Natural Science Forum / Earth Science / Oceanography / November 2003

I think you are completely off track.
Some time ago I designed and wrote a program for the optical
modelling of car lights.
Here one has a source which emits a light flux, these are
represented by 3D vectors in space.
When these collide with a reflective surface, a part is
reflected (angle of incidence is angle of reflection specular
to the surface normal), and a part is absorbed.
Transparent materials follow Snell's law which is dependent
on the index of refraction of the material.
Hence light is considered to travel in a straight line along a
vector untill it hits something and then is transformed in
a vector with other intensity or is split in two vectors
(refraction and reflection in transparent body).
The technique is also called "ray-tracing".
Fluid simulators wether gas (weather) or liquid (seas,
streams, etc. are treated as potentials like pressure.
These are often simulated with finite elements in streams.
Thus a high pressure causes particles to flow to low pressure
zones.
A water particle doesn't zoom along a vector from high to
low, it is displaced into the adjacent cell of finite elements
which in its turn does the same thing until mass is transformed
to a low pressure zone, hence until there is no longer a pressure
differential. If there is a source like a mountain source
of course the game goes on until equilibrium is reached.
These problems are often solved with differential
equations.
If you reason a bit logically you can ponder on the following
example: A river flows downhill from a mountain spring
to a lake. This is the difference in potential.
It encounters a waterfall which is a break in the steady
slope of the river. A waterfall will form, which is not just
dropping down but follows the laws of a falling body.
***Did you ever see a light waterfall?***
No, you did not, because the mechanisms are quite
different.
Here you will have to contend with gravity,
viscosity etc.

If one one wants to write a numerical simulator
one needs the following skills:

Be a far above average technical programmer
consider Fortran because you can buy math libraries.

Be cognizant with the physical phenomena you are
dealing with. With all respect, from your question,
you need to spruce up your physics.

Be a fairly good mathematician in order to transform
physical phenomena into a mathematical model.

Good luck.

Joe Rat  8:o)