When the wind is calm and the snow falls straight down, it is fairly accurate. Within a couple of tenths of an inch or so.
If it gets sunny and affects the temperature sensor, the "speed of sound" calculation can get off and depress the reading to as much as a half inch.
It is in a really bad location and the wind swirls around the deck and either sweeps the target board clear, or piles it up. I have straight RS-232 serial runnng from it to my Linux box. Some scripts parse the data string and subtract the "no snow" distance and convert it to inches.
With RS-485, the unit from Stefan's outfit could be interesting. I would need to feed power out to the sensor as well as the data line. The Davis rain interface could be interesting. I figure I could adapt my existing scripts if the Davis compatible version doesn't come to fruition.
It might be interesting to see some sample data output from this sensor.
Greg H.