2021/06/10 by Devabrat Sharma, Santu Das, Sharma, Devabrat +3
Environmental Science · Physics and Astronomy · #Climate variability and models #Hydrological Forecasting Using AI #Hydrology and Drought Analysis #physics.ao-ph
paper · pdf · doi:10.48550/arxiv.2106.05543
44 pages, 6 figures, 11 extended figures, 1 table
arxiv created 2021/06/11 · arxiv updated 2021/06/14
Scientific basis for long-lead seasonal prediction of Indian summer monsoon rainfall (ISMR) critical for water resource and crop strategy planning is lacking. Using a new predictor discovery method, here we show that the depth of 20 degree isotherm (D20) is least influenced by atmospheric noise and that the 18-month lead forecasts of ISMR have high potential skill (r = 0.86). The high potential predictability is due to smaller initial errors associated with the 18-month lead initial conditions and their slow growth associated with the El Nino and Southern Oscillation (ENSO). The potential skill arises not only from the correlation between ISMR and large-scale slowly varying D20 but also contributed significantly by that with the interannual small-scale D20 anomalies indicating a seminal role of the nonlinearity on the potential predictability. It is, therefore, imperative that a nonlinear predictor discovery as well as nonlinear prediction model is essential for realizing this potential predictability.