Variability of the Madden-Julian Oscillation (MJO) inferred from tropical in-situ observations during 1940-2023 is examined, and compared to that derived from reanalyses. MJO assessments are challenging because characterizing MJO behavior outside of the satellite era suffers from a lack of detailed information about its observed state, remote signals and often relies on climate models’ typically poor MJO representation. This study shows that, while tropical soundings are spatially and temporally sparse, they are still useful for better understanding multi-decadal MJO amplitude variability. The similarities between variability of the MJO derived from (imperfect) reanalysis products and (sparse) tropical station observations suggest that decadal changes during the last 60 years are physical and not necessarily related to changes in the observational network. The larger differences in MJO amplitude between station and reanalysis prior to 1960 suggest that sparse tropical observations are a barrier to fully characterizing long-term MJO evolution during the 20th century.
