Ocean optics · open questions

Where the sources disagree

Some things in ocean optics have stayed unsettled for decades: measurements that disagree with each other, or with what other measurements say they should be. The package jerlov does not settle them. That takes new measurements. What it can do is put numbers on how far apart the published values are, and on how much of the gap the conditions a source leaves unrecorded could explain. Four questions, one figure each.

Question 1

How much do the published absorption spectra of pure water disagree?

Pure water is one substance, so its absorption should be the same in every careful laboratory. Below about 450 nm it absorbs so little that the measurements are hard.

Found
  • Pope & Fry (1997) over Mason, Cone & Fry (2016), both measured in an integrating cavity: .
  • In pure sea water the spectra give remote-sensing reflectances up to 6.85 times apart at 380 nm and 2.62 at 443 nm. In Jerlov IB, whose own contents absorb more, only 1.12 to 1.34 times.
Not known

Which laboratory value is right below 450 nm. No newer laboratory measurement than Mason et al. (2016) was found. jerlov does not choose: every function that needs a pure water absorption takes it from the caller.

Absorption of pure water, 1/m, log scale. Each source as jerlov ships it; gaps are where a source gives no value.

Question 2

What happens to light scattered at less than 0.1°?

Most light scattered in the sea turns by only a small angle. Petzold's measurements of 1971, still widely used, start at 0.1°. Below it his report extends the curve by a power law.

Found
  • The extension holds of all the scattering at the eight stations (right).
  • Bogucki et al. (2004) measured below 0.1° and found scattering on turbulence, which depends on how the water is being mixed that day, of 1 to 10 per metre. Petzold's extension cannot contain it.
  • It hardly matters for daylight. Dropping the extension moves Kd by −0.63 to +0.13 %: light turned by so little goes on as if unscattered.
Not known

How much of the light near the beam turbulence takes, against particles, in a given sea.

Top: phase function (β/b, per steradian) of Petzold's eight stations at 530 nm, from 0.1° to 180°, log-log. Nothing to the left of 0.1° was measured. Bottom: the share of b that the report puts below 0.1°.

Question 3

Do measured IOPs give back Jerlov's own Kd?

Williamson & Hollins (2022) measured absorption a and scattering b in waters they grouped by Jerlov type. Jerlov printed how fast daylight fades in the same types, Kd. If the two describe the same water, a radiative transfer calculation given the first should give back the second.

A Monte Carlo ran their a and b for six types at twelve wavelengths. The sun, the shape of the scattering and the pure water, which neither source records, together move Kd by times.

Found

Everywhere but the blue end, the two sources agree within what those conditions can make. The outlined cells lie outside: IB is too high at 425 to 500 nm, and the coastal types too low at 425 to 450 nm.

Not known

Why Jerlov's 1976 coastal blue Kd are higher than both his own 1968 values and the waters later typed by them. Which of IB's six campaigns are right.

Kd computed from Williamson & Hollins's a and b, over Jerlov's (1976). Blue: the IOPs give less attenuation than Jerlov printed; orange: more. Outlined: outside what the unrecorded conditions can reach.

Question 3, continued

Jerlov against Jerlov

Between the 1968 and 1976 editions of his tables, Jerlov raised the coastal blue Kd, by up to times. His 1976 text names the change at 600 nm in the clear types, and says:

it seems that the last word is not said about the behaviour of the irradiance attenuation in the "slope" region of 600 nm

For the coastal blue, no reason has been found in any text read so far. Williamson & Hollins's coastal campaigns measured blue Kd nearer the 1968 values.

Kd in Jerlov's 1976 table over his 1968 table, at the wavelengths both print.

Question 3, continued

Two waters with the same name

Solonenko & Mobley (2015) also give a and b for each Jerlov type, inverted from Kd rather than measured. Their beam attenuation c = a + b for a type is not Williamson & Hollins's, and the gap is not even on one side.

Williamson & Hollins (2022), measuredSolonenko & Mobley (2015), inverted

Beam attenuation c = a + b, 1/m, 400–700 nm.

Question 4

Does the Monte Carlo give back HydroLight's Kd?

Question 3 rests on a Monte Carlo written for it. Solonenko & Mobley printed, beside their a and b, the Kd that HydroLight, the code most of the field uses, made of them. Given the same a and b, does the Monte Carlo give back that Kd?

Found
  • At 143 cells from 350 to 700 nm, the Monte Carlo's Kd is a median 0.970 of HydroLight's, with 94.4 % of the cells within 10 %. The criterion written down before any run, a median within 0.95 to 1.05 and 90 % of the cells within 10 %, is met.
  • HydroLight's Kd lies at every cell within what the paper's unprinted conditions (the phase function, the sun, the depth) make of the Monte Carlo's.
Not known

Which of the remaining differences are HydroLight's average-particle phase function, its sun and its depth, and which the codes. The paper prints none of the three.

Every figure on this page comes from jerlov : the tables it ships with their sources, and the Monte Carlo runs kept in its repository (examples/data/sm_closure/). The full account of each question, with every figure recomputed by a test, is in its open questions, with the data's history in DATA.md and the design in DECISIONS.md.

Sources: Pope & Fry (1997), Appl. Opt. 36, 8710; Mason, Cone & Fry (2016), Appl. Opt. 55, 7163; Lee et al. (2015), Appl. Opt. 54, 546; Smith & Baker (1981), Appl. Opt. 20, 177; Morel et al. (2007), Limnol. Oceanogr. 52, 217; Petzold (1972), SIO Ref. 72-78; Bogucki et al. (2004), Appl. Opt. 43, 5662; Jerlov, Optical Oceanography (1968) and Marine Optics (1976); Williamson & Hollins (2022), Appl. Opt. 61, 9951; Solonenko & Mobley (2015), Appl. Opt. 54, 5392.