GPS spoofing attacks consist of precisely crafted false signals that exhibit highly similar time and spectral characteristics to genuine signals. Spoofing detection methods utilizing array antennas exploit spatial feature differences to detect spoofing interference, while coprime arrays can obtain broader spatial information, providing potential solutions for spoofing detection in underdetermined scenarios. The spoofer can successfully take over the target receiver covertly by adjusting the transmission power to slightly exceed that of the genuine signal, which poses a significant challenge for detecting spoofing attacks. To address the challenge of establishing power detection thresholds under low-power spoofing, a novel GPS spoofing detection algorithm using coprime array antennas is proposed. The proposed algorithm fully utilizes space-time domain characteristics of incident signal sources. Specifically, angle-of-arrival (AOA) estimation is employed to perform spatial separation of the received signals. Then, virtual domain equivalent power comparison is combined with temporal cross-correlation function peak monitoring, resulting in reliable detection of low-power spoofing. Numerical simulations have verified the effectiveness of the proposed algorithm in multiple scenarios.