There's no JVM overhead like for Spark computation. The dask array methods use the numpy C-API, which are implemented in C and run on the physical machine.

I think you might have misunderstood my comment; I was referring to the bullet point under "What didn't work":

> Reduction speed: The computation of normalized temperature, z, took a surprisingly long time. I’d like to look into what is holding up that computation.

Apart from rendering static html plots or plots with client-side JS callbacks, you could look into using the new bokeh server: http://bokeh.pydata.org/en/latest/docs/user_guide/server.htm...

It allows for building streaming visualizations or plots with using websockets (implemented using tornado).

That's super cool.

Yes - there's the ability to generate the raw html of a plot or a JS script and div that can be embedded in an HTML doc.

Source: http://bokeh.pydata.org/en/latest/docs/user_guide/embed.html

Also, check out http://colour-science.org/

It has python implementations of a ton of colorspaces and transforms. Also some cool support for spectral data.