water have complicated nature, water acts as a big obstacle in the propagation optical light which ends up with reduction of the optical transmittance in very short ranges which can be in the order of a few tens of meters. on the other hand, we can use the visible band of the electromagnetic spectrum (450nm-550nm), where water is relatively transparent to light and absorption takes its minimum value.

I think the biggest problem are dissolved salts which makes the water relatively conductive. EM waves are quickly absorbed and thus even radio communications is hardly possible in the oceans. In terms of practical applications: sees, oceans, lakes other dissolved substances, dirt etc. make the water very scattering medium. In this sense looking at any band of spectrum of pure H20 does not make sens.

On the other hand there exist some optics method to image through the scattering medium like water on earth. Look at the article:

Optics and Photonics News, Vol. 24, Issue 7, pp. 30-37 (2013)

http://dx.doi.org/10.1364/OPN.24.7.000030

Hi,

There are a wide range of obstacles for underwater communications. They can expand from underwater current since the water refractive index changes on areas with high current, water bubbles and the amount of minerals in the water. Even temperature difference in different water depth affect the laser beam in the water.  

Using high peak power of visible laser sources in (450-550) probably will increase the detecting disatance or communication distance in some extent.

Formally, the optical fiber is not a wire. So, drop the optical fiber into your water and you can communicate across a hundred kilometre.

That is true that fibre is not wire. However, in free space optic the medium cannot be a fibre. Likewise, in underwater optical communication, when the term “wireless underwater optical communication” is used, it is strictly referring to the case that the medium or communication channel is water. Therefore, the fibre still cannot be used. Nevertheless, if for underwater optical communication you use fibre it is referred to “fibre optic communication”. I think we need to be very careful about the definition of an individual type communications.

Optical wireless communication systems’ operating in underwater environment is a very challenging task to investigate. Due to its rather complicated nature, water acts as a big obstacle in the propagation of light which ends up with reduction of the optical transmittance in very short distances which can be in the order of a few tens of meters. Depending on the type of underwater medium such as sea, ocean, shallow, deep, mid-water, harbor or other types, the behavior of optical transmittance will show big variation when examined versus the wavelength. On the other hand, the degradation on the incident light imposed by water is considerable which results in achieving very low bit rates even in short distances. Theoretical and experimental studies have been made to a certain extent to model the underwater propagation to include the scattering and absorption effects due to various constituents existing in water. The effect of underwater turbulence on the optical wave propagation is also being scrutinized. The high bit rates in relatively long atmospheric optics links are today quite widely used in access telecommunication networks. In the underwater environment, communication is realized mostly by acoustics which allows only very low data rates. The industrial, scientific and defense needs are forcing underwater wireless communication to be held at much higher data rates as in the atmospheric wireless optical communication. Currently the use of the underwater optical wireless communication is mainly in the sensor network communication applications where data is collected from the ocean to be used in various scientific research such as monitoring the pollution, oil and gas reserves, seismic investigations and in industrial applications like maintaining the functioning and security of oil and gas pipelines.

In the conducting medium it is the skin depth which decide at which depth radiation can reach. By taking the value of conductivity of sea water you can calculate the required frequency to reach at a depth of say a few Km under the sea. The corresponding wavelength will be very large (~few tens of Km) and such waves can not be generated as we need very big antennas. (For details please see the book  Introduction to Electrodynamics by D J. Griffiths)

I read some comments about fiber and wire.. I think, the initial question should be reformulated: what to you want: to promote some optical systems or to communicate through water without any cable? In the last case, I think, the acoustic waves are more efficient. Over-vice, you question looks like

"Which scissors are better to hummer a screw into a wood wall?"

I think it is important to devide the obstacles to  two part which are related to Absorption & Scattering and Turbulence. In first part , the concenration of some material and the size of  particle are important. In second part, temprature and salinity play an important role.  

Fish.

More seriously: There are three significant impediments for optical communication in water:

1) the absorption in water is completely beyond comparison to optical fiber. I have no figures for the first one, but the last one is about 0.2dB/km. For comparison, in terms of photography, that is one stop per 15 km!

2) The light in an optical fiber is guided by the fiber and does not stray. In water the light is not contained and the free space loss will add on top of the attenuation.

3) The detector in an optical fiber is aimed at the fiber itself and doesn't "see" anything else. A detector in water will probably also see other sources of light (e.g. the sun), especially if magical low-loss water is used.

Also, for those interested, the full text of Shailesh Kumar Jha's contribution, the text by Yahya Baykal and Roger Green can be found here:

http://www.google.be/url?sa=t&rct=j&q=&esrc=s&source=web&cd=1&cad=rja&uact=8&ved=0CCEQFjAA&url=http%3A%2F%2Fopticwise.uop.gr%2Fuseful%2FSIG%2520UW-OWC%2520Scope.pdf&ei=_RgXVJa3J8j8ywOt1AI&usg=AFQjCNGrFt3er5ZUwNdgbHCry-wUgCGFXg&bvm=bv.75097201,d.bGQ

Think of it this way. H2O absorbs photons in a wavelength-dependant manner. Salt in H20 causes more absorbtion. Dirt, chemicals, and fine particles absorb and disperse the photons. Think of it as attenuation of the 'signal.' Gert was only half kidding when he said fish! Any large objects will temporarily block the photon beam.

One of the best ways to design some reliability in the communication system is to use packet transmissions and error detection and correction coding. That way, lost packets can be retransmitted and some error correction can reduce packet loss. This is done is satellite systems.

I think that practical environments such a sea water are also frequency dispersive which puts a theoretical limit on the bit rate of the water 'channel,' e.g., you might transmit good sharp, high rate, photon pulses but they get smeared out at the reciever.

Good luck