Previous studies showed that atmospheric feedbacks play an important role in modulating sea surface temperature (SST) during ENSO. Based on kernel methods, we further decompose radiative feedbacks into contributions from water vapor, air temperature, surface temperature and clouds. Besides by using a newly developed set of radiation kernels, we can get radiative feedbacks not only from top pf atmosphere (TOA) but also from surface. In term of TOA, our results show that longwave (LW) and shortwave (SW) radiation feedback compensate with each other and thus result in a near 0 net feedback. For surface, negative SW feedback cause by cloud and negative latent heat (LH) feedback dominate, which help to damp SST warming during El nino. In the atmosphere, both radiative and dynamical feedbacks are important. Cloud LW feedbacks are strong positive while dynamical feedback caused by convergence of moist static energy (MSE) in the atmosphere is negative, which indicate that there is more poleward energy transport during El nino years. Qualitatively our result is in accord with previous results. However in terms of water vapor feedback and cloud LW feedbacks, there is a quantitative discrepancy between our results and previous results, in which cloud feedbacks are quantified by cloud radiative forcing (CRF) method.