Microcontrollers and Sensors [closed]

Broad question, leads to a broad answer. Some things to think about (lessons learnt from my post grad days, doing wireless sensor networks):

1. Have you looked at the Arduino platform, there are many shields you can get that are just "plug and play". When your a bit more confident you could make up your own board. Cheap Chinese replicas are available (this opensource hardware after all) off sites such as ebay, gearbest - if you don't mind waiting.
2. Waterproofing - electronics and water don't mix (obviously), so a good water tight case and perhaps look at some kind of potting compound to protect the electronics.
3. Storing - you will need something to store the data on while your away getting on with life. So perhaps an SD card shield to buffer the sensor data between data downloads.
4. Getting the data - you could ether anchor the sensor node to something and retrieve the node later? taking out the SD card and downloading the data manually. Although you might damage the waterproofing in the process.
5. Wireless data collection - perhaps look into bluetooth, WiFi, Zigbee? The first two are cheap and easy to get modules for Arduino... Zigbee is what I used for low power consumption, but it is relatively quite a bit more expensive. Potential issue here is Radio transmission through water? Not sure about that one? Another option for underwater is sonar. There was an MIT wireless sensor project that used sonar to transmit data along water pipes.
6. Power - how are you going to power this? Lithium battery packs seem to be the way to go if you can. This would provide a relatively long lasting power source. I'd also recommend a voltage regulator, as we had issues with strange sensor values when the voltage reduced at the end of the battery life. The Micro (MSP430) could still run and transmit sensor values over ZigBee, and yet the sensor didn't have enough voltage to run correctly.
7. Digital filtering - a digital filter is easy to implement, consider an EWMA (exponentially waited moving average). It's simple to implement (see wikipedia) and does the job for most things. If you want anything more complex, see Kalman filter. But really I'd recommend starting out simple and making it more complex only if you have to.
8. Data storage format - if you don't have much storage space and yet you still want a reasonable sample rate. Perhaps think about using a high-ish sample rate and then storing the filtered sensor values ONLY when they deviate from the last stored value by a chosen value. I used this method in my postgrad work and significantly reduced the data storage/transmission needs (I'm taking 99% reduction - depending on the signal).

Hope this gets you started...