and ice, chiefly Wegener,Reference Koch and Wegener
3
Sorge,Reference Sorge
4
JülgReference Jülg
5
and Sverdrup.Reference Sverdrup
6
Suggestions for conduct of a program of temperature measurements are grouped under two headings: DIRECT, those pertaining to the measurements themselves; and COROLLARY, those pertaining to other measurements which must be made at the same time if the program is to have maximum value.
Direct
Method
Electrical means of determining temperatures at various depths below the surface of the firn are indubitably superior to readings of mercury or alcohol thermometers, either suspended in individual holes (WegenerReference Koch and Wegener
3
) or inserted into holes in the side of a shaft (SorgeReference Sorge
4
). These electrical elements may be dropped down to the required depths in holes drilled in the snow or ice for the purpose, and the holes filled in immediately; the elements are not recovered at the end of the program.
Instruments
Resistance thermometers (used by WadeReference Wade
7
) may be more durable than thermocouples (used by SverdrupReference Sverdrup
6
and Hughes and SeligmanReference Hughes and Seligman
8
) but may have larger errors due to the larger mass of the thermal element. Three wires are required in the leads of resistance thermometers, as opposed to two for thermocouples; this difference may be important when very long leads are planned. In formulating any future program of temperature measurements, the relative advantages of resistance thermometers and thermocouples must be carefully weighed.
Accuracy
Temperatures should be accurate to the nearest tenth of a degree centigrade, at all depths. Greater accuracy is probably not needed for over-all usefulness. The annual variation below 20 m. is so slight that it will not be shown clearly unless readings are truly accurate to the nearest hundredth of a degree centigrade, an accuracy which will probably be unattainable.
Location
The holes in which the thermal elements are dropped should be grouped very close together, and should be at least 50 m. from the nearest building or other drift-causing obstruction or source of heat. (In discussing their observations, both WegenerReference Koch and Wegener
3
and JülgReference Jülg
5
were forced to study extensively the possible effects of heat from buildings on temperature measurements underneath them.)
Depths
Levels at which temperature measurements are to be made should not be spaced uniformly, but rather according to an exponential law. For practical purposes a geometric progression may be used, such as 
, 1, 2, 4, 8, 16, 32 and 64 m. Readings at these levels will provide information adequate in detail for any analysis of the thermal structure and regime of a relatively homogeneous sheet of firn. Additional readings, for instance at 3 m. or 10 m., would not add materially to the information obtained; the suggested levels are sufficiently close together for satisfactory interpolation.
Accretion
The exact depth below the surface must be known, within r per cent. for each thermal element. Since the surface will change from day to day and probably will rise during the year, the height of the actual snow surface above the original surface level must be measured frequently. A series of vertical graduated poles, placed around the area of the holes, will provide the required information. Such sticks were used for measuring accretion at “Little America,”Reference Fleming and Alton
2
but were not originally related to the sub-surface temperature measurements. Naturally, the side for observing both temperatures and accretion must be free of unnatural drift; depending on the general location the 50 m. suggested above may not be enough to avoid artificial drifts, especially from large buildings.
Frequency
Readings need be made no more than once daily in the uppermost 5 m., and only every 5 to 10 days at greater depths. However, at each observation a series of three or even five readings, at intervals of three minutes or so, should be taken for each level, to ensure accuracy. (Wade’s observations at first were made four times daily, later twice daily, and finally once daily.)
Duration
Measurements preferably should be continued for thirteen or fourteen months; observing a complete annual cycle with overlap will determine the presence of any year-to-year variation or secular change. However, much value can be obtained by careful measurements over a much shorter period. Even one day’s readings of temperature at the lower levels (32 and 64 m., or 30 and 50 m.) can offer invaluable information, provided the disturbing effects of the drilling have disappeared by the time readings are taken. In such a short-term investigation, the holes will be drilled, the elements dropped in, the holes filled up as well as possible and readings taken frequently until equilibrium is attained. Since the annual variation is almost undiscernable at 32 m. (it is only about half a degree centigrade, at 16 m. at both “Little America” and “Eismitte”), temperatures at these lower levels will indicate the permanent thermal gradient and thus the direction and amount of heat flow.
Corollary
Density
Any computation of heat flow in shelf ice or glaciers involves both the thermal gradient and the thermal conductivity, which depends on the density. Consequently, determinations of density should be made at the same levels as those at which temperatures are measured. Admittedly, this is far more difficult than drilling a hole and dropping into it a wire with a resistance thermometer or thermocouple on the end. Nevertheless, without density values, the temperature readings merely describe conditions and cannot be used to explain them.
Crevasses
In some cases crevasses will be penetrated by the holes into which the thermal elements are to be dropped. Since they will affect the temperature regime markedly, they should be avoided if possible. If measurements must be made in a crevassed area, the exact depths and sizes of all crevasses or other cavities must be recorded.
Weather
Ordinary weather observations are almost certain to be made at the same camp as the subsurface temperature readings, so they need not be mentioned in detail. Preferably, both air temperature and wind speed data should be available from two levels, such as 2 and 6 m. Barometric pressure, records of cloudiness and precipitation times and types may also be of value in interpreting the sub-surface temperatures; precipitation amounts probably cannot be obtained other than from the snow poles mentioned above.
Radiation
For a complete explanation of the temperatures observed in firn and ice, knowledge of the radiation balance is required. Intensity of solar radiation should be measured, and also the amount of outgoing (“nocturnal”) radiation.
Surface
The temperature of the actual snow surface is of immense interest, but there are almost insuperable difficulties in obtaining it. Because of the low density at the very surface, even a fine thermocouple cannot be expected to measure the snow temperature unaffected by radiation. Some sort of radiometric measurement of the snow surface temperature may be feasible.
Thickness
The total thickness of the firn-ice sheet being investigated should be determined, by seismic or other means.
General
Other variables enter into a complete analysis of the internal temperatures of firn and ice, depending on the particular circumstance. For floating shelf ice, the temperature, salinity and currents of the underlying water are important. For glaciers, grounded shelf ice and ice caps the rate of motion and the nature of the underlying rock are important.
, 1, 2, 4, 8, 16, 32 and 64 m., and readings (if repeated to ensure accuracy) need be taken only once daily down to 4 m. and weekly or so at lower levels. Measurements should be free of external influences such as heat from buildings or deep drifts. Only when all other necessary data are available and all precautions observed will it be possible to explain the thermal structure of the ice sheet as shown by the sub-surface measurements of temperature.
You have
Access
Temperature measurements at various depths in Polar shelf ice, glaciers and ice caps need not be very intensive to be of value, but several precautions must be taken in their gathering, and certain allied data must be obtained concurrently, if the sub-surface temperature distribution is to be not only described but fully explained.
The following paragraphs have been written to amplify one point in the “Discussion on Glaciological Research in the Antarctic,” held by the Society on 14 May 1947. They are based on intensive study and attempts at interpretation of the original observations of sub-surface temperature made at “Little America III” by Wade,Reference Wade 1 , Reference Fleming and Alton 2 and of reports of others who have measured sub-surface temperatures in snow and ice, chiefly Wegener,Reference Koch and Wegener 3 Sorge,Reference Sorge 4 JülgReference Jülg 5 and Sverdrup.Reference Sverdrup 6
Suggestions for conduct of a program of temperature measurements are grouped under two headings: DIRECT, those pertaining to the measurements themselves; and COROLLARY, those pertaining to other measurements which must be made at the same time if the program is to have maximum value.
Direct
Method
Electrical means of determining temperatures at various depths below the surface of the firn are indubitably superior to readings of mercury or alcohol thermometers, either suspended in individual holes (WegenerReference Koch and Wegener 3 ) or inserted into holes in the side of a shaft (SorgeReference Sorge 4 ). These electrical elements may be dropped down to the required depths in holes drilled in the snow or ice for the purpose, and the holes filled in immediately; the elements are not recovered at the end of the program.
Instruments
Resistance thermometers (used by WadeReference Wade 7 ) may be more durable than thermocouples (used by SverdrupReference Sverdrup 6 and Hughes and SeligmanReference Hughes and Seligman 8 ) but may have larger errors due to the larger mass of the thermal element. Three wires are required in the leads of resistance thermometers, as opposed to two for thermocouples; this difference may be important when very long leads are planned. In formulating any future program of temperature measurements, the relative advantages of resistance thermometers and thermocouples must be carefully weighed.
Accuracy
Temperatures should be accurate to the nearest tenth of a degree centigrade, at all depths. Greater accuracy is probably not needed for over-all usefulness. The annual variation below 20 m. is so slight that it will not be shown clearly unless readings are truly accurate to the nearest hundredth of a degree centigrade, an accuracy which will probably be unattainable.
Location
The holes in which the thermal elements are dropped should be grouped very close together, and should be at least 50 m. from the nearest building or other drift-causing obstruction or source of heat. (In discussing their observations, both WegenerReference Koch and Wegener 3 and JülgReference Jülg 5 were forced to study extensively the possible effects of heat from buildings on temperature measurements underneath them.)
Depths
Levels at which temperature measurements are to be made should not be spaced uniformly, but rather according to an exponential law. For practical purposes a geometric progression may be used, such as
, 1, 2, 4, 8, 16, 32 and 64 m. Readings at these levels will provide information adequate in detail for any analysis of the thermal structure and regime of a relatively homogeneous sheet of firn. Additional readings, for instance at 3 m. or 10 m., would not add materially to the information obtained; the suggested levels are sufficiently close together for satisfactory interpolation.
Accretion
The exact depth below the surface must be known, within r per cent. for each thermal element. Since the surface will change from day to day and probably will rise during the year, the height of the actual snow surface above the original surface level must be measured frequently. A series of vertical graduated poles, placed around the area of the holes, will provide the required information. Such sticks were used for measuring accretion at “Little America,”Reference Fleming and Alton 2 but were not originally related to the sub-surface temperature measurements. Naturally, the side for observing both temperatures and accretion must be free of unnatural drift; depending on the general location the 50 m. suggested above may not be enough to avoid artificial drifts, especially from large buildings.
Frequency
Readings need be made no more than once daily in the uppermost 5 m., and only every 5 to 10 days at greater depths. However, at each observation a series of three or even five readings, at intervals of three minutes or so, should be taken for each level, to ensure accuracy. (Wade’s observations at first were made four times daily, later twice daily, and finally once daily.)
Duration
Measurements preferably should be continued for thirteen or fourteen months; observing a complete annual cycle with overlap will determine the presence of any year-to-year variation or secular change. However, much value can be obtained by careful measurements over a much shorter period. Even one day’s readings of temperature at the lower levels (32 and 64 m., or 30 and 50 m.) can offer invaluable information, provided the disturbing effects of the drilling have disappeared by the time readings are taken. In such a short-term investigation, the holes will be drilled, the elements dropped in, the holes filled up as well as possible and readings taken frequently until equilibrium is attained. Since the annual variation is almost undiscernable at 32 m. (it is only about half a degree centigrade, at 16 m. at both “Little America” and “Eismitte”), temperatures at these lower levels will indicate the permanent thermal gradient and thus the direction and amount of heat flow.
Corollary
Density
Any computation of heat flow in shelf ice or glaciers involves both the thermal gradient and the thermal conductivity, which depends on the density. Consequently, determinations of density should be made at the same levels as those at which temperatures are measured. Admittedly, this is far more difficult than drilling a hole and dropping into it a wire with a resistance thermometer or thermocouple on the end. Nevertheless, without density values, the temperature readings merely describe conditions and cannot be used to explain them.
Crevasses
In some cases crevasses will be penetrated by the holes into which the thermal elements are to be dropped. Since they will affect the temperature regime markedly, they should be avoided if possible. If measurements must be made in a crevassed area, the exact depths and sizes of all crevasses or other cavities must be recorded.
Weather
Ordinary weather observations are almost certain to be made at the same camp as the subsurface temperature readings, so they need not be mentioned in detail. Preferably, both air temperature and wind speed data should be available from two levels, such as 2 and 6 m. Barometric pressure, records of cloudiness and precipitation times and types may also be of value in interpreting the sub-surface temperatures; precipitation amounts probably cannot be obtained other than from the snow poles mentioned above.
Radiation
For a complete explanation of the temperatures observed in firn and ice, knowledge of the radiation balance is required. Intensity of solar radiation should be measured, and also the amount of outgoing (“nocturnal”) radiation.
Surface
The temperature of the actual snow surface is of immense interest, but there are almost insuperable difficulties in obtaining it. Because of the low density at the very surface, even a fine thermocouple cannot be expected to measure the snow temperature unaffected by radiation. Some sort of radiometric measurement of the snow surface temperature may be feasible.
Thickness
The total thickness of the firn-ice sheet being investigated should be determined, by seismic or other means.
General
Other variables enter into a complete analysis of the internal temperatures of firn and ice, depending on the particular circumstance. For floating shelf ice, the temperature, salinity and currents of the underlying water are important. For glaciers, grounded shelf ice and ice caps the rate of motion and the nature of the underlying rock are important.