Water Column correction involving DII and BRI methods. Details see the link below: http://spie.org/Publications/Proceedings/Paper/10.1117/12.977188

Submerged Aquatic Vegetation Mapping Algorithm (SAVMA) coupled with thermal band mapping should be your best solution, however, take care to ground truth extensively, preferably with an airborne AisaEagle hyper-spectral sensor to account for atmospheric interference.  Your biggest challenge will be variations in water depth along the water body, variation in water quality (contamination), either will affect the quality of your data and lastly, adjacent vegetation (along the river or stream you are measuring) since you intend to use satellite sensors anything to close to the river will affect electromagnetic radiation interactions. Good luck!

Victor,

Can you give a little more information about the water conditions you are looking at ?  Are they relatively clear or quite turbid (i.e., water quality)?  Also, to what water depths are you wanting to map down to ?  What are typical spatial size and distributions ?

Not sure if satellite, airborne, or shipborne imaging and data collections would be the best at this stage.  Besides satellite and airborne imaging I have used dual-frequency acoustic profiling systems, along with sonar imaging, to look at bottom sediment characteristics in very turbid waters and coral reef habitat mapping in clear coastal waters.  Very difficult to see thru turbid waters using satellite and airborne imaging.

Pat

Dear Victor Klemas

An interesting question and i will put my utmost effort to provide you with good answer.

- When you want to map something, you must be able to detect it. Either directly or indirectly. When you want to use aerial camera, it would be a direct approach. If the visible reflectance (what we see by our own eyes) is strong enough, probably because the colonies are in large extent and shallow water, possibly it can be detected.

-Every plant with chlorophyll will do photosynthesis. If there is a living plant in the water, of course the PAR (photosythetic active radiation) can penetrate the water, thus photosynthesis happen and the plant can lived. Therefore, the water must have high turbidity.

- In that case, detection at 520-550nm (green), 580-620nm (red) and some portion in 710-800nm are approriate. I suggest you to examine the correlation between the presence of the submerged plant and reflectance from those regions. Use of suitable indices are recommended. However, i have no reference for that.

-Satellite altitude affect resolution (the lower the platform, the finer pixel they can get) and also amount of atmospheric attenuation (the lower the altitude, the lower attenuation they can get).

- The things i concerned is that at satellite altitude is the pixel size and spectral resolution. The pixel may be coarser even tough nowadays we can get 3-5 meters. Normally, it will be distorted based on nadir location. The spectral will be wider (~100nm) in order to capture more signal (This is to increase the Signal to noise ratio).

- If i were you, i would select a satellite data with fine resolution (5-10m) and having high spectral resolution from 520-800nm (three to four bands). Then, i would do the field sampling to determine the plant density. Grid sampling is preferred. After that, i will correlate those parameters and determine its predictive ability.

- If i dont want to or i have restriction or limitation or else that hindered me from obtaining any in-situ information, i would use any indices that suitable to be proxies for green plant and try to calculate it in your area of interest.

- Any effective technique i believe should be started from good hypothesis.

Best wishes,

Rizal

In order to study vegetation, optical sensors might be the most appropriate for you. However, as Chavez and Mahmud said, the transparency of the water you are observing is a limiting point for your study. Turbidity might mask what you are trying to see.

If the water under study allows you to see the vegetation underneath, I recommend you 2 types of sensors: if you need to study any plant function (i.e. chlorophyll content, NPP, pigment concentration, species diversity, etc.), I better recommend you a hyperspectral sensor. If you only need to know the distribution of the vegetation communities, maybe it is enough with aerial CIR photos.

Another key question is the resolution you need. If you don't need very high resolution, you might also use sensors of mid-spatial resolution, such as Landsat, which can be downloaded for free.

It is always a trade-off between sensor features and price. If you provide more details on your research maybe we can better assess you.

Good luck!

virginia