Observing a mountain valley: a reference design

01What a mountain valley needs to see

A valley observing site exists to answer one question for the people below the slopes: is it safe to be here now, and will it stay safe? Avalanche warning services define danger as the potential for avalanches to cause damage to something of value¹.

That danger is judged from the stability of the snowpack, how widely weak spots are spread, and how large an avalanche could become¹. European warning services publish it on a common scale with five levels, from low to very high¹.

Forecasters also name the kind of problem they face. For new snow, the amount of fresh loading on the old snowpack is the crucial factor, and how critical it is depends on air temperature, wind and the old snow surface². Those are measurements a valley site can make continuously.

Snow is not the whole story. Rockfall, debris flows and creeping slopes share the same valleys and respond to the same rain, melt and freeze. A seismometer on bedrock can register large slides and rockfall as ground motion, which gives the site a sense the weather station cannot provide.

The reference site therefore brings together an automatic weather station, a compact radar, a seismic vault and a private network tower, joined to an edge module on the valley floor. Each answers part of the question; together they give a forecaster, a road crew and a lift operator one shared picture.

02Siting around the terrain

Mountains decide what every instrument and radio can see. The international guide to observing practice covers snow measurement, automatic weather stations and weather radar in separate chapters, because each has its own exposure rules³.

A weather station for avalanche work stands where it samples the wind that loads the starting zones, not in the sheltered spot that is easiest to reach. A snow depth sensor needs undisturbed ground, clear of drifts built by buildings, fences or the mast itself.

Radio links fail first in the shadow of a ridge. Designers of fixed line-of-sight links use international propagation methods⁴, and when terrain blocks the path the signal bends around obstacles by diffraction, which has its own prediction methods⁵.

A clear straight line between two antennas is not enough. The radio wave occupies an ellipse around that line, called the Fresnel zone, and a ridge or a snow cornice that intrudes into it weakens the link even when the far mast is in plain view. The simulator shows how wide that zone becomes at mid-path.

This is why the reference design puts the radar on a high shoulder and the seismic vault on firm rock, then checks every link against a terrain model before anything is built. Where a link cannot clear the ground, a relay on a spur or a different band is chosen at the design stage, not after a winter of dropped readings.

03From sensor to decision

Every instrument on the site joins one private mobile network run for the valley's own use. Mobile standards describe such non-public networks, and note that keeping network entities on site lets an outage be handled locally⁶.

The edge module admits each station, checks what it is, stamps its data against the site clock and keeps a copy. Seismic records travel in an open format that can also carry the recording system's state of health, with a companion description of each station's location and response⁷.

Decisions that cannot wait are made on the valley floor. A sudden rise in wind loading or a burst of ground motion near a road can raise a local alert at once, while the same values continue upstream to the warning service for its regional forecast.

When the site does publish an alert, it uses an open, non-proprietary message format for all kinds of warnings, with geographic targeting and support for updates and cancellations⁸. A road authority, a resort and a public warning channel can then read the same message.

04What travels with each value

A wind speed from a ridge is only useful if the reader knows which ridge, at what height, measured how, and whether the sensor was iced. Each value leaving the site carries its time, place, instrument, method and quality flags through an open interface for sensor observations and their metadata⁹.

Quality management is part of observing, not an afterthought, and the international guide gives it a volume of its own³. In the valley that means flagging a rimed anemometer or a buried snow sensor at the edge, where the evidence is fresh, rather than discovering it in a forecast office days later.

Weather observations can also flow to national and international services. The global system for sharing meteorological data is moving to open web standards and a subscribe-and-notify model¹⁰, so a valley site can contribute without a special link.

05When the valley is cut off

The storms that load the slopes are the same storms that bring down power lines and ice up antennas. The reference site is designed to keep observing, deciding and recording while its link out of the valley is lost.

Local alerts keep running on the edge module, the site clock holds its time through the outage, and stored observations are sent on in order, with their original timestamps, once the link returns. The warning service receives a complete record, not a gap.

A failed instrument is handled the same way. Its silence is recorded as a gap with a reason, never filled in quietly, so the forecaster can see at once which part of the picture is missing and weigh the rest accordingly.