Example

Ocean Colour

Simulation of ocean colour observed remotely from aircraft or satellites

Observations

1. Satellite images of ocean colour correlate well with samples of phytoplankton pigment concentration in the mixed layer. Remotely sensed ocean colour is therefore a useful surrogate for ship measurements.
2. Ocean colour images exhibit mesoscale ( order 10 km) patchiness, with an amplitude equivalent to a factor of ten in phytoplankton biomass.

The challenge

1. To predict ocean colour from first principles without empirical calibration against ship samples of phytoplankton pigment. i.e. to use virtual ecology.
2. To test the model used to predict ocean colour by comparing the annual cycle of ocean colour in observations and predictions.
3. To develop a method for relating predictions of ocean colour from the one-dimensional virtual ecosystem derived from external forcing averaged over 100km to the patchy observations of ocean colour averaged over 100km.

Optical Monte Carlo modelling

Ocean colour is diagnosed from a virtual ecosystem by the method of optical Monte Carlo modelling. The solar radiation in each of 64 wavebands in the range 400 - 700nm is represented by five million photons. Each photon approaches the Earth from the Sun, passes through the atmosphere, interacts with the waves on the sea surface, enters the ocean, and interacts with the seawater and plankton. At three stages (clouds,waves and plankton) the interactions are computed as random samples according to prescribed probability distributions. The complete radiation field is computed by summing all five million photons, each treated in that way. Most of the photons are absorbed in the seawater and plankton, but a tiny fraction are scattered back in the direction of the satellite. They contribute to the received light in this waveband. Repeating for all 64 wavebands produces the spectrum of light received by the satellite after scattering by the plankton. Matching that spectrum to the sensors on the satellite produces the value of ocean colour.

Comparison with empirical methods

This procedure predicts ocean colour from the emergent plankton profile, which depends on the primitive equations in the VE model. It does not depend on empirical calibration against ship observations. Comparing the results with empirical formula reveals errors in the latter, which are attributed to neglecting the contribution of plankton below the mixed layer.