October 25, 2007

In the summer of 2007, we revised the Plumes & Blooms processing system by using interpolated calibration coefficients as opposed to long-term averages.

We essentially returned to an older method, which calculated the coefficients by interpolating between calibration events.  Since 2003, we had been using long-term averages for each sensor and were assuming that there were no major shifts for those particular channels.

When we applied the new processing system, we found that there were, in fact, significant differences between the two approaches.

For all channels (412 - 555), there was nearly a 30% discrepancy, which resulted in much larger reflectance values at the surface.

A comparison of surface reflectances (and their subsequent processing schemes) was documented by the figures in this folder.

The plots in the left column compare the early data (black) with the most recent data (red).

In this case, both datasets used interpolated calibration coefficients, however, the new surface reflectances were calculated by using a regression of the top 12 meters as opposed to the top 20 meters.  Any differences between the two datasets are most likely due to the number of available measurements in each of those vertical profiles.

The plots in the second column compare surface reflectances processed with long-term averages as opposed to those processed by using interpolated calibration coefficients.

The plots in the first row of both columns show the variation in surface reflectance over time.  Interpolation was used from the start of Plumes & Blooms (1996) to approximately 2003.

At that point, we began using a long-term average approach.  We stopped using the long-term averages in the April of 2005.

A "one-to-one" overlays the first plots to illustrate the bias in each approach.

The plots in the second row of both columns show the ratio of the values and supports the observations of significant differences.

Below are a list of files that have been replaced.

• Radiometric Surface Data
• Radiometric Profile Data
• IDL SAVESET (with below (0-) & above (0+) surface reflectance values)
• Radiometric data from the on-line database

These new files can be found at:

/home/data65/pb/PRR/

For any further information, contact:

dcourt@icess.ucsb.edu