Looking inside the ocean skin: differential absorption techniques to sense the interface temperature gradient

The crucial role the air/sea interface plays in heat and gas transfers and satellite sea surface temperatures is well known. A full understanding of interface effects is delayed by the difficulties of making measurements inside this thin (~1.0 mm) and erratically moving zone. Atmospheric sounding interferometers have been developed which use the frequency variations in atmospheric absorption to retrieve the temperature profiles (Smith et al., 1983). Using a similar strategy, such instruments have “sounded” the interface temperature gradient. Frequency variations of water´s absorptive properties have been used to measure the temperature gradient inside the interface (McKeown, 1995). The measurement extends less than 1.0 mm into the water; however, it is precisely this zone that is the most relevant to heat and gas transfers and the most difficult to sample. This technique also avoids the problems of thinness and erratic motion that plague mechanical measurements. The water molecule has quantum resonance features which cause the optical properties to vary with frequency. For example, the absorption coefficient drops 7 orders of magnitude from infrared to optical frequencies. A useful parameter is the inverse of the absorption coefficient, called the effective optical depth or EOD. EOD quantifies the range of depths in which a frequency´s radiation originates. Since the radiant flux ofa 300 K ocean is weak at 2.2 μm, the 3.8 μm region is the most practical for existing instruments. To explore the technique, the radiance spectrum emitted from a small well known water body was measured in a laboratory setting. Temperature change over time measured by thermistors allowed calculation of the interface heat flow and the gradient required by that flow (required gradient) to be known. Since gradient measurement is the goal, the brightness temperature spectra were plotted at a frequency´s EOD rather than the frequency itself. This plot, called a “spectral gradient”, facilitates comparison of two gradients