1. Ocean Color Algorithm working Group
ZhongPing Lee

2. The objectives: 1. Algorithm cross comparisons
2. Algorithm recommendations
3. Report on algorithm progress
The algorithm here is the mathematic processes that using water color as input to derive in-water properties.
As we all aware, there are many issues regarding algorithm activities.

3. Major issues regarding algorithm activities:
1) there are many algorithms;
2) many of them are empirical;
3) many of them are only partially tested/validated;
4) limited/no common ground for cross comparison.
IOCCG Report #3, Page 117:
… desirable to set up an international data base …, with free and easy access to all interested researchers, …
To address these issues, we need community effort.

4. This OCAG plan to do:
1. “To assemble data from field measurements”
Link/merge with SIMBIOS data
slowly; measurement error, ..
!2. “To synthesize data based on known relations”
IOCCG/agents support? ..
Before we get to why we want to do these, let me briefly review what is already done.
3. Algorithm cross-comparison/evaluation
!4. A place to store data/algorithm software, ..
A webpage under IOCCG

5. For [chl], cross comparison/evaluation has been performed, and operational algorithms have been suggested.
However, this comparison was focused on [chl] concentrations.
Why we want to compare others?
(O’Reilly et al. 1998)
Only limited comparison/validation for other properties.

6. Traditional color data flow scheme:
[chl] algorithms
many are empirical!
color
IOP,
[chl]
Historically, we need [chl] for the estimation of other properties and primary production.
So, in the past and today, tremendous work has been focused on how to get [chl] directly from color.
auxi.
PP, …

7. . From Ocean Optics theory, Multi-variant, two-step control! [chl] [G]
less understood
better understood
[chl]
[G]
[S] . 1 2
IOP, ..
color
Before we get into color inversion, it is necessary to know how color is formed.
Each step has its own variations.
Multi-variant, two-step control!

8. . A true inversion looks like: Multi-variant, non-linear problem!
[chl]
[G]
[S] .
IOP, ..
color
[chl]
color
Then it is not that hard to know one-step inversion includes all intrinsic variations.
It is time to change the one-step approach to two-steps approach.
Empirical algorithms do not help us understand the science.
Empirical algorithms do not help in error analysis.
Traditional one-step inversion.

9. The limitations of one-step inversion are clear:
One-step empirical algorithms do not help us understand ocean-optics mechanism, and harder to diagnose the error sources.

10. . . Two-step inversion: [chl] [G] [S] 1 2 IOP, .. color easier,
harder,
less accurate
easier,
more accurate
[chl]
[G]
[S] . 1 2
IOP, ..
color
Light field
Heat transfer PP .
auxi.

11. This OCAG will focus on algorithms for the IOPs.

12. 1. Algorithm cross comparisons 2. Algorithm recommendations
After we have:
1. assembled data from field measurements
2. synthesized data
1. Algorithm cross comparisons
2. Algorithm recommendations
3. Report on algorithm progress