1. ## N(t) = γA(t)

This should be easy if you know what to do. I just don't.

N(t) = γA(t)

N(t) is a known number (of peatlands) at time t, A(t) is a known area (deglaciated) at time t. γ is an unknown number (of peatlands that can form per unit of deglaciated area). How do I decide what is the one best fit value for γ for a range of N's and A's? This needs some kind of computer programme to calculate it?

2. Originally Posted by etsija
This should be easy if you know what to do. I just don't.

N(t) = γA(t)

N(t) is a known number (of peatlands) at time t, A(t) is a known area (deglaciated) at time t. γ is an unknown number (of peatlands that can form per unit of deglaciated area). How do I decide what is the one best fit value for γ for a range of N's and A's? This needs some kind of computer programme to calculate it?
So:

$\displaystyle y=\frac{N(t)}{A(t)}$

so compute the average of the ratio of $\displaystyle N(t)$ to $\displaystyle A(t)$ for the data you have and use that.

If you know a bit about the errors in the known $\displaystyle N$'s and $\displaystyle A$'s you could do a bit more.

RonL

3. Average, that makes sense. Thanks a lot. The article I'm reading (Gorham et al 07) says about their dataset "No value of y allowed this simplest model to fit the data adequately". That made me think I should somehow try a number of values to find the fittest, but perhaps it just means different average values like mean and median?

Anyway, it seems to work now. I calculated the mean, used it for a simple model (mean*A(t)), correlated that with the actual data N(t) and got a realistic R2.