Bryan Franz
NASA Ocean Biology Processing Group
16 January 2007
Revised:

Introduction

The OBPG is currently in the process of transferring the full MODIS Terra Level-0 mission archive from MODAPS for ingest into our processing system. This effort is 60-70% complete. In the interim, the data covering the period from January 2005 to present has been fully ingested and is available for distribution on the Level-1 browser. We have been running standard SST processing since 2005, and we have now started ocean color processing in the forward stream. We are using standard MCST calibration and prelaunch characterization, but there are a number of issues with MODIS/Terra ocean color processing that are enumerated below. MODIS/Terra ocean color products currently produced by the OBPG are not considered to be of sufficient quality for quantitative analyses.

SST Processing

Full production in forward stream

Reprocessing from January 2005 to present complete, L2 & L3 products available

Reprocessing from 1999-2004 pending completion of mission ingest

Fully validated with quality similar to MODIS/Aqua SST products

Ocean Color Processing

Processing algorithms identical to MODIS/Aqua standard production

Preliminary vicarious calibration

Forward stream production started 1 January 2007

Reprocessing from 1999-2007 pending completion of mission ingest

Unvalidated, and with many known artifacts and issues limiting quality

Ocean Color Issues

Temporal change in response

In the worst case (412nm), the responsivity of MODIS/Terra has degraded by as much as 40% since launch (compare to 15% for Aqua). The changes are tracked through the solar diffuser (SD) and lunar (SV) calibrations, but these calibration measurements have assiciated uncertainties (e.g., degradation of the SD). A plot of the SD and SV calibration trends for mirror-sides 1 and 2 (MS1 and MS2) is here.

Change in mirror reflectivity

The change in instrument response is due in large part to changes in mirror reflectivity. The amount of change differs significantly between mirror sides. This is believed to be due to an overheating event during prelauch thermal vaccuum testing (but subsequent to most prelaunch characterization efforts), wherein the mirror was damaged by smoke. As of the beginning of 2007, the mirror-side reflectivity difference at 412nm has grown to as much as 7% (compare to 1% for Aqua).

Change in response versus scan angle (RVS)

The amount of change in mirror-side reflectivity, as well as the relative difference between mirror sides, varies with angle of incidence (AOI) on the mirror (where AOI varies with scan angle or scan pixel). The MODIS onboard calibration system can only track changes at two discrete AOI (SV and SD), which allows for a linear adjustment (rotation) of the prelaunch RVS characterization, but this may not represent the true RVS. Residual cross-scan error is likely.

Change in mirror polarization response

Given the fact that the mirror response has changed dramatically over the mission lifespan, presumably due to degradation of the mirror coatings, it must be assumed that the prelaunch characterization for polarization response is no longer valid. We have no proven technique for determining the polarization response on-orbit. We are developing and testing vicarious approaches, but this may be an insurmountable obstacle to quantitative science quality ocean color measurements from MODIS/Terra.

Other instrument instabilities

Through-out the mission, but especially within the first few years, the instrument response has been shown to vary with changes in instrument state (e.g., a-side to b-side electronics, into and out-of safe hold). These changes are removed through the SD measurements, but the corrections are imperfect and some discontinuities may be evident in the ocean products.

Solar diffuser degradation

The ability to track changes in radiometric response is limited by uncertainty in the temporal stability of the solar diffuser (SD). The reflectivity of the SD decreases with time, but that degradation has accelerated since mid-mission when the SD door mechanism failed, leaving the SD continuously exposed. The SD reflectance at 412nm has degraded by 35% since launch. The SD degradation is tracked and corrected via the solar diffuser stability monitor, but that correction also has uncertainties, and the reduction in available signal for calibration means that SD calibration uncertainty is increasing with time.

Detector striping

The OBPG has not made any effort to reduce residual detector striping artifacts beyond that which MCST has derived from the SD measurements. These effects may also vary by mirror-side and scan angle, as the detectors view the mirror at slightly different AOI, which greatly complicates any striping reduction efforts.

Ocean Color Validation

We have not yet assessed quality of the ocean color retrievals relative to in situ sources. This is pending completion of the mission ingest process. However, we have performed analyses to assess residual cross-scan and relative detector & mirror side differences in the derived products, and to assess initial differences relative to MODIS/Aqua

comparison to MODIS/Aqua

comparison of mirror-side 2 to mirror-side 1

cross-scan response (RVS and polarization effects) (compare to MODIS/Aqua)

detector differences (compare to MODIS/Aqua)

Future Work

We are working closely with MCST to further characterize the instrument behavior and to develop and test alternate interpretations of the onboard calibration. We are also developing vicarious techniques that may provide a mechanism for resolving residual RVS trends and polarization sensitivities. Once we have the complete MODIS/Terra dataset ingested, we will generate the retrospective SST products and begin mission-long analyses and cross-sensor comparisons of the ocean color products.