Independent component analysis (ICA) has been used to identify and remove confounding factors in fMRI data. While spatial ICA (sICA) is more typical for fMRI, temporal ICA (tICA) can better distinguish between temporally independent but spatially overlapping components compared to sICA. We investigate if tICA and sICA perform differently in identifying different physiological components of the resting-state fMRI (rs-fMRI) signal. Our results show that noise sources with widespread effects across brain (e.g. PETCO2, FD, DVARS) are better identified by tICA. However, sICA performed significantly better than tICA in removing the effects of affine head motion parameters.