Introduction
In order to detect avalanche victims, we want to show by radiometry that there is thermal radiation of a human body buried in a snow layer, that is to say a dielectric area more or less absorptive, placed on the ground. The thermal radiation transmitted above the snow is collected by an antenna which changes this radiation into a slight noise characterized by an apparent antenna temperature.
In order to use antennas of not too large a size, it is necessary to work with high frequencies. On the other hand, the rather low radio frequencies are the only ones that can penetrate a compact, wet snow without excessive attenuation. Therefore a compromise is required. It seems that an interval of frequencies from 2 GHz to 8 GHz can be used.
Now, the detection of a slight noise signal can only be obtained with an ultra-sensitive receiver which uses a pre-amplifier with slight noise and high gain. Besides, the power of the useful noise that characterizes the presence of a thermal source becomes greater and greater as the band width of the receiver becomes wider.
The technology of today allows the production in the S band of a radiometer sufficiently reliable to detect a variation of apparent antenna temperature of about one kelvin (Reference LivaLiva, 1975).
In order to detect avalanche victims, the radiometer is composed of two similar, mechanically coupled antennas that can be moved about over the avalanche area which is to be explored.
A continuous electric signal is obtained at the radiometer output proportional to the difference between the apparent temperatures of the antennas. A servo-motor can cancel out the continuous signal voltage above so that an average local temperature of the snow layer can be defined.
The time constant chosen for the servo-system must be large enough to obtain for a speed of movement of about 0.3 m/s a signal that can be detected when there is a victim under a snow layer. A certain number of false alarms will also be obtained.
We present a model to calculate the difference of the apparent temperatures of antennas, pointing out the importance of the two parameters: water content of snow and thickness of snow (Reference LivaLiva, unpublished). This model can explain why there are false alarms, and can predict the signal detected when there is a human body under a snow layer.
Finally, a systematic study of detection on snowy ground can assess the feasibility of the application of the radiometric method for the detection of avalanche victims.
Model for calculating the difference of the apparent temperatures of antennas
The location of the antennas and of the human body buried in the snow is shown in Figure 1.
This supposes that the antennas are similar and directive enough so that only the A antenna is able to receive the thermal radiation from the victim. We also suppose that the irregularities of the avalanche area surface are slight compared with the working wavelengths.
We respectively call the temperatures of the air, snow, victim and ground, T air, T s, T v, and T g respectively.
The variation of the apparent temperatures of the antennas can be written:
where θ v and θ g represent the rates of transmission through the snow and e v and e g the emissivities of the snow above the human body or the ground.
In practice T g—T s remains small so that the last term in the equation for ΔT app can be neglected. The values of e v, e g and θ v represented respectively on Figures 2, 3 and 4 are calculated from a plane model with three elements (Reference LivaLiva, unpublished).
The values given for ΔT app in Table I are approximate to within a few degrees, and can also be cancelled out.
The difference between the apparent temperatures of the antennas is largest when one of the antennas is over the victim. When the antennas system is in action above the victim, ΔT app undergoes an inversion passing through zero. The distance between the antennas is chosen to be about one metre.
The distance within which the system of antennas goes from one extremum of ΔT app to the other depends on the size of the victim and is between 0.3 m and 1.5 m. A variation of ΔT app of the same type can also be obtained when there is no victim, as a result of inhomo-geneity of a snow layer. In that case, we can write:
where e gA and e gB represent the emissivities of the snow beneath the A and B antennas respectively. If the emissivity of the snow presents local fluctuations on a scale of about one metre, we obtain a false alarm.
A logical device placed in the chain of treatment of the noise signal connected with ΔT app, allows us to take into account only variations that could come from a victim.
Signal detected when there is a human body buried in a snow layer
The snowy ground is explored by transiting parallel to the antennas. The speed must be about 0.3 m/s. The response of the radiometer to movement over dry snow, about 1.6 m thick and with density about 0.38 g cm−3, give fluctuations in the output potential lower than one volt. So the servo-system can keep the level of the output signal within reasonable limits, that is to say can define a local reference temperature.
Figure 5 represents the type of response of the radiometer to an empty hollow in the snow layer that has previously been defined. We obtain a signal with a maximum amplitude of about 4 V. Near a human body placed in the hole, the maximum amplitude of the output signal increases by 1.5 V or so.
It should be noted that if the dry snow thickness changes a lot over distances of the order of one metre, fluctuations of the output potential increase and cause false alarms.
For a wet compact snow with a water content of around 2% in volume and a density of around 0.43 g cm−3, the amplitude of the detected signals rapidly decreases. It is impossible to detect a human body in such snow except if its thickness is below 0.4 m.
False alarms on snowy ground
The directivity of the antennas used requires a systematic exploration of the snowy ground to be studied through parallel traverses separated by 2 m or so. Detection is made by an acoustic alarm which triggers from a 2 V signal. This is associated with an indicator like a galvanometer.
The ground studied had thickness irregularities of about 0.2 m. For dry snow 1 m thick, we obtained a false alarm every 20 m2.
For a wet snow 1 m thick and 2% water content, there are practically no false alarms. In that case, we repeat that a victim can only be detected within 0.4 m of the surface, whereas in a dry snow, detection is certain up to 1 m, with an error rate less than 1%.
On an avalanche field, the systematic exploration by parallel traverses at 2 m intervals is impossible because of the irregular configuration of the snow layer.
Conclusion
The radiometric method is not a practical solution to the detection of avalanche victims, though contrary to the conclusion of Reference Enander and LarsonEnander and Larson (1976), it seems possible to detect a human body under a snow layer if the victim is buried at a reasonable depth and if the water content of the snow is not too high.