Flowcapt : A New Acoustic Sensor To Measure Snowdrift And Wind Velocity For Avalanche Forecasting

Chritin V., Bolognesi R., Gubler H.

Abstract : Wind can create even greater unstable accumulations of snow in mountainous areas than heavy snowfalls. But knowing wind conditions is not sufficient to predict these accumulations because their formations also depend on the snow quality of the snowpack surface upwind of the release zone. Consequently, assessment of snowdrift is required to improve avalanche forecasting. In accordance with this assumption, a new acoustic sensor was developed. The sensor includes a mechanical part designed to form a closed acoustic enclosure. The acoustic enclosure contains a microphone connected to an electrical amplifying and filtering device. Because the output information delivered by the instrument is proportional to the wind velocity and to the flux of solid particles (ice grains) drifted by the wind, the instrument is called an anemo-driftometer. Prototypes of the instrument were first tested in a wind-tunnel and then at an experimental site in the Alps. Then an operational version, called FlowCapt, was developed and connected to an automatic weather station at 2700 m in the Aminona ski resort (Switzerland). During the winter, snowdrift is recorded on the test site along with other meteorological parameters, and avalanche activity, to provide extensive on-site calibration and testing of the sensor. The experiment demonstrates that the instrument is a useful component of the avalanche forecasting chain.

KEYWORDS : Snow drifting, snow engineering, avalanche forecasting, acoustics, jet roof.

1. A new snowdrift sensor to improve avalanche forecasting

To improve the reliability of local avalanche warning systems, parameters directly related to avalanche danger or slab stability have to be measured close to and within the potential avalanche release zones that endanger the area to be protected. Because wind can create even greater unstable accumulations of snow in mountainous areas than heavy snowfalls, snowdrift is a very predictive parameter. Wind speed measurements are not sufficient to predict these accumulations which also depend on the snow quality of the snowpack surface all around measurement stations. Knowing snowdrift by direct measurements is subsequently of high importance for avalanche forecasting.

At present time, snowdrift can be measured by ski patrol men (Fig. 1) with the so-called Driftometer (Bolognesi et al., 1995). This simple instrument makes possible quantitative snowdrift assessments, but it requires the presence of a human observer on the sites. If manual measurements are not possible, snowdrift would have to be estimated from other parameters with significantly lower reliability.

Fig. 1 : The Driftometer catches blown snow particles into a
collector through a tube by the combined effects of filter
and pressure fall. Weighting the collector directly gives a snowdrift index.

The new FlowCaptTM acoustic sensor gives the possibility of a continuous and automatic recording of the snowdrift (~kg.m-2.s-1). Installed upwind of the release zone, FlowCapt provides additional information on the snow accumulation process within the release zone, deformability of the forming slab and erodibility of the snow surface. Because it was shown that snowdrift data increases the reliability of avalanche predictions (Bolognesi, 1996), this information is used by the decision support system NivoLogTM to establish local avalanche predictions (cf. § 5).

2. Principle of the FlowCapt anemo-driftometer

The FlowCapt anemo-driftometer determines both wind velocity and snow particles flux. The detection principle is based on mechanical-acoustical coupling. The sensor is composed of closed pipes containing electro-acoustic transducers and a powering, filtering and amplifying unit. When the sensor is placed into a snow particles flux, the particles shock the sensor pipes, inducing acoustical pressure (fig. 2, left). The pressure is picked-up by the transducers. The electrical outputs are filtered and time-averaged in given frequency ranges to provide a signal proportional to particles flux Q (kg.m2.s-1). The formal relation between the measured acoustic pressure and the snow particles flux Q requires the determination of the mechanical-acoustical coupling equations for the sensor, according to suitable hypothesis about particle impacts. The wind velocity is determined on a similar principle : the wind interacts with the body of the sensor and induces acoustic pressure into air enclosures (fig. 2, right). Suitable sensitivity can be obtained optimising the body shape and structure to the expected wind velocities.

Fig. 2 : Left - Saltation ice particles shocking a pipe. (from V. Chritin). Right - Visualisation of a turbulent flux around a cylindrical obstacle. (from H. Werlé, ONERA).

Because snowdrift happens during windy periods, it is necessary that the sensor strongly discriminates wind from snowdrift. This property can be obtained by an appropriate design of the mechanical-acoustical coupling. With no mobile components and full protected transducers (inside closed cavities), the FlowCapt is very suitable for stringent topographical and climatic environments.

3. Prototyping of the anemo driftometer

Theoretical and experimental campaigns have been carried-out at the Swiss Federal Institute for Technology (EPFL) to develop FlowCapt prototypes (fig. 3).

Fig. 3 : Top left - Prototypes tested in wind tunnel, at LASEN-EPFL. The acoustic response of
cylindrical and spherical forms excited by wind were characterised in the 0 - 12.5 m/s range (from Th. Castelle).

Top right - Calibration with controlled particles flux on test-bench, at LEMA-EPFL (from Th. Melly).

Bottom - Prototype tested at Anzère ski resort (2400 m). Comparison of acoustic records to manual Driftometer indexes (from Th. Melly, V. Chritin).

Results obtained from the validation experiments carried-out with a reference snowdrift measuring device and a reference anemometer show good accordance.(fig. 4, 5).

Fig. 4 : Comparison of snowdrift measured by the FlowCapt prototype and the Driftometer.

Fig. 5 : Wind velocity measurements (m/s vs. time in s.) in wind tunnel.

(a) calibrated MiniAir5ª anemometer response,

(b) FlowCapt prototype response.

On the basis of the obtained results, the industrial development of the FlowCapt began. A particular attention was paid on the calibration and reliability to ensure precise quantitative snowdrift information. The sensitivity of the sensor to wind velocity was calibrated in a wind-tunnel, by comparison with reference anemometers. To calibrate snowdrift, no suitable reference instrument exists. Thus, it is necessary to define a specific method to find the calibration parameters under various conditions (fig. 6) : (1) sensitivity measurements in bench tests with controlled particles flux, (2) continuous meteorological and snowdrift measurements for two winters on a test site, (3) comparison of the FlowCapt response to Driftometer indexes during storms.

Fig. 6 : Left - Bench test. Centre - Test site with an automatic weather station. Right - Manual driftometer measurement place (from V. Chritin).

4. Operational use

In December 1997, FlowCapt has been installed on a standard automatic remote station (Gubler, 1996), at Aminona ski resort (Switzerland). The station is located at an altitude of 2700m on a south facing slope of Mt. Bonvin. Values are recorded every minute (time constant =1s). Snowdrift is integrated from ground to 1m height and between 1m and 1.2 m (fig. 7).

Fig. 7 : Wind and snowdrift recorded at Aminona ski resort, between January 24th and 26th, 1998.

The upper sensor additionally measures wind speed. The station (fig. 8) is equipped with a set of sensors and a number of features that allow for a significant improvement of the assessment of the actual local avalanche danger : snow surface temperature measured by a special infrared radiation thermometer, surface reflected short wave radiation, ground temperature, wind, air temperature and humidity allow for a direct onsite indexing of the formation of weak layers, one of the key parameters for the formation of dry slab avalanches (cloudiness, near surface energy flux balance, dendricity and sphericity (Brun et al, 1992) grain size of surface layer, formation of surface rime, surface melt).

Fig 8 : Swiss Standard Remote Snow Station with additional sensors.

The snow profiler measures the snow stratigraphy at an index point within the release zone, settlement, snow accumulation, fracture height and penetration/ damming of meltwater as well as refreezing. The indication of damming of meltwater at a certain depth within the snow cover at a time resolution of about 30' improves the short term forecasting of wet slabs and surface slides. A specially prepared, low priced TDR sensor attached to the ground surface indicates the arrival of meltwater at the snow ground interface, an important parameter for the assessment of the danger from full depth avalanches. In the near future this sensor will possibly also provide additional information on the state of the base layer. A geophone attached to the system indicates avalanche activity and checks remote control of explosives for artificial avalanche releases. A reliable sensor for a direct measurement if initial fracturing as a precursor for slab avalanche formation is still missing but is currently under development.

5. Combination with jet roof

Protecting the FlowCapt with a jet roof is crucial for many applications where the optimal placement of the sensor regarding to snow drift is close to a ridge within the flow path of drifting snow. At these locations often large cornices may form. In such cases snow accumulation on the sensor is significantly reduced by placing the instrument below a jet roof.

Fig. 9 Jet roof placed on a ridge crest to protect an optical snow drift gauge from snow accumulations (Gubler, SFISAR 1982).

The jet roof locally increases wind and drifting speed and therefore avoids permanent snow accumulations in the vicinity of the sensor. A combined structure has been developed and is currently being installed on a site in the Swiss Alps.


The FlowCapt system is battery powered and includes all necessary electronics to be externally powered by a solar panel or other AC or DC current sources. Total power consumption including the CR500 logger is 40mW. Interfacing to other systems or base stations include radio and GSM transmission, SDI serial protocol, and analog output signals. These features make the sensor extremely useful for remote installations.

As snowdrift quantities is a parameter of high importance to analyse the avalanche release probabilities, the current work consists in interfacing FlowCapt with the NivoLog avalanche forecasting support system (Bolognesi, 1998). This link between hardware and software is a first step towards an automatic prediction system.

Fig 10 : NivoLog Graph software module displays simple graphics from FlowCapt data in order to show quickly snowdrift trends.

Fig 11 : The FlowCapt snowdrift sensor and the NivoLog forecasting support systems can be interfaced to an
automatic weather station to constitute an automatic prediction chain.

7. Conclusion

FlowCapt, an acoustic sensor, is the first automatic anemo-driftometer. It can be connected to weather automatic stations located in stringent environments to provide quantitative snowdrift measurements, which are significant data to predict avalanches. Thus FlowCapt is an essential component of the automatic avalanche forecasting chain : sensor - automatic weather station - decision support system.


Ch. Wuilloud (Service des Forêts et du Paysage du Canton du Valais), J.-C. Amos and F. Meyer (Tél-Aminona SA), A. Dussex (SAREM Anzère).