Interesting link for ocean color measurement



This sort of goes right along with the OROV project. In-situ measurements of ocean color can give measurements of chlorophyll in the water thereby providing a measurement of primary productivity in the water column. Something like this would be wonderful for coral reef or estuarine research.

For anyone interested, I have also included a peer-reviewed article in PDF format outlining the design of a basic chlorophyll sensor that might work.

370-Leeuwetal2013sensors.pdf (631 KB)


That looks awesome. I have a couple of these laying around I wonder if they would work.

They do some pretty basic hyperspectral analysis using 3 colored LEDs and look at the light coming back to determine the color of the sample. Its amazing how accurate they are. You can put it on any color fabric and it will match the color for example.

The drawback would be the low amount of backscatter in fairly clear water, and the interfereing light inside the system.


The lack of backscatter would be very telling in terms of primary productivity. Clear water in a natural environment is actually an indicator of a weak primary producer population (i.e. algal cells, diatoms, rotifers, other plankton, etc). In other words, the clear water would indicate low chlorophyll content and therefore low primary productivity.

As for the interfering light, I agree with you completely, the answer here would be to place a shield between the receiver and the emitter blocking them from light emitted by the ROV...or you could just turn the lights off while making a measurement.


Yes you are right. Low backscatter is indeed usefull data! Its such an easy thing to implement into a sub, you already have the acrylic window and it can be inside the main hull. Thank you for the great idea.


Carsten was starting to work on something similar. Not sure how far he's gotten since this post:



The sensor I'm using for the 8 colors requires a much bigger pcb than the small TCS3200-DB Mike have.

There's a better option than the TCS3200-DB - use the TCS34725 from AMS that have I2C.

Specs for the TCS:

Specs for AMS:

AMS sensor is capable og the same as the TCS but at the same time splits Measurements up in 3 colors. TCS only delivers amount of light you have not how the color change as you go deeper and deeper.

At moment my idea is to Mount the TCS34725 inside the ROV as I'm not sure my pcb with 8 colors will fit inside. Even if I redo the design as it requires lot of components.


Seems to me that you would want the IR blocking filter removed. The paper I posted shows a blue LED with a red blocking filter. In other words, you want a setup that will detect light in the NIR IR bandpass filter if you will. I haven't done this in an aquatic environment using these methods, but I can tell you that this is very similar if not the same as the NDVI analysis that we do using a photography technique called infrablue. I have a camera that is specially modified to allow IR light to reach the imaging sensor. The red channel is replaced with the NIR band light to allow detection of ground vegetation in geo-referenced imagery using this equation:

NDVI =(rNIR - rVIS)/(rNIR + rVIS)

You can also use the blue channel reflectances in a similar way to locate water on the ground using the NDWI analysis:

NDWI = (rho(0.86 mu m) - rho(1.24 mu m))/(rho(0.86 mu m) + rho(1.24 mu m))

In reading the paper again, I have come to the conclusion that the chlorophyll detection technique they use could probably be used with the same NIR reflectance as used in NDVI analysis. It would be an interesting experiment to say the least.



You can see the color band of sensor I'm planning to use in comment #2 here:

My idea is to sample the colors at specific time interval and later on build a LED aquarium light that will replicate all 8 colors. Think sample each 1 minute would be ok or even 5 minutes. Logger has a 2GB uSD Card so don't think amount of data would fill it...


Not sure about your storage, but your light wavelengths are interesting. The upper wavelengths correspond to chlorophyll reflectances as seen in this paper (link to abstract):

Maybe I am missing something here, but the sensor you spec'd ( appears to have an IR blocking filter meaning that it may not be able to quantify reflectances at the upper wavelengths (670 nm and 695 nm) in your blog post comments. If it will, then it is certainly a candidate for what I was originally talking about here.



The TCS34725 is only 3 colors and "clear", but easy to find on breakout boards that is small and should fit inside the ROV.

What I understand from specs are these:

465 nm , 525 nm, 615 nm.

The "Clear" would be same value as the TCS3200-DB only delivers.

Problem with the 8 color sensor is that at moment my skills can press it Down to a pcb size of 7,5 x 7,8cm. I have all components on one side so a pcb with components on both will be smaller.


I see now. I was a little off in the wavelengths I was referring to. Yours are still in the visible portion, but not by much. NIR starts (according to some sources) around 700nm or so. I have several filters that allow NIR to pass at 695 and 700 nm, but they block out everything else. The resulting photos are really cool to see.


My 8 color sensor span are:

695nm 10nm bandwith
670nm 20
615nm 15
557nm 15
515nm 15
490nm 10
460nm 10
428nm 10


I have seen that same paper and it seems a pretty easy thing to make the fluorometer. When I get my kit built I will give it a try. Unfortunately.. if our role is to be ground-truth for the satellite measurement, we would still need a lab measurement of Chlorophyll from water samples.



The above pic was taken using my Olympus e-420 DSLR with a 695nm bandpass filter. I had to bump the exposure time up and steady the camera for this shot. I did tweak it a bit using a monotone color ramp, but it is otherwise untouched.

Here is the same image with a different color ramp applied. I am showing these to you so you can see the difference in detail when you view the IR spectrum through a lens or if using a photodetector such as the ones we are discussing here.


I definitely agree that the reflectance results wouldn't be nearly as accurate as a lab sample, but for most purposes, the in-situ measurements with the fluorometer would be accurate enough for research. The time and effort required to obtain uniform chlorophyll samples in a marine environment with good primary productivity can be overwhelming. The last time I did it in the 2002 National Coastal Assessment (Corpus Christi, Texas and surrounding bay systems), I used a portable vacuum pump attached to a vacuum filtration apparatus that had to be sterilized after each use at each sampling site. Each resulting sample was collected on a filter paper, wrapped in aluminum foil, then stored in dry ice for transport to the lab. LOTS of work! If you have turbid water with really good populations in the plankton, these samples can require an inordinate amount of time to filter. Sand and detritus entrained in the water samples only increase the time required to process these filter samples.

I can certainly see the utility in actual water samples. Perhaps using OROV with the fluorometer AND a water sampling apparatus (discussed in other posts) would be a valid method of redundant sampling to help with final analyses.

Interestingly enough, our water sampling apparatus was almost identical to this one.

Ours was tied to the end of a nylon rope and the trigger weight was placed around the rope. We had depth markings on it so we would know what depth we were sampling from. Again, this sampling tube was required to be sterilized after each sample. This design works beautifully. I think I just came up with another idea for a peristaltic pump that could be mounted to OROV. You could figure your purge volume from the depth (length of sampling tube) and pretty much eliminate the need for sterilization between samples.


Chlorophyll has a reflectance dead spectrum from about 500 to 610 or will give almost 0 reflectance, that includes chlorophyll a and chlorophyll be. The original paper can cited use one photodiode to quantify chlorophyll and it’s highly reflective region.
I personally would like to use it to only the color spectrum. I would choose a 480 nm excitation to reflect off of both or chlorophyll a and chlorophyll B. And I would also use a 550 nm LED to remove debris as a subtraction step. More channels is a lot more complicated unless you are well versed in Band difference analysis and principal component analysis. You would also you want to make a library for everything you’re looking at.


TOTALLY agreed! Keep it simple! I hadn't thought about the 550nm LED to remove debris, but it seems like a sound idea for sure. The original paper used a 425nm LED for the experiment. Their analysis showed what I believe to be really good results. Their regressions had really good R2 values.


WOW impressive Pictures!


While we are on the subject, your sensor could be used as an analogue to the Secchi disk measurement. It could be used to measure the amount of light attenuation with increasing depth.