The rain value from the wxt has been a constant zero:

http://www.eol.ucar.edu/isf/projects/SOAS/isfs/qcdata/plots/20130606/precip_20130606.png

The current output message looks like so:
0R0,Dm=103D,Sm=0.8M,Ta=20.5C,Ua=94.2P,Pa=1.0072B,Rc=0.00M,Hc=0.0M,Vs=12.8V\r\n

I rserial'd in and displayed the current precip configuration with the "0RU cr lf" command:

0RU,R=00000000&10010000,I=1,U=M,S=M,M=C,Z=A,X=10000,Y=6553\r\n

Switched it from M=C to M=T:

0RU,M=T ctrl-M ctrl-J

From the manual, there is an indication that tipping bucket mode may loose some resolution:

M=Autosend mode
  C = tipping bucket: The transmitter sends a
      precipitation message at each unit increment (0.1
      mm/0.01 in). This simulates conventional tipping
      bucket method.
  T = time based: Transmitter sends a precipitation
      message in the intervals defined in the [I] field.
However, in polled protocols the autosend mode
tipping bucket should not be used as in it the
resolution of the output is decreased (quantized to
tipping bucket tips).

After the above change the wxt is now showing non-zero rain counts.

Jun 9 3 pm
Changed reporting interval to 10 seconds in case it improves the resolution.

0RU,I=10\r\n
0RU,R=00000000&10010000,I=10,U=M,S=M,M=T,Z=A,X=10000,Y=6553\r\n

However the wxt reports rain with .01 mm precision, so accumulating over 10 seconds may not be necessary and it could be changed back to a 1 second reporting interval. I plan to leave it in this mode until the next rain event.