Originally published February 13, 2013 on paleopix.com/blog by Penny Higgins:

A while ago, I proposed an experiment in which I collected snow at regular intervals during a Lake Effect Snow event. I made some predictions and collected the snow, and have now finally succeeded in analyzing the waters. The results weren’t quite what I expected.

I had predicted that isotopic values in the snow would not change over the course of the event. This was because all the snow would be forming directly off the lake, which is only a few miles away from the collection point. (This is in contrast to other synoptic storms, where we have precipitation coming from a single vapor mass, which will evolve isotopically over time. Read more about that here.) The temperature of the lake water, and its isotopic value would not change consequentially over the course of such a short event.

What I saw instead was an increase isotopic values overnight, and then a decrease the next day.

Results from January 22-23 Lake Effect Snow Event
Results from January 22-23 Lake Effect Snow Event. Click to enlarge.

I compared the isotopic values with measures of air temperature during that period of time. I selected first the air temperature measured at the Rochester International Airport (ROC). Isotopic values do track temperature changes, thus I realized that what is most likely happening is the fractionation of isotopes (the selective evaporation of the heavier versus the lighter water) in both the formation of the water vapor off the lake and more importantly the freezing of that vapor into snow which is changing over time due to temperature.

I realized that ROC is actually sufficiently removed from the lake, that its measured temperatures are likely to be different than those directly adjacent to the lake. Shoreline temperatures are moderated by the warmth of the lake water itself. Temperatures between ROC and the lake shore are known to differ by as much at 20 or 30 degrees. I retrieved data from a WeatherBug weather station right on the lake shore (Forest Lawn Beach, FLB on the plot. Thanks to Parker Zack and Kevin Williams for helping me find this.) as it happens, for this snow event, temperatures at ROC and at FLB track each other quite closely for much of the event, until the event peters out. In either case, isotopic values track air temperatures.

The snow gets isotopically ‘heavier’ during the colder overnight hours. Does this make sense?

Under warmer conditions, more of the heavier isotope will be incorporated into water vapor. In isotopic terms, this means that δ18O and δ2H of the vapor will be more positive when air temperatures are warmer. For freezing (or crystallizing snow), one might expect that more of the heavy isotope would remain in the vapor when the air temperatures are warmer. Or, since we’re measuring snow, warmer air temperatures means isotopically ‘lighter’ snow. If it’s colder, more of the heavy isotopes go into the snow, causing the δ18O and δ2H values of the snow to become more positive.

Oh thank goodness! It does make sense! That is if the changes in isotopic value of the snow is directed by air temperatures during the crystallization of the snow and we assume that air temperatures have minimal effect on the fractionation during evaporation.

Can we make the latter assumption?

I think we can. The temperature of the water is close to freezing (approximately 4 degrees C, data found here). Evaporation stops if the water freezes. The difference in fractionation of evaporating water at 4° C and 0° C is negligible (see article here). We can assume it is essentially the same. Thus any isotopic change we see must be due to changing air temperatures during the freezing of snow.

Other observations

Snow was collected at two sites in Wayne County affected by this Lake Effect event. One site in the Town of Williamson, and one about seven miles further west in the Town of Ontario. Isotopic values of snow from these two sites are essentially the same and follow the same pattern. Thus we can say there is likely to be little lateral isotopic variation in snow isotopic values. That makes sense given that the snow is all coming from evaporation off the same lake.

Further work

If the isotopic value of the original lake water is known, along with air and water temperatures, it is possible to look at the extent of fractionation both during evaporation of the lake water and crystallization of the snow. We were unable to collect a lake water sample at the onset of this event, but we do have one collected from November of 2011, as well as snow measurements also from 2011. Alas, for the November 2011 event, we lack temperature data. But we can make some assumptions and try to look at fractionation. I’m working on those calculations now. And they make my head hurt.

For the next lake effect event, I’m hopeful we can get a sample of Lake Ontario water for a starting point. We will also collect snow from the weather station at FLB to see if there is a gradient in the snow isotopes from nearer the source to farther outboard (like Williamson). Sublimation may be occurring in the clouds, which might cause the snow to be isotopically heavier than ordinary fractionation would predict, in which case we would predict that shoreline snow would have more positive δ18O and δ2H values than snow collected further inland.