Electromagnetic scattering bounds on subwavelength structures play an important role in estimating performance of antennas, radio frequency identification tags, and other wireless communication devices. An appealing approach to increase a scattering cross section is accommodating several spectrally overlapping resonances within a structure. However, numerous fundamental and practical restrictions have been found and led to the formulation of Chu-Harrington, Geyi, and other limits, which provide an upper bound to scattering efficiencies. Here we introduce a two-dimensional array of near-field coupled split-ring resonators and optimize its scattering performance with the aid of a genetic algorithm operating in 19-dimensional space. Experimental realization of the device is demonstrated to surpass the theoretical single-channel limit by a factor of >2, motivating the development of tighter bounds of scattering performance. A superradiant criterion is suggested to compare maximal scattering cross sections with the single-channel dipolar limit multiplied by the number of elements within the array. This empirical criterion, which aims to address performance of subwavelength arrays formed by near-field coupled elements, is found to be rather accurate in application to the superscatterer, reported here. Furthermore, the superradiant bound is empirically verified with a Monte Carlo simulation, collecting statistics on scattering cross sections of a large set of randomly distributed dipoles. The demonstrated flat superscatterer can find use as a passive electromagnetic beacon, making miniature airborne and terrestrial targets radar visible.