We usually measure the fractal dimension of architecture (I have never measured a river system) using our own software we are developing, but we have used "Benoit", it is a commercial software but not particularly expensive. We have used it to measure architecture and coastlines and it seems quite accurate.

http://www.trusoft-international.com/benoit.html

If you have line drawings of the river network you could use it.

I saved the line images as bitmaps and then inverted them and then analysed them using the box-counting option. (other fractal analysis options are available). We have a paper coming up at the next ANZASCA conference which outlines the method for preparing drawings for fractal analysis (using the box-counting method) for maximum reliability of results. If you are interested I can upload it to research-gate once in has been published.

I've used, during my works, Benoit.

Thank you very much for your support!

Hi, Ion. I was just wondering if you've tried the local connected fractal dimension in FracLac for this work?

See also the paper on Tree-ordering , which I have uploaded today. This is a good-oldy from 1975. A more recent reference is Uylings, H.B.M. and Van Pelt, J. (2002) Measures for quantifying dendritic arborizations. Network: Computation in Neural Systems. 13:397-414. This paper is also available via ResearchGate.

Thank you very much for your support!

Hi Ion,

sorry for the late response. I think R does a fine job. Its just a little tricky. You need R 2.10.1 and the package fdim. It uses box counting and you can use the forecast package which includes the FARIMA model that produces the H parameter. I would suggest reading: Statistics for Long Memory Processes by Beran. I saw that you downloaded my paper and most of my work is based on Beran. He even has his own R package.

Thank you. Fractal Gography by Andre Dauphine it is very intersting and usefull for me! 

Maybe you can try "fraclab".