Snow amounts: The ratio problem.

You may have noticed that some of the region's snow-accumulation forecasts have a range big enough to drive a plow truck through.

Of course, this has something do to with the region's complicated topography and the science of trying to determine precisely just how much a given storm will develop.

But it's also very much tied to the vagaries of the snow-to-liquid ratios. Computer models now are seeing about 0.25 inches of liquid being squeezed out of this storm in the immediate Philadelphia area.

Even if that turns out to be dead-on, it would be difficult to estimate precisely how that would convert to snowfall.

In the old days, it was assumed that an inch of rain yielded 10 inches of snow, or 0.1 inches of rain for every inch of snow.

Recent studies, however, have shown that to be an understimate. The ratios vary widely depending on the types of snowflakes, and those are governed by how cold and humid it is through the atmosphere.

Among the snowflake types, the most-efficient accumulator is the dendrite. Those come in the six-sided stellar shapes so favored in holiday decorations.

They tend to be less watery than other flake types. They have space -- or air -- between their arms, and they stack more efficiently than other flakes.

Forecasting what percent of the snow would be constituted of dendrites is a bridge beyond science right now, although computers have been able to identify areas where dendrite-growth is possible.

Meteorologists are getting better at all this, says Gary Szatkowski, chief meteorologist at the National Weather Service office in Mount Holly. These days they take a hard look at the atmospheric profile and how it might effect the snow-liquid ratios.

For this one, in the Philly area he's looking for a ratio of about 12-to-1 or 15-1. Doing the math, that would put the total accumulations in the 3 to 3.75 range.

But if the atmosphere is a tad warmer than expected, that hold down totals; vice versa if it's colder. That, in part at least, explains the range in the forecast.