Model Reduction of a Noisy 2D Experimental Data

I am trying to understand the following MATLAB code:

    Reduce(Reduce(u))

where u is a n×m matrix and is a 2D measurement from an experiment which contains noise. Does Reduce take into account the uncertainty in u? Would someone kindly help me have a better understanding of what the above mentioned code is doing?