I've mentioned before on this blog how I joined the CoCoRaHS organization a number of years ago, which involves submitting daily precipitation reports. Measuring rainfall is quick and easy. Measuring snow, not so much. They don't just want to know how much new snow fell, they want to know the density of the new snow (aka, the snow-to-liquid ratio), as well as the total snow (old and new combined) still on the ground. All of that can take a while to measure!
There's a whole protocol for measuring the snow-to-liquid ratio that involves taking a core of snow from a snow board and determining the amount of liquid it represents, by either melting or weighing the core (weighing is much faster). It's kind of tedious, but also kind of fun, because I like snow, and I like collecting data, and I like contributing to science.
Which is why I was thrilled to get an email from CoCoRaHS last week, out of the blue, telling me that the snow data I've submitted was "used in a recent scientific publication about using machine learning to better predict snowfall and storm impacts."
The email mentioned how a couple of winters ago, CoCoRaHS sent out a survey to people who regularly recorded snow measurements to verify that we really are following the protocols. I had forgotten all about that survey. Based on the responses, I was one of over 900 observers whose data was used in the paper. And, using that data, the model that researchers created “outperformed existing methods” used by the National Weather Service.
How cool is that?!
The email also linked to a long and interesting article about snow density and how this study will help create better snow forecasts. You can read it here:
https://waterdesk.org/2026/05/machine-learning-snow-to-liquid-ratio/
I'll just say it again: How cool is that?! Getting that email pretty much made my week.