That's quite a bold assertion, my friend. I have to say that Python's superiority for that particular field is not obvious to me, if such superiority even exists. I'd be very interested to read how you know this to be true.
First of all, I was sloppy with terminology. I shouldn't have said "scientific computing" because that has a more restricted meaning than what I intended to refer to. I intended to refer to any domain where computers are extensively involved in gathering empirical knowledge, which is basically all over the place. The intended meaning was what scientific computing really means as well as statistics, machine learning, etc.
Anyway, no other very high-level language has anything really comparable to all the advantages offered by numpy, scipy, scikit-learn, TensorFlow, Sage etc. in this area. There's still a lot of stuff I would only do in R (as much as I really don't want to) and Matlab is still pretty widely used, though I don't know much about it. A lot of big data stuff appears to be done in Java and fellow JVM language Scala. Raw C, C++ and Fortran are still relevant if you have a real need for speed. But in many respects Python has become the de facto standard for these things when special requirements don't need to be met. For example, I'm enrolled in the Coursera course Data-Driven Astronomy and it's all in Python. This is equally true of a number of other courses on said website. Results from Google are also indicative of Python's preeminence in data science / machine learning, more than I expected really. In academics, Python is replacing other languages for introductory programming courses and has replaced Common Lisp in the leading AI textbook Artificial Intelligence: A Modern Approach. And so it goes.