Power law for analysing and predicting best rowing times

[Nomath]

I am well aware of the relationship between velocity and the power needed to overcome wind resistance. ...

This is not wind resistance, but air resistance!! In the absence of wind, the power for air drag is proportional to v³.
The fact that as a runner you feel the wind direction means that the air drag power is substantial. Maybe running legs cannot profit as much as wheels and gears shifting can from a tailwind. Moreover, a running body has a much bigger frontal area than a cyclist bent over the handlebar in an aerodynamic position. As I said before, I am not familiar with the power analysis of running, but the air drag in running should not be discounted.

The exponents in the power law that I evaluated from the cycling ergometer data (1.04-1.06) are of course in a stationary position, hence without air drag.

I will come back to the CP issues after I have studied the Rowing Analytics article.

[Allan Olesen]

You are not telling me anything about running power, which I don't already know. I might add that I am an engineer too.

The bottom line is that power usage for running is roughly proportional to velocity. Yes, it has a non-linear component, which makes it less than perfectly linear, but in the context discussed here - empiric methods for predicting output of other durations - the impact of this nonlinearity is quite small. I could try to show you this with calculation examples, but I am not really interested in using so much time on this, and I assume that you are capable of making those calculations yourself.

Feel free to read the rowing article on CP, but as I said, I haven't read it. So if the article has errors, I can't comment on them.

Edit:
I am out of this discussion. I came here to give some input, which you might find useful. It has become clear to me that your goal is a longpissing competition, and I have taken part in too many of those.

[Nomath]

I am sorry that you take offense to my replies.

I am not interested in modelling of running, except when it gives credibility to the usefulness of the power law in time or speed predictions.
There is a recent free-access research paper about modelling endurance running which states as a Conclusion:
"... it is likely that logarithmic and power-law models are the most accurate models for the predictions of long-distance performances (maximal running speeds for 30 and 60 min or marathon) by extrapolation."

I feel encouraged to explore what the power law can do in rowing.

[jamesg]

The only FISA race is 2000m, so there doesn't seem to be much point in forecasting other distances.

Perhaps two tests such as 1 and 5k or ½/2k and a Watt/log m graph or even just the Watt ratio could be enough for you?

That's what I'd use if I thought it might help to understand what training I need next for racing, if I didn't trust the Interactives and suchlike.

Rowing has a specific aspect, that we have to train both the stroke itself and also endurance enough to use it over the distance at the necessary rating. The relationship between two times will be of little help with the first.