The identification of PeV gamma rays over the cosmic ray background requires an excellent discrimination capability, usually achieved through the measurement of the shower muon content. However, the deconvolution of secondary muons from the electromagnetic shower component is a difficult task that substantially increases the cost of the experiment.
In this work, we propose a novel approach based on the analysis of the azimuthal fluctuations of the shower footprint at the ground measured by a low-density array of water Cherenkov detector units. Using a large statistical simulation sample, we demonstrate that it is possible to reach a gamma/hadron discrimination as high as the one achieved by counting muons.