Superposed multicarrier transmission is one of the schemes that achieve efficient use of frequency resources. In the scheme, desired signals are interfered by other systems sharing the same frequency band, which causes unreliable LLR (log-likelihood ratio) setting and leads to degradation of the BER (bit error rate) performance. In many conventional schemes for OFDM (orthogonal frequency division multiplexing) systems, the interference power for each subcarrier is estimated independently. Otherwise, based on the assumption that interferences have the same average power, one interference power is estimated and applied to all subcarriers. However, when there exist multiple wideband interferences with different average power, it is better to detect each interference and estimate each power based on all subcarriers within each superposed band to make it accurate. In this paper, we propose a scheme to suppress multiple wideband interferences based on the probability with which each subcarrier is superposed, referred to as the superposition probability. In the proposed scheme, we derive the PDFs (probability density functions) of residual powers for superposed and non-superposed bands. Based on residual powers and their PDFs, we calculate the superposition probability for each subcarrier. Furthermore, we perform superposed band detection iteratively according to the probabilities and update each estimated interference power. Through simulations, we show that the detection rate of superposed bands is improved by iterating superposed band detection based on superposition probabilities. We also show that the proposed scheme achieves the better BER performance than that of conventional ones.