The effect of broadband radio frequency interference(RFI) is often overlooked in the terrestrial communication literature. However, such an RFI can render very poor equalization for a reception over a frequency selective channel by making the channel estimator perform the worst. Accordingly, the paper firstly proposes an efficient matrix-based semi-blind channel estimator (MB-SCE) for single-input multiple-output(SIMO) systems suffering from severe broadband RFI. Secondly, more complex yet more performant tensor-based semi-blind channel estimator (TB-SCE) is proposed while showcasing a trade-off between performance and complexity. The paper then compares MB-SCE and TB-SCE with the existing matrix-based blind channel estimator (MB-BCE) and tensor-based blind channel estimator (TB-BCE), respectively, that ignore RFI. Eventually, Matlab Monte-Carlo simulations have corroborated that MBSCE and TB-SCE outperform MB-BCE and TB-BCE, respectively. In addition, TB-SCE outperforms MB-SCE especially for a short observation interval.