L1B - L2 l2gen very slow

[kimlotte]

Here is what I am running

l2prod1=(chlor_a,chl_oc2,chl_oc3,Kd_490,sst,sst4,tsm_swim)

l2gen ifile=$l1bfile geofile=$geofile ofile=$l2file l2prod=\(${l2prod1[@]}\) resolution=1000 proc_land=true albedo=0.054 maskland=true proc_ocean=1 maskhilt=false maskglint=false aer_opt=0 aer_opt=-99 cloud_thresh=1.0

I am processing PDS file from L0 - L1B which is quite fast, but L1b to L2 is very slow. Is there a particular reason why this is the case?

I will appreciate your reply. Thanks

James

[OB.DAACx - SeanBailey]

James,

Going from L0 to L1B is a reasonably trivial format conversion (Lo->L1A) and count-to-radiance conversion (L1A->L1B), so those will be quite fast.
Processing TOA radiance to derived geophysical products at the ocean surface...that is a wee bit more computationally intensive...OK a LOT more computationally intensive.
Throw in a semi-analytical inversion model (like SWIM) and its a REALLY, REALLY computationally intensive operation.

BTW, unless you're processing data with an appropriate benthic albedo ancillary input (breflectfile (ifile) = input NetCDF file for bottom reflectances and bottom types)
or are processing over an optically deep ___location, you probably aren't going to get believable TSM from the SWIM algorithm...even if you are, it's not a well vetted product.

Sean

[kimlotte]

Thank you Sean.

Just curious on what product replaced tsm_clark available in SeaDAS 6.4?

Regards,
James

[OB.DAACx - SeanBailey]

No product replaced it.

[lmckinna]

Hi James,

Please note that the tsm_swim product is an evaluation product that was developed for optically shallow coral reef waters. The idea was to estimate tsm from the magnitude of derived particulate backscattering coefficient. I would not recommend using the product (I developed it, so feel qualified to make that assertion) and it is likely that the product may not be included in future releases of the code.

Cheers,
Lachlan

[kimlotte]

Thank you all for this information.

Best regards,
James