L'évolution de la neige déposée à moyenne altitude

The behaviour of snow on ground at medium altitude

Docteur-Ingénieur.

Résumé

This report presents the results of four years of operation of a laboratory set up at Col de Porte in October 1959. Necessary emphasis is laid on the difficulties associated with measurements of this type, which are partly due to the researcher not having full control over his experimentation, and partly to the medium investigated, which requires reliable equipment. This is why only a single set of usable data has been obtained for certain parameters in three years. The initial part of the report considers the accumulation of snow on the ground, and the second part deals with thermal conditions in snow. The measurement of precipitation in the form of snow is a very delicate operation in which wind is a disturbing factor and a perpetual source of error affecting measurements with any type of equipment or instrument, from a simple tin can to the most elaborate snow recorder. An essential requirement for reliable results is a site adequately sheltered from the wind, though the obstacles providing this protection must not impede the circulation of the snow. This condition is satisfied at the Col de Porte snow recording station. Original equipment at this station included a snow and rainfall recorder and a totalising instrument. By 1963, its precipitation-measuring equipment comprised two totalising instruments, three snow gauges and five rain gauges. Comparison of data measured with these various instruments led to conclusions which are no doubt equally applicable to any site sheltered from violent winds. The best instrument beyond any doubt is the totalising instrument with a Nipher cone. Data measured with two totalising instruments 30 metres apart show a discrepancy of only 1 %, which also suggests that precipitation was evenly distributed over the measurement platform. This impression is confirmed by comparison of the results obtained with two conventional rain gauges which, while giving figures about 15 % lower than the totalising instruments, differ from each other by only 2 %. The tipping through type snow and rain recorder is not at all suitable for the measurement of this type of precipitation, because of the bulky obstacle provided by its cylinder measuring 50 cm in diameter and 130 cm in height. Considering R = [(P - T)/T ] (where P and T are the amounts of precipitation received over a long period of time by a rain recorder and at totalising instrument respectively), it is found that R = 0 for precipitation in the form of rain and snow if there is more rain than snow, that R = - 6 % if the amounts of rain and snow are roughly the same, and that R = - 22 % to - 25 % for snow alone. Losses by evaporation due to cone temperature rise only amount to a few hundredths. Where all the precipitation occurs in liquid form, the totalising instrument cannot collect as much water as the rain recorder. The Bendix rain recorder gives better results, probably because of its more compact dimensions. They are roughly 8 % higher than those given by the through-type instrument. Ordinar rain gauges give results about 15 % lower than the totalising instrument. It seems possible to improve these measurements by using a smaller instrument protected by an Alter cone at a sheltered site. A totalising instrument comprising a 300 cm2 cross-section tube and an Altercone has been set up at Plan des Aiguilles near Chamonix. Its indications have been found to agree with the mean snow depths measured by taking core samples from a roughly 4,000 km2 area nearby. An alternative method of determining the amount of snow on the ground is by measuring the water equivalent of the snow layer, either by taking snow samples of given cross sectional area and determining their mass, or by measuring the absorption of radiation from a Cobalt 60 source. The accuracy of the Cobalt 60 snow gauge developed by DTG is independant of the snow layer thickness. The instrument is accurate to within 2 mm, which is quite remarkable. Allowance should also be made for interference due to the snow gauge support, however, which may appreciably modify the absolute snow quantity determined. Measurement accuracy by core-sampling methods may vary between 10 % and a few percent. Comparison between direct bulk density measurements and recordings does not show any abnormal discrepancies, especially for 1962 and 1963. As can be seen from the following equations, bulk density varies linearly with time: 1960-1961 ρ (t) = 3.0 t + 239 1961-1962 ρ (t) = 3.0 t + 233 where ρ (t) is in kg/m3 and t in days. The quantities of snow during the two above winters and the time they remained on the ground are both comparable. By the time they had finished accumulating, their respective water equivalents were 650 mm and 630 mm. The two following equations were found for 1962-1963 : - ρ (t) = 5.3 t + 208 ρ (t) = 2.5 t + 264 The first of these equations typifies a period following a substantial fall of snow, with a mean air temperature in the neighbourhood of 0 °C. Scatter in the space distribution of ρ at a given time is about the same as that affecting snow layer thickness data. Knowing ρ at a given point on the platform, the water equivalent can be calculated for any point at which only the layer thickness is known. Measured snow quantity data are tabulated below (in mm): 1960- 1961 1961-1962 1962-1963 Precipitation during the accumulation period... 700 860 1,320 Water equivalent after accumulation... 650 630 860 The melting of snow and its internal heat transfer conditions are discussed in the second part of this report. Melting rates are measured by lysimeter, and continuous recording of the rate of flow is provided by a counter. This arrangement only worked satisfactorily in 1962-1963, but some use can nevertheless be made of the 1961-1962 recordings as well. The flow of water from the melting snow through the lysimeter amounted to 660 mm in 1961-1962, compared with the water equivalent of about 630 mm after accumulation. The flow recordings show a daily cycle featuring two extremes, with a minimum at about 12 noon and a peak at the end of the afternoon. These times depend among other things on the thickness of the snow layer and the rate of flow. The folIowing relationship has been found (where M is the melting flow in mm and θm is the mean positive air temperature) : ΣM = 4.14 Σ0m Daily flows rates vary between 0.3 and 6.5 mm/hr. A characteristic effect when melting starts is arise in the mean temperature from - 2.2°C (during accumulation) to + 7.8 °C while melting is in progress, with a simultaneous rise in the absorbed solar radiation rate from 70 ly to 233 ly daily. In 1962-1963, this instrumentation comprised three lysimeters, two of which remained in continuous operation. The melting flow amounted to 710 mm, compared with the water equivalent after accumulation of 850 mm. The difference can be explained by evaporation. Melting lasted for 47 days, compared to only 19 days the year before. Mean temperatures during the accumulation and melting periods were - 3.7 °C and + 4.9 °C respectively. Mean solar radiation absorption amounted to 130 ly, i.e. less than 56 % of the previous year's rate. All these facts taken together seem to substantiate the assumption that evaporation during the melting process is by no means negligible. The loss of flow amounts to 16.5 % of the overall snow quantity. The following measured evaporation data are given as an example : During a single night, Church measured evaporation amounting to 2.5 mm, with snow frozen on the surface, and Baker measured 75 mm for a total flow of 550 mm. i.e. 13.5 %. Results measured at St. Louis Creek (I) and the Upper Columbia Snow Laboratory (II) are listed in the following table : - Table I : 1946-1947 1947-1948 1948-1949 1949-1950 14th May to 5th June 1948 : 53mm Total precipitation (mm)... 1100 1 175 850 1375 19th May to 20th June 1949 : 56mm Evaporation (mm)... 340 260 285 350 13th May to 28th June 1950 : 122 mm Evaporation (% precipitation)... 31 22 34 25 The flow deficit observed for the entire season in 1962-1963 amounted to 320 mm, i.e. about 22 % of the total precipitation, and is thus of the same order as the evaporation rate measured at U C S L. As for the previous year, the flow recordings again show up the daily melting cycle, but the extreme conditions occur later in the day because of the large quantity of snow on the ground and the low rates of flow (minimum between 15.00 hrs and 18.00 hrs, maximum between 21.00 hrs and 24.00 hrs). Maximum rates of flow are generally less than 4 mm/hr. Not until the end of the melting period does the cycle show features comparable with those of the year before. Using the same notation as before, the following equation was found: ΣM = 4.36 Σ0m Allowing for the fact that the melting rate for 1961-1962 was underestimated by about 5 %, the two relationships are seen to be compatible. Assuming the mean air temperature to remain positive and to only vary slowly, and that no precipitation occurs, the average daily melting rate M in mm is connected to mean air temperature by the relationship M = 3.64 θm, + 5.2. The correlation factor is r = 0.84 and the associate standard deviation σ M,0m= 6.3' mm. Snow depletion measured by Cobalt 60 snow gauge between the 10th April and 18th May is exactly the same as the flow determined by lysimeter. The next few paragraphs consider temperatur distribution and heat transfer effects in the snow. Conventional equipment was used to measure temperatures and the vertical temperature gradient, comprising a series of copper-constantan thermocouples and an electronic potentiometer recorder. This method has been abandoned at the new snow laboratories near Chamonix, where entirely new equipment developed in the electronics and mechanical design shops of the C.N.R.S. Alpine Glaciology Laboratory is being used. The data obtained with this equipment are accurate to within less than 0.1 °C, compared with the accuracy of about 0,25 given by the thermocouple method. The author is the first in France to have plotted the equal temperature line distribution within a snow layer, which shows up the propagation of "cold" and "hot" waves and the insulating properties of snow (this irrespective of the temperature at the snow surface ; the temperature at the ground surface under a 30 cm covering remains in the vicinity of 0 °C). The plot also shows up the importance of transfers by changes of state. Percolating water in the liquid state causes a rapid temperature rise in the snow, whereas the presence of free water attenuates the effects of a "cold" wave, as part ofthe "frigories" (French unit of cold = 1 frigorie) are used up in freezing this water. An approximate net balance has been established for certain periods. For a mass of snow with a water equivalent of 425 mm and containing 5 mm to 10 mm of free water, calculation by equation φ1 = - K (d0/dZ) shows that "frigorie" penetration amounts to about 100 to the square centimetre. Between 90 and 110 "frigories" to the square centimetre are necessary to freeze half the free water content and bring about the final temperature distribution. The variation in time of temperatures at various distances from the surface shows how the "cold" waves damp out with increasing depth. At 30 cm from the surface, for instance, the amplitude of these waves is about 6 °C, but flaws off to 3.5 °C at 60 cm, and finally to about 2 °C at 90 cm. A further example shows up the important effect of changes of state on thermal snow conditions. An atmospheric "cold" wave similar to the one just considered, but propagating through wet snow, cools the snow down by 10 °C at 10 cm from the surface, and by 2.5 °C at 40 cm. Part of the "frigories" are used up in freezing the free water. A qualitative indication of the importance of the transfers taking place in the various layer elements is provided by the vertical temperature profile at a given instant of time. The thermal conduetivity of snow (K) has been calculated for two different densities. Results were K = 52 X 10-5 C.G.S. for ρ = 0.25 g/cm3 , and K = 60 X 10-5 C.G.S. for ρ = 0.32 g/cm3 The quantity of "frigories" stored by the snow during a season has also been calculated, as a result of which it has been found that most of the energy involved in melting the snow originates from solar radiation. Comparison of minimum temperatures at the snow surface with air temperatures shows that, given a clear sky, the snow surface temperature may be as much as 15 °C below the air temperature. This difference becomes much smaller when the sky clouds over.