Some comments on the precip data, in case there is any confusion.
The PIC processor on the ETI gauges generates a tip count as shown below:
data_dump -i 4,80 -A isff_20101216.nc 2010 12 16 05:42:25.8211 5.04 9 ETI8 0\r\n 2010 12 16 05:42:30.8613 5.04 9 ETI8 0\r\n 2010 12 16 05:42:35.9015 5.04 9 ETI8 1\r\n 2010 12 16 05:42:40.9424 5.041 9 ETI8 0\r\n 2010 12 16 05:42:45.9823 5.04 9 ETI8 0\r\n 2010 12 16 05:42:51.0220 5.04 9 ETI8 0\r\n
The above snippet shows what appears to be a false tip at Dec 16 05:42 UTC, (Dec 15 22:42 MST). No other stations reported precip, and the longwave in (Rlw.in) doesn't indicate the presence of clouds.
A tip is .01 inch of precip in 5 seconds. In the example above, the ETI at station 4 is reporting every 5.04 seconds. Using a value of 5 seconds is in error by less than 1 part in 100.
NIDAS converts .01 inch / 5 seconds to a rate, called Rainr, in mm/hr by multiplying the tip value by 182.88, so that one tip is a momentary rate of 183 mm/hr. 60 (or 59) of these 5 second points are averaged over 5 minutes. So one tip in a 5 minute period will result in an averaged rate of 182.88/60=3.05 mm/hr or 182.88/59=3.10 mm/hr depending on how many samples were received from the ETI in the 5 minute period, which is why there is a little jitter in the value of one tip on the rain plot.