absorption coefficients retrievals

By Claudia Date 2010-11-22 12:59

Hi,

I am trying to obtain a ratio between a670 and a443, and another ratio between aph670 and aph443 from ocean color data using SeaDAS6.1. These ratios are used for bio optical traits estimates (Aiken et al. 2007).

My problems are three:

First: when I apply any of the available algorithms (giop, gsm01,pml,las) for IOP processing, the ratio aph670 to aph443 results in values between 0.245 and 0.265, although in situ values range between 0.2 and 0.7.

I've found that the the resulting absorption coefficients are given in Short Integer between -32767 and 32767 as raw values, whilst the geophysical range is between -0.776700 and 5.77670. The slope is 0.000100000 and the intercept is 2.50000 to transform from raw to geophysical values with LIN as Scale.

This results are for the R9009.1 SeaWiFS data for S2002288103644.L2 processed with SeaDAS6.1 with both updates. Absorption coefficients checked were: a_443, a_490, a_670, aph_443, aph_490, aph_670, for the following algorithms:

GIOP
GSM
LAS (except a670: Raw Data Type : SHORT INTEGER, Raw range:-32767 to -32767, Raw Units: N/A, GeoPhys range: -0.776700 to -0.776700, GeoPhys Units: m^-1, Band Raw to Geophysical Scale Info: Slope: 0.000100000, Intercept: 2.50000, Scale Type: LIN)
PML

QAA could not be used as one of the bands was missing.

The second problem is: I used User Defined Band Operations of SeaDAS via both the GUI and command line for ratio calculus (result= B2/B1). The version 6.1 of Seadas is giving me a Warning of Invalid band number:0, although the version 6.0 was giving me those odd ratio values stated above.

I wonder about this difference in slope and intercept values as I would expect 1 and 0, repectively. This is also what is given in the downloaded from OceanColor Web L3 dayly 9km aph443 (QAA) S2002046.L3m_DAY_QAA_aph_443_qaa_9km
Band Data Info:
   Raw Data Type :    SINGLE-PRECISION FLOAT
   Raw Min: 0.00464000
   Raw Max: 999.900
   Raw Units: m^-1
   GeoPhys Min: 0.00464000
   GeoPhys Max: 999.900
   GeoPhys Units: m^-1

Band Raw to Geophysical Scale Info:
   Slope: 1.00000
   Intercept: 0.00000
   Scale Type:    LIN

Then the third point is: Although I can retrieve aph and a values, I wonder about their reliability because the between-band ratios of them are giving me odd values and also because their slope and intercept values are different of 1 and zero.
Are the retrieved IOPs all right?

Hope you can help me.

Best regards.
Claudia

By tjsm Date 2010-11-22 14:16

Hi Claudia,

Good to see that the work of Prof. Jim Aiken is being used out there! I can certainly answer your question with regard to the PML model retrieval of aph 670. I suspect that most of the problem you are seeing is that the IOP models simply aren't that good at retrieving IOPs the further into the red you go. This is due to the fact that the absorption due to pure water plays an increasingly large part. Certainly in the Smyth et al. (2006) paper we don't even report the values at 676 for this very reason. The authors of the other models might want to add something on this.

I hope this is of some use.

Cheers,

Tim Smyth

By Claudia Date 2010-11-22 15:22

Hi Tim

Thank you for your reply. Yes, as I told you some time ago, that was the idea, to apply the Aiken's work to investigate phytoplankton functional types.

I actually downloaded SeaWiFS data for the Benguella current as stated in the OceanColor Web Data browser for L1 for the same date as Jim's paper: October 15 2002, although they used MERIS data. I would expect good retrievals but I failed to reproduce similar values for the ratios they propose. Then I started to investigate the aph and a estimates using all the available IOP algorithms and the results were always the same, for different regions and times: the aph670/aph443 is always between 0.245 and 0.265. The aph443 varies and follows chlorophyll variation in the image. That's why I reckon there is some odd thing when the IOP algorithms are applied...

The only strange thing that I could detect was, for all the aph and a estimates from the IOP algorithms, the slope was 0.0001 and the intercept 2.5 for the retrieval of geophysical values from raw values.

Would you suggest me anything I can try?
Many thanks
Cheers

Claudia

By bryan Date 2010-11-22 15:57

Claudia,

The LAS model does not produce aph.

The GSM model uses a fixed aph* spectrum, and thus any aph spectral ratio is a fixed value based on the assumed spectrum from Maritorena et al. The ratio should be the same everywhere, all the time. Here are the numbers:

> cat $SEADAS/run/data/common/aph_default.txt

/wavelength(nm),normalized aph
412,0.00665
443,0.05582
490,0.02055
510,0.01910
555,0.01015
670,0.01424

The default behavior of GIOP in SeaDAS 6.1 is to reproduce GSM (with minor differences), so the same applies to GIOP unless you change the aph option. Refer to the documentation:

http://oceancolor.gsfc.nasa.gov/MEETINGS/OOXX/franz_and_werdell_2010_ooxx.pdf

I'm not clear about your concern on the slopes/offsets. The Level-2 IOPs are stored as integers, so slope and offset are needed. The Level-3 are stored as float, so slope/offsets are not needed.

-- bryan

By zplee Date 2010-11-22 16:21

Hi Claudia:

Following what Tim has said, aph670 could not be "analytically" derived from the measured Rrs670 for most oceanic waters. One remedy for such is to model aph670 from aph derived at other wavelengths (such as aph443). The options include Carder et al (1999) and Lee et al (1998).

Best,
zhongping

By Claudia Date 2010-11-25 16:28

Hi, Zhongping and Bryan

Indeed. I've checked those algorithms. Unfortunately, Carder's algorithm is not available in the newest version of SeaDAS. I can retrieve the ratios using Carder's algorithm from SeaDAS 6.0.

Many thanks for your help.
Best regards.
Claudia