Repeated K-means clustering to create a tree diagram

I have several time-series and I want to use the k-means function for a cluster analysis (distance = correlation).

1st step: I want to partition my sample into 2 clusters

2nd step: I want to partition each of my 2 clusters from step 1 into another 2 clusters

3rd step: I want to partition my 4 clusters from step 2 into another 8 clusters

4th step: ...

I’m doing this to create a tree diagram. Has anyone an idea how to realize this? My first idea failed due to the iterative property of the function.

Thank You!!!

[Christian Maschner]

I found a way:

input = linkage(data','centroid','correlation')
figure()
dendrogram(input,'orientation','left')