Mentioned the issue to a local hobby tinker forum
in the Ann Arbor, MI area, full of techie geeky people :-)
A possible scenario- cut/pasted from the discussion
- Perhaps go with IR or ultrasonic sensor to
measure distance to an enclosed float.
A float attached to a rotational encoder might
give a larger range of accurate distances.
- If there's a requirement to use visual verification
of a water line, process the still images locally
(at the collection site) and send data sparingly.
- Cell phone networks will be costly.
You'll probably have to use a consumer cell phone
and data plan which might be a theft target.
The best option may be low-bandwidth mesh network.
There are a variety of XBee radios designed for
embedded use that would work perfectly.
One of them has a 1200m outdoor range using 50mW to transmit.
If your measurement sites are more than 1200m
from a base station, you'd either chain sites
together (and collect measurements from each
while also relaying other sites) or use a more powerful radio.
The max range is rated 80km (yes! 80,000 meters),
but you'll need a high gain antennae, more power
and clear visibility.
Urban ranges are much, much lower.
At the base station you hopefully have power and
better security to connect to the Internet.
If not, use one of the longer range radios
and keep relaying through them until you reach
somewhere you have an Internet connection.
The whole thing will run off a 9V battery,
so a cheap solar panel can power it.
-You can prototype a complete collection site with
an Arduino Uno, XBee, sensor, solar panel and
PVC pipe enclosure for about $150.
-Adding a camera will jump the cost by at least
$50 and you'll have to use a more powerful
micro-controller to process the images.
The fast, new Arduino Due is software compatible
with the Uno, so you can start with the Uno
and upgrade if necessary.
You'll still use the same low-bandwidth radio solution.
It will just take longer to process and transmit,
so it needs more power.
I had hit up the USGS guys for gauges on the
Shiawassee River in Michigan a few years back.
I'm definitely interested in exploring alternatives
to patch a dying USGS network of equipment.