How do forecasters predict snowfall?

Sunday, 16 November 2014

It's  one of the hardest things to predict in the weather forecasting business - just how close to sea-level snow is going to fall. In the past, the phrase 'snow to low levels' was about as precise as forecasters were able to get in warning New Zealanders of the white stuff.

Forecasting whether icy rain, sleet or snow will fall at your place is a knife-edge kind of prediction, as rain can turn to snow in the space of 100 metres or less as you drive up a hill.

However, it is a crucial forecast to be able to make, given that most Kiwis live within a few hundred vertical metres of sea-level.

MetService has been fine-tuning its snow forecasting and its findings are helping forecasters pinpoint with greater accuracy the likely snowline and snow depths in a bitterly cold southerly.

Traditionally, forecasters around the world have used two measures of how cold the atmosphere is - the 1500m temperature, and the 'thickness' of the air between two pressure levels - to determine roughly how low snow will fall before melting into sleet and then rain.

If the temperature around the 1500m mark is -5 degrees Celsius, the average lapse rate (increase in temperature) of 1C for every 200m means it will be around 2C at sea-level, around the point where wet snow turns more into sleet.

'Thickness' is just that, a measure of how deep the atmosphere is between two levels, usually between the surface and about 5000m up, but actually expressed in terms of air pressure, 1000 hectopascals (hPa) and 500hPa respectively. The colder the air in that vertical slice of the atmosphere, the dense and heavier it is, and the lower the thickness is in metres.

Generally, for snow to sea-level forecasters look for a value of around 5250 metres or lower.

By way of contrast, in a hot summer nor'wester the thickness between the same two points is often around 5800m.

Snow forecasting using thickness and the 1500m temperature has generally worked well in the past but does not take into account the nuances in the atmosphere which cause localised changes in the intensity and altitude of snowfall.

MetService severe weather forecaster Paul Mallinson, who has been refining the mathematical equations used by weather models for snow prediction, said precipitation above New Zealand in the winter always started as snow.

'It'll fall to some level where it will eventually begin to melt and turn to rain.

'There used to be a golden old rule that the level snow would fall to was about 300m below the freezing level. In heavy snow it would go lower than that.

'But we wanted to do better than that. We wanted a much more objective gauge of it.'

In pursuit of that, Mallinson studied 35 cases where it snowed to different levels around Invercargill, one of MetService's upper-air observing stations which gathers data from scheduled balloon flights.

He was interested in the link between the freezing level and the wet-bulb temperature - the temperature air surrounding the thermometer has if it is cooled to saturation, from which humidity and dewpoint can be calculated. It's like the temperature felt on your skin when you run out of a swimming pool on a breezy day.

'The idea was to come up with an [mathematical] expression that connects the freezing level with the wet-bulb zero level.

'In dry air, that wet-bulb zero level can be quite a bit below the freezing level, but as the air gets moister and moister then the freezing level comes down to the wet-bulb level,' he said. 'That effect can make a difference of hundreds of metres in how low the snow falls. In the cases I looked at, on average, for precipitation falling at an average rate, the snow level comes to about 240m below the wet-bulb zero level.'

However, Mallinson said heavy snow was a game-changer. The 240m figure worked for about 1cm of snow falling an hour. However, heavy snow at the rate of 8-10cm an hour could drop the snow level by as much as 800m.

In terms of improving warnings of how much snow might fall, Mallinson said he came across a model developed in Quebec that could be tweaked to fit New Zealand conditions.

Its calculations of snowfall amounts depended on three things - the strength of the upward motion in the storm (the stronger it is, the more snow that can fall), the relative humidity, and the temperature of the area where the snow is generated (the largest flakes generally form where the temperature is -15C to -20C).

With the new guidance, forecasters had more confidence to predict snow to just 100m or so above sea-level. 'We are pretty well aware now when it is going to get down to sea-level or just above. It simplifies things a bit.'

MetService monitors the accuracy of its severe weather warnings and last year reported 100 per cent success in detecting heavy snow.