What is a "double-boxcar filter"?

Question

I am smoothing data according to a research paper, and it says they apply a "double-boxcar" filter of width X". I know how to boxcar filter in python, i.e signal.boxcar and signal.convolve, but I'm not sure how to interpret this statement.

Do you guys think they are doubling box width or are they just smoothing twice over?

Answer 1

It probably means they apply the boxcar filter twice. The Central Limit Theorem tells us that repeated box convolutions will converge to a Gaussian convolution [1]. 2 passes is equivalent to a "tent" filter; usually 3 passes are used for a nice smooth approximation.

[1] http://nghiaho.com/?p=1159