Two calibrations were performed for each instrument - a pre-deployment calibration (made at room temperature, 20^oC) and a post-deployment calibration (made near 0^oC, close to the deployment temperature). Both sets of calibration files (20^oC and 0^oC) yielded negative irradiances when used to calibrate the measurements. To overcome this problem, the following procedure was used:

The data were first converted from binary to ASCII format (as raw counts) using the Satlantic, Inc. program asciicon.exe. The measured dark values (found between 23:00 and 01:00 solar time, a total of 20 measurements/day) were then averaged for each day, yielding a set of daily dark values. The raw irradiance data (in counts) were then converted to irradiance units by first subtracting the appropriate dark value for the day, and then multiplying by the appropriate calibration coefficients and immersion corrections from the post-deployment calibration file.  Any resulting negative irradiances were set to zero.

Daily dark values were used rather than an average dark value for the entire time period, since one of the sensors showed temporal variations in the dark values. This sensor (on mooring 6) showed a peak in dark values between days 350 and 380, and otherwise remained relatively constant. The dark values for the other moorings all remained relatively constant throughout the period of deployment. However, mooring 9 showed anomalous dark values on day 430. The irradiance data from this day were removed from the mooring 9 data set.

The data files were then truncated to contain only data between the start deployment time and end recovery time. Anomalous data caused by measurements made during the actual mooring deployment and recovery were then removed by ignoring the first and last 100 data points (10 hours of data at each end).

Note that the resolution of the higher wavelength measurements (665, 683 and 700 nm) is much lower than that of the shorter wavelength measurements. For example, the range in counts of measurements made at 412 nm in one data set is 0 - 14000, whereas the range in counts for the 665 nm sensor in the same data set is only 0 -70, a factor of 200 lower.

Webpage by Jasmine S. Bartlett, Oregon State University.