Geopotential-based Multivariate MJO Index: extending RMM-like indices to pre-satellite era

Model simulations suggest that Madden–Julian oscillation (MJO) activity changes under the anthropogenic climate change background. However, satellite observations, which provide information on MJO convection activity, are not available before the 1970s, hindering research on the long-term historical variability of MJO. This study aims at extending the data length of MJO indices that include both MJO circulation and convection features, such as the widely used Real-Multivariate MJO (RMM) Index, to the pre-satellite era. This paper introduces a new MJO Index construction method, in which the outgoing longwave radiation (OLR) input is derived from upper-level geopotential, and names it as the Geopotential-Based Multivariate MJO (GMM) Index. The GMM Index is derived from 1902 to 2008 based on the 20th century reanlaysis product, by assuming that the relationship between OLR and geopotential does not change over time, and is compared with the filtered version of the RMM (FMM) Index during 1981–2008 and historical observed precipitation records in the 20th century. The GMM Index is shown to (1) have the same climatological properties as the FMM Index, (2) be statistically highly correlated to the FMM Index, and (3) be able to indicate MJO activities and MJO’s convection features in the pre-satellite era. The overall bivariate correlation between the FMM and GMM indices based on ERA-20C is 0.964. Evaluation results confirm the validity of the proposed MJO Index construction method, which could capture MJO convection activity in the pre-satellite era and can be applied to all MJO indices that require OLR inputs. This study provides an alternative way that overcomes the difficulty of historical MJO studies and will be beneficial to our understanding of the long-term change of MJO.