[werk_ag]

I also seem to understand that this could be improved even further by sending the Arduino a "sleep x seconds and wake up to take new readings" command.
EDIT: Now that I think of it, perhaps it can not go to sleep as it must count the wind cup rotations..?

It's true, we can't put it in sleep mode. It uses interrupts to count wind cup rotation and rain meter clicks.

So, can you gentlemen please tell me what's wrong with my math? What am I underestimating/missing? (See attached table pic)

I've never done so detailed calculations, but it is interesting!
Maybe you can make a more complete table, or even publish the excel sheet.

The consumption of the Arduino nano v3 alone is rated at 35mA
The TX and FAN leds when on, not consume more than 10mA.
The TX unit with all sensors installed and activated, transmits data about every 3 seconds, but just by a fraction of a second.

It's important, also consider the consumption of the solar regulator, which at night also feeds power from the battery. The one I use, consumes around 30mA.

Just as an example, my solar powered TX unit don't have any fan attached (the RS with the fan, is connected to a second TX unit that is mains powered). I'm using a 14Ah battery with a 10W solar panel. In the last two years it never failed, but occasionally after three or four days without sun at all, the battery voltage drops to around 12.0V which is low for a solar battery (solar regulator switch off the output at 10.8V). If I ever consider change anything, what I will do is increasing the power of the solar panel to 15 or 20W. This will increase the charge capacity, on low sunlight days.