Full calibration of a HydroRad or WALRUS includes the following procedures. Since some of the steps in the comprehensive calibration measure characteristics that are normally very stable, we offer a radiometric-only option that includes the most critical subset of procedures. We recommend a comprehensive calibration at least once every two years, with a radiometric calibration every six months to one year.
We connect your instrument, using your own cables if you supply them, and test all the applicable instrument functions, including batteries, depth transducer, switches, and of course the fiber optic cables and collectors.
Each channel is illuminated by a Mercury-Argon source containing various well-defined spectral lines. We observe which pixels on the CCD are illuminated by specific lines, then calculate a polynomial function to translate pixel number into wavelength. The polynomial coefficients are included in the calibration file.
Nominally, the raw signal from the spectrometer is directly proportional to the light energy it receives, but this relationship is not perfectly linear. We measure this effect by illuminating each channel with a constant light source, then varying the integration time over a wide range to produce a range of raw signals. Because the relationship between received signal and integration time is very accurate, any inaccuracies can be attributed to the nonlinearity of the CCD. A table of compensating values is calculated and included in the calibration file.
Electronic delays can cause a slight discrepency between the effective integration time and the time designated by the instrument firmware. This is manifested as a constant offset, which is measured as part of the linearity procedure described above. This offset value is included in the calibration file.
The spectrometer produces some signal even in the absence of light. These dark offsets vary as a function of temperature, integration time, and from pixel to pixel. We slowly change the instrument temperature over its complete operational range (0 to 35C); at each temperature we also measure at integration times from 21 ms to 20 s. The resulting data set allows us to calculate a comprehensive dark-correction function. Dark-correction coefficients for each pixel are included in the calibration file.
The radiometric calibration is done in air, but immersion in water has a strong effect on the response of a collector. For a radiance collector, the immersion effect can be accurately calculated from the known effect of refraction at the interface between the window and water, which have well known indices of refraction. The effect is less predictable for irradiance collectors, so they must be measured. We shine a stable light source onto the collector, first in air and then under water. We can calculate and remove the effect of refraction at the air-water interface, leaving only the effect of the collector's immersion.
The instrument is enclosed in a water-filled pressure chamber, and exercised over its rated pressure range. The pressure transducer's raw readings are recorded and fit to the responses of a calibrated transducer in the chamber.
This is the heart of the calibration: each collector is illuminated from a NIST-traceable calibrated irradiance source. This allows us to calculate the ratio between each pixel's raw output and a known radiometric quantity. These values are included in the calibration file.
After the calibration is complete, we apply the calibration file to data collected in front of the same light source used for the radiometric calibration. Any problems with the calibration file will be revealed as a mismatch between the HydroRad's readings and the known output of the calibration source.
As a "real-world" quality check, we take the calibrated instrument outside and measure the ambient solar light. We look for good wavelength and magnitude matching between channels (when applicable), the correct shape of the well-known solar spectrum, and reliable performance of the instrument's functions.