Comment on “Unraveling the Mystery of Indian Summer Monsoon Prediction: Improved Estimate of Predictability Limit” by Saha et al.

Abstract

In a recent study, Saha et al. (2019, https://doi.org/10.1029/2018JD030082) examine the correlation between boreal summer seasonal mean rainfall over India and rainfall variance as a function of subseasonal time scale and find that the correlation has a local maximum (exceeding a value of 0.6) for synoptic time scales (2-5 day periods). They claim these results to be a major advancement in understanding monsoon predictability but do not provide a clear physical explanation. Here we examine the sensitivity of this relationship to the details of the analysis and only consider the observed correlation identified by Saha et al. (2019, https://doi.org/10.1029/2018JD030082). There is large sensitivity in the correlation maximum between spatially averaged seasonal mean rainfall and synoptic scale rainfall variance averaged over the same domain. The correlation maximum is weaker over the longer period of 1901-2015, and more notably it is highly sensitive to the domain of prior averaging. A correlation peak is not found outside central India and is neither found within central India when averaging over a smaller domain. Averaging over a larger domain results in a disproportionate reduction in synoptic variance (that is of most interest). The peak in correlation between seasonal mean and 2-5 day variance only emerges after first averaging over a sufficiently large region; thus, the maximum appears to be an artifact of spatial averaging. It is further pointed out that a positive mean-variance relationship is an intrinsic property of rainfall, and thus, its existence alone does not necessarily imply any physical, causal, and/or predictive connection between time scales.

Role of Microphysics and Convective Autoconversion for the Better Simulation of Tropical Intraseasonal Oscillations (MISO and MJO)

“…The same methodology was also used in Swenson et al. (2020) study for similar analysis. The rainfall is spatially averaged (land region only) over central India (74°E–83°E, 15°N–25°N).…”

Role of mean, variability and teleconnection of clouds behind Indian summer monsoon rainfall

“…NOAA interpolated Outgoing Longwave Radiation (OLR) data are also used in the study (Liebmann & Smith, 1996). To find the relationship between the subseasonal variance (cumulative) of cloud ice with the ISMR and ENSO, we adapted the method from Saha et al (2019, 2020, 2021) which was also validated by Swenson et al (2020). The seasonal (June–September average) mean ISMR and Niño3.4 SST anomaly correlated with the cumulative subseasonal variance of cloud ice at various time bands.…”

Reply to Comment by E. T. Swenson, D. Das, and J. Shukla on “Unraveling the Mystery of Indian Summer Monsoon Prediction: Improved Estimate of Predictability Limit”

“…Swenson et al (2020) (hereafter SN20) have chosen to raise several questions on the technical aspects of the part of Saha et al (2019) (hereafter SA19) that deal with correlations between the amplitude of subseasonal fluctuations of rainfall over the region and seasonal Indian summer monsoon rainfall (ISMR). SN20 have used a similar method as in SA19 (section 2) to reconstruct subseasonal variance at various bands.…”