# Correlation length

#### Evyatar

##### New Member
Hi all,

I'm working on a hydrological model.
I supposed to build a mesh with correlation length in the permeability field. I know what is the meaning of the term 'Correlation length', It says that the permeability values that are near each other are relate and there is no drastic change in the values between them.
But what i don't know is what are the statistic principle beyond correlation length.
Does anyone can help me?
Evyatar

#### hlsmith

##### Less is more. Stay pure. Stay poor.
By correlation length, do you me the correlation coefficient?

#### Miner

##### TS Contributor
From the context, I think the OP may mean autocorrelation and the number of significant lags.

#### hlsmith

##### Less is more. Stay pure. Stay poor.
You are a clever one if you are right - Miner!

#### Evyatar

##### New Member
Hi,

First of all thanks for the quick response.

I think that the term autocorrelation is more accurate for what i'm looking for. But i'd like if Miner can tell me also about the correlation coefficent.

Evyatar

#### Miner

##### TS Contributor
I am guessing here because you have provided very little context on which to proceed. Possibly this information on spatial analysis might help. Spatial analysis looks at correlations within geographical areas.

#### Evyatar

##### New Member
It's a big problem because i don't even know how to explain to myself and therefore not to you. But i will try.
From what i understood auto-correlation is for checking the relationship between points, and if so it is not what i need. I am ''smoothing'' the values, and instead of getting a completely random distribution (value of 10^-13 can be near value of 10^-17) i'm getting a random one but with some connection between them (near 10^-13 will be 3*10^-13 or something like that).
I really hope i explain my self properly now.
Again Thanks a lot for trying.
Best regards,
Evyatar.

#### rogojel

##### TS Contributor
Yes, that would be the autocorrelation as Miner said. It basically means that close to a high value you are likely to get higher values and close to a low value other low values. The lag would roughly mean, how far the influence of a high (low) value is extended . if you look at the plot of a one-dimensional strongly autocorrelated variable it very much looks like a chain - it has a kind of inertia,

regards