I have been measuring CDOM absorption off the west coast of Ireland in 1998. This data set have been published in the paper by Darecki at al., 2003 Continental Shelf Research (it can be downloaded from the my Research Gate profile). Unfortunately did not measured DOC at that time. The problem that your describing is well known and it can be explained by CDOM source analysis. If the CDOM and DOC have the same source and physical mixing inn the primary drive of the variablity of both parameters you can have a very tight relationship between aCDOM and DOC. If sources of both parameter are different and there are number other non-linear relationships that can act  as sources or decomposition of CDOM the DOC and aCDOM will decoupled, see also discussion in my 2010 Marine Chemistry paper on aCDOM DOC relationship in South Atlantic Bight. There is no general and common relationship between aCDOM and DOC, and relationships between those two parameters can be only established locally (not globally) see the text book be Wozniak and Dera Light Absorption in Sea Water, Springer for detailed explanation. I would also looked at your data set for location of the coastal front, that separate the Atlantic domain water (blue waters) from the coastal waters (green waters). It is common especially in subtropical and tropical oligotrophic water that you have in the mixed layer very low CDOM absorption high DOC, and below mixed layers you have much higher aCDOM absorption and low DOC. Probably these data points mess up your relationships.  see Nelson and Siegel Review, 2013, and number of papers by Nelson and Siegel, Swan, Kitidis and myself (Marine Chemistry, 2013).

In general a clear selection of your data based on the location of the coastal front shell help you to solve the problem.

Cheers,

Piotr

Following up w/ Piotr's comments (Hi Piotr!) I would suggest a few more analytical steps, and a dip into the literature.

DOC and CDOM dynamics have been a hot topic for 20+ years and there is an extensive literature. Recent (5+ years) developments in sensors and sampling have fueled this even more. But, it is still a frontier as the pools are not well constrained with respect to basic chemistry (origin, availability as substrate, degradation rates, etc) and as you have seen in your data, tend to be highly dependent on the methodology.

I suggest:

1) Compare the salinity to CDOM and salinity to DOC relationships between your sites. Generally salinity and CDOM are inversely related in coastal environments, but the specific relationship is related to the provenance of the freshwater input. Hence it should not be surprising that the two sites have different DOC and CDOM relationships.

2) DOC is the big pool that CDOM and fDOM are fractions of. All three have measurement biases and those biases are probably dependent on the specific components. Further, heterotrophic activity and bleaching are going to differentially impact the total DOC and the fractional CDOM and fDOM. One way to constrain this is to look at deviation from the salinity/DOC or salinity/CDOM relationships. Sort your data on salinity and depth and look for correlations between temperature and DOC and CDOM. The idea here is to use salinity as the conservative tracer and track inputs and/or time on the surface with temperature.

For an example, check out the two profiles from this Navis BGC float:

http://navis.sea-birdscientific.com/profilemap.php?floatId=32

The most recent profile (221, July 24) is just 8 km northeast of a profile made on May 17.  The fDOM profile has flipped from near surface input to near surface depletion modes as the surface water has warmed and chlorophyll concentration decreased.

Note that I use 'as' here fully knowing that it is not the same water mass, but reflects a general trend in the North Atlantic where the Labrador Current influences the surface water.

With respect to your data, if I had to bet, I would guess that the high DOC/low CDOM samples had been exposed to more light and that the water mass had a higher temperature.  

Hi guys,

great comments. Firstly, all CDOM is DOM, but not all DOM is CDOM. Similarly, all FDOM is CDOM, but not all CDOM is FDOM. The ratio of CDOM/DOC is called the 'carbon specific absorbance' at the wavelength at which you measure CDOM. Here, it sounds like you have differences in this metric between the 2 bays, which would suggest different sources of DOM or processes as suggested in previous comments. For example photobleaching of CDOM would decrease its C-specific absorbance.

Another way of separating DOM sources is to look at the spectral slope of the CDOM (see Helms et al., 2008, Limnol Oceanogr. or my 2006 paper in Deep Sea Research). This is the slope of the exponential decrease in absorbance with increasing wavelength. If possible, use a non-linear exponential fit rather than a linear fit to ln-transformed data, as this introduces bias. You can use different slopes and their ratios to get a handle of what is going on.

What wavelength are you looking at for CDOM absorbance? I presume it's in the visible (412 nm or 440 nm). Try looking in the UV (300 nm or 350 nm) if you have scans. CDOM dominates absorbance in the UV, so you get a much better signal to noise ratio. Off course, this is pointless if you're interested in remote sensing applications since you can't see UV absorbance from space. Another way of increasing the signal to noise ratio is by increasing the pathlength of our optical sample cell (e.g. from 1 cm to 10 cm), though this won't help you with the data you already have.

Hope this helps,

Hi all thank you so much for your comments they have been a huge help. The spectral slopes have been useful for identifying different populations of CDOM. Yes I am looking at the absorbance at 440nm because of the remote sensing applications. I have also looked at other relationships (temp, salinity etc) no significant relationships between the temp and DOC/a440 but at least I'm getting somewhere with the data and seeing some interesting results.  Thanks again for taking the time to post your replies and for pointing me in the right direction. It is much appreciated!

As the previous posters have pointed out, variability in the relationship between DOM quantity and optical characteristics have been a hot topic for some time. I have a few thoughts and resources you may find useful:

1. I suggest you read this website for a succinct background on why organic molecules absorb light. Although not peer reviewed, it is an excellent introduction (you do need to know some basic chemistry for it to make sense): (http://www.chemguide.co.uk/analysis/uvvisible/theory.html#top).  The main point is that uv-light absorbance is a a function of molecular structure and not all molecules absorb UV or visible light.  For example, glucose, one of the most important biomolecules, has no appreciable UV-vis spectrum. Alternatively, photobleaching (as mentioned previously) can also degrade the molecular bonds responsible for UV-visible light absorbance.   

2. As you interpret your data, I suggest you think about the landscape and ecological processes producing DOM in your bays (e.g. allochthonous vs autochthonous sources) and consider how that might affect the composition of the molecules. 

http://www.chemguide.co.uk/analysis/uvvisible/theory.html#top