Reading the map. Each map averages ERA5 anomalies (from the 1991–2020 normal, with the long-term trend removed) on days in the chosen MJO phase whose date falls in the window around the chosen day; the tapered window weights days near the centre most, so maps change smoothly as you step. With a lag, the anomalies are taken that many days after each MJO day — what tends to follow the phase.
Stippling marks where the signal is unlikely to be chance, judged by resampling whole years, with a window across New Year counted as one season (consecutive days in one MJO event are not independent). Light grey dots: significant at 95% at that grid point — probably real, though on a map of pure noise about 1 point in 20 would pass by luck. Black dots: also pass a strict map-wide test (false discovery rate 10%) that allows for checking thousands of points at once — robust. N counts days, separate MJO events, and years; the events and years are the honest sample size.
MJO index: Bureau of Meteorology RMM. ENSO: NOAA CPC Oceanic Niño Index. The earlier composite tool remains available.