MOBY Oct 2016 Reprocessing (Rev July 26, 2018)

This document describes how MOBY data are processed after Oct 2016 when these modifications were implemented. These are in addition to the Jan 2005 processing changes (see this doc).

Three problems are being addressed:

1. The method to extrapolate radiance attenuation coefficient (Kl) to the surface has been improved to include corrections for Raman scattering and chlorophyll fluorescence (doc)
2. The arm depths used in retrievals was increased by 0.234 m (doc)
3. A single pixel shift in the data for the red spectrograph (RSG) collected at a bin factor of 384 for deployments MOBY203 to MOBY253.

Initially only MOBY254 to current deployments were reprocessed. These deployments did not have the RSG single pixel shift problem.  So for MOBY254 to current deployments only correction 1 and 2 are applied.  As the older deployments are processed the data will be added to this directory.  MOBY203 to MOBY253 will have had all three corrections applied.

1) Improved method to extrapolate Kl to the surface - applied to all deployments (doc)

Abstract: The upwelling radiance attenuation coefficient (KLu) in the upper 10 m of the water column can be significantly influenced by inelastic scattering processes, and thus will vary even with homogeneous water properties. The Marine Optical BuoY (MOBY), the primary vicarious calibration site for many ocean color sensors, makes measurements of the upwelling radiance (Lu) at 1 m, 5 m, and 9 m and uses these values to determine KLu and propagate the upwelling radiance directed toward the zenith, Lu, at 1 m to and through the surface. Inelastic scattering causes the KLu derived from the arm measurements to be an underestimate of the true KLu from 1 m to the surface at wavelengths greater than 570 nm, thus the derived water leaving radiance is underestimated at wavelengths longer than 570 nm. A method to correct this KLu , based on a model of the upwelling radiance including Raman scattering and chlorophyll fluorescence has been developed which corrects this bias. The model has been experimentally validated, and this technique can be applied to the MOBY data set to provide new, more accurate products at these wavelengths. When applied to a 4 month MOBY deployment, the corrected water leaving radiance, Lw, can increase by 5% (600 nm), 10% (650 nm) and 50% (700 nm). This method will be used to provide additional more accurate products in the MOBY data set.

 

The improved Lw product is designated by adding a “2” to the front of the existing product designator. So for Lw1, the improved value using the modeled Kl product is Lw21. And so on for Lw2 and Lw7. Also note that not every good or questionable MOBY file had a modeled Kl product. Because of the way the model works some data sets, even on good days, will not have modeled data.

2) Arm depths increased by 0.234 m - applied to all deployments (doc)

While working on a complete uncertainty budget for MOBY, we have recently determined that the depth reference point on the buoy that has been used for MOBY data reduction is not correct. A better estimate of the depth would add 23.4 cm ±5 cm to the depth of the top arm, effectively increasing the derived Lw and Lwn. For VIIRS blue bands the correction is less than 1% , less than 2% for 550 nm and ~5% for the 670 nm band.

3) Bin factor 384 RSG pixel shift - applied to deployments 203-253

For older deployments the program that reads in the MOB files was not handling an anomaly in the raw MOB files correctly. When the bin factor is set to 384 (which occurs primarily in the Lu data) the RSG sends only 511 pixels not 512. Because of this the RSG data were placed in the array one pixel to the left from where they should be. This problem was fixed at the beginning of MOBY254, so the current data have never had this problem. For most of the bands this 1 pixel shift makes no difference; for bands around 674 nm the difference is about 0.5%.

MOBY system response status: SLC ver#, thermal cor applied, deployment calibration status and lamp calibration status

BOL = Beginning of Life Lamp calibration done

EOL = End of Life Lamp calibration done

Lamp Pre/Post = BOL/BOL means both pre and post calibration lamp have only the Beginning of Life Lamp calibration done, EOL/BOL means the precals have the End of Life Lamp calibration done, etc. N/A means that calibration, pre or post, does not exist and so there is no lamp calibration to keep track of

Pre = Pre-deployment system responses applied

Post = Post-deployment system responses applied

Mean = Pre and Post-deployment system responses averaged and applied

2Int = Blue and Red spectrographs are using difference integration time calibrations

1Int = Blue and Red spectrographs use the same integration time calibrations