Lakes as snowfall sensors: solving the precipitation problem inthe mountain cryosphere

The mountain cryosphere is so large, varied, inhospitable and changeable that we must rely on models of snowfall to map and manage this water resource and to predict how it will evolve. This is an important goal because snowmelt released each summer is an extraordinary generator of wealth and wellbeing in rich and poor countries, but it is among the most sensitive of all major ecosystem-services to climate change. Unfortunately, the water content of snowfall is notoriously difficult to measure accurately on a large enough scale, and conventional gauges are too sparse, small, and bias-prone to constrain precipitation climatologies or weather models. As a result, there are large biases in state-of-the-art assessments of mountain precipitation worldwide, and future projections are even more uncertain.

I present an innovative approach to measuring snowfall that solves many of the problems of existing instruments. I show that lakes, which are common in the mountain cryosphere, can be used as pressure-sensing surfaces to yield accurate observations of the water content of snowfall simply, cheaply, autonomously, and over areas that are thousands to billions of times larger than conventional gauges. Using a standard water-pressure sensor submerged on a lakebed, this approach quantifies directly, as a pressure signal, the mass of winter precipitation as it reaches the lake surface. The measurement precision is readily quantifiable and comparable to, or better than, that of commercial precipitation gauges. More importantly, it avoids the measurement biases of other instruments (like pluviometers or snow pillows) that interact with snow as it falls or accumulates. Crucially, it also senses snowfall averaged over the whole hydrostatic lake surface above, over large areas comparable in scale to model grid cells. The largest single lake tested to date, for example, has an area 100,000 times greater than all the world’s conventional pluviometers combined. In winter conditions, this new approach can therefore largely eliminate both the measurement biases and scaling biases of the global array of conventional instruments, biases that until now have propagated into errors in the output of models developed, calibrated and validated on flawed observations. Initial tests of advanced operational forecast models Arome Arctic and COSMO-1 against these novel measurements reveal model biases of up to 100%, demonstrating the potential for this new approach to transform our ability to measure and model snowfall.