“…As a result, the subdivision of the filtering boxes into two categories excluding the spectral peak region may cause overfiltering. However, Roundy (2012a) shows that based on the linear regression for signals along the Kelvin wave dispersion curves at zonal wavenumber 4, for equivalent depths of 90, 25, 12, 8, and 5 m, the wave structures continuously change with the equivalent depth. Moreover, temporal correlations of the rainfall anomaly between faster and slower components averaged over the warm pool are 0.056, 0.016, 0.011, and 0.002 for Kelvin, EIGn0, WIGn1, and WIGn2 waves, respectively.…”