The colour of the light at depth

Sunlight loses its red within the first few metres of any water. What it keeps after that depends on the water: clear ocean water passes blue best, and coastal water, with more dissolved and suspended matter, passes green. This guide follows the downwelling light down through four Jerlov types and shows how the share of blue, green and red changes.

The light at the surface

The calculation needs a spectrum to start from. Here it is daylight with the shape of CIE illuminant D65, which the package ships for colour work. That is an assumption, not a measurement: the real spectrum depends on the sun’s height, the sky and the surface, and a measured one should replace it when there is one.

>>> import numpy as np
>>> import jerlov
>>> wl = np.arange(400.0, 701.0, 1.0)
>>> d65_wl, d65 = jerlov.d65()
>>> surface = np.interp(wl, d65_wl, d65)

The shares of three bands, blue 400-500 nm, green 500-600 nm and red 600-700 nm, of the light at each depth:

>>> def shares(ed):
...     bands = [(wl >= 400) & (wl < 500), (wl >= 500) & (wl < 600), wl >= 600]
...     return [ed[band].sum() / ed.sum() for band in bands]

Down through one water type

The diffuse attenuation coefficient Kd sets how fast the downwelling irradiance falls: Ed(z) = Ed(0) exp(-Kd z) at each wavelength. Jerlov’s own Kd, from his 1976 book, is source="jerlov1976":

>>> for water_type in ("I", "III", "1C", "5C"):
...     kd = jerlov.water(water_type, source="jerlov1976").kd(wl)
...     for depth in (0.0, 5.0, 20.0):
...         blue, green, red = shares(surface * np.exp(-kd * depth))
...         print(f"{water_type:>3} {depth:4.0f} m   blue {blue:.2f}"
...               f"   green {green:.2f}   red {red:.2f}")
  I    0 m   blue 0.36   green 0.35   red 0.29
  I    5 m   blue 0.53   green 0.39   red 0.08
  I   20 m   blue 0.70   green 0.30   red 0.00
III    0 m   blue 0.36   green 0.35   red 0.29
III    5 m   blue 0.47   green 0.45   red 0.09
III   20 m   blue 0.49   green 0.50   red 0.00
 1C    0 m   blue 0.36   green 0.35   red 0.29
 1C    5 m   blue 0.35   green 0.54   red 0.11
 1C   20 m   blue 0.23   green 0.77   red 0.00
 5C    0 m   blue 0.36   green 0.35   red 0.29
 5C    5 m   blue 0.23   green 0.60   red 0.17
 5C   20 m   blue 0.07   green 0.90   red 0.03

The red has all but gone by 20 m in every type. In the clearest oceanic water, Jerlov I, what is left is mostly blue; in the coastal types it is mostly green, and more so the more turbid the water.

The water changes on the way down

Jerlov classified the water by its top 10 m, and the water below is typically different: turbid coastal water clears with depth, and the clearest oceanic water turns a little less clear. Treating 1C as 1C all the way to 45 m is therefore a choice. jerlov.descend() changes the type every 10 m, following the types Williamson & Hollins (2023) found typical at each depth across more than 2500 measurement campaigns, with Jerlov’s Kd for each, and stops where those profiles stop:

>>> descent = jerlov.descend("1C", 45.0, wl)
>>> for top, bottom, water_type in descent.layers:
...     print(f"{top:4.0f} to {bottom:4.0f} m: {water_type}")
   0 to   10 m: 1C
  10 to   20 m: 1C
  20 to   30 m: III
  30 to   40 m: III
  40 to   45 m: II

The light at 45 m, both ways:

>>> kd_1c = jerlov.water("1C", source="jerlov1976").kd(wl)
>>> for name, ed in (("1C all the way", surface * np.exp(-kd_1c * 45.0)),
...                  ("descend", surface * descent.transmittance)):
...     blue, green, red = shares(ed)
...     print(f"{name:>15}   blue {blue:.2f}   green {green:.2f}   red {red:.2f}")
 1C all the way   blue 0.12   green 0.88   red 0.00
        descend   blue 0.26   green 0.74   red 0.00

The clearer water below lets more blue through, so the light at depth is less green than a single type would say.

Where to go next