Gauging Rain Gauge Accuracy

A few years ago I installed a heater in the Davis Vantage Vue weather station that I installed at AccuWeather HQ. The point of this is to get meteorologically-relevant "liquid precipitation" data during the winter. Now, with both a local newspaper and engineering firm asking for the data, I thought it would make sense to make sure it was accurate. So I pulled the readings from that rain gauge, mine at home, the local official station at PSU, and a radar estimate.

The graph above shows the total amounts over the last 5 months (my home station is omitted for October because of the snowstorm - I have no heater). For the first few months they were all similar but this summer they started diverging, and I'm not sure I can explain that fully. The AccuWeather gauge has been consistently underperforming (75% for May-Sep) compared to Penn State (1.7 miles NNE), which worries me.

But of course rain can vary greatly over short distances, and it isn't that far off from my raingauge at home (1.2 miles southeast). Rain gauges are generally not instruments that have random inaccuracy - if maintenance is performed frequently. I may have been a little behind this year. The most common problem is that birds will deposits which will clog the hole in the bottom of the gauge. When I saw the Accu gauge reading low (in early September), I also cleaned the "tipping buckets" (shown below) which had mold growing on them.

Since they are supposed to tip (and record 0.01") when enough rain falls in, I thought maybe them being heavier would be a problem (though common sense tells you that should over, not under-read the rain). I did the same thing at home and hoped that would make things more accurate -- but the mystery continues.

*My gauge was adjusted by 0.97", the amount from the Accu gauge on Sept. 27, a rain event which I missed.