Convectively coupled equatorial waves (CCEWs) are often identified by space-time filtering techniques via a fast-Fourier transformation (FFT) that make use of the eigenvalues (frequency and zonal wavenumber) derived from the linear shallow water theory. Here, instead, a method is presented for identifying CCEWs by using a combined FFT and empirical orthogonal function (EOF). We show that this technique is better at isolating CCEW’s signals from noises and undesirable spectral mixtures among the modes. In particular, using lag-regression analysis, the structures associated with each eigenvector signal resemble equatorial wave features consistent with a linear wave theory. The first eight EOFs of the Kelvin-filtered outgoing long-wave radiation (OLR) at the equator represent Kelvin waves with zonal wavenumbers 2, 3, 4, and 5, respectively. The horizontal structures of MRG (n=0) and ER (n=1) waves are well isolated by only the first two EOFs, while the higher EOF modes capture spectral mixtures among the wave modes. On the other hand, the first ten EOF modes of Tropical Depression (TD)-type-filtered OLR anomalies represent TD-type wave activities across different regions; where the first four modes indicate TD-type wave activity over the South East China, while the modes of 5-6 and 9-10 indicate the TD-type wave activity over Africa and Central America, respectively. This study highlights the importance of the combined space-time FFT-EOFs analysis to better capture the horizontal structures of CCEWs that occur across a range of spatial scales.