Comparison of two filter approaches for characterizing boreal summer intraseasonal oscillation

Abstract

Two commonly used filter approaches for boreal summer intraseasonal oscillation (BSISO) are investigated with the observations during 1979–2021. One is the real-time BSISO index and the other is based on 20–80-day Lanczos bandpass filter. Both two filter approaches are able to isolate spatial structure and life cycle of BSISO realistically in the Indo-NWP region. But the BSISO phase pattern match does not guarantee consistent propagation characteristics and phase occurrence frequency. The real-time monitoring methods such as real-time BSISO index does not use future information beyond the current date, so it is influenced by seasonal mean anomalies induced by El Niño on the interannual timescale over the Indo-NWP region. During post-El Niño summers, the real-time BSISO index shows a weaker northward propagation than bandpass filter approach. Phases 2–4 of BSISO show relatively higher occurrence frequency than other phases during post-El Niño summers in real-time BSISO index, resulting in the seasonal mean of active eight BSISO phases largely resembling the El Niño induced mean state anomaly pattern with depressed convection and low-level anomalous anticyclone in the tropical northwest Pacific region. Bandpass filter approach exhibits a more uniform phase occurrence frequency for both normal and post-El Niño summers, which is consistent with the random nature of BSISO as one of atmospheric internal variability. Bandpass filter approach is recommended for studies of BSISO northward propagation and its relationship with low-frequency variability such as ENSO.