Understanding Variability of Indian Summer Monsoon Rainfall in Different Time Scales Using Global Coupled and Regional Models

Abstract

The Indian Summer Monsoon (ISM) is a well-organised land-ocean-atmosphere coupled system known as the southwest monsoon. The southwest monsoon season (June-September) in the annual cycle is distinguished from the rest of the seasons in terms of onset, propagation/maintenance and retreat phases. The ISM exhibits significant spatial and temporal variability in monsoon circulation and associated rainfall. Accurate spatial and temporal rainfall information helps to understand the hydrological processes on a country scale. The shortage of observational networks or damage caused by geophysical extreme events limits the actual picture of rainfall characteristics and variability of ISM. The observational network is limited to providing insights into the possible reasons for the characteristic variability of that particular observation. In particular to ISM rainfall (ISMR) variability, the evolution of different ocean and atmospheric conditions in relation to the teleconnections is vital, and this information is not readily available from the observational network. However, satellite products have emerged, delivering data at a different spatiotemporal resolution. The availability of satellite precipitation products is expected to present realistic monsoon rainfall characteristics at different time scales. Further, the climate, seasonal, and regional models can provide numerous ocean-atmosphere variables at different time and spatial scales, which are alternatives to the observational network for understanding ISMR variability. However, these models have shown limited skill in predicting ISM rainfall and its variability. The representation of slowly evolving modes in the tropical ocean, as well as the accompanying ISMR variability, is a tough challenge for operational and research communities. The magnitude, location, and timing of ENSO occurrences influence the performance of the El Niño-Southern Oscillation (ENSO). Several Coupled General Circulation Models (CGCMS) studies mostly focused on the effects of developing El Niño and La Niña on summer rainfall over India. Recent studies have discovered that, like growing ENSO, the decaying stages of ENSO modulate the ISMR. The significance of ENSO decay phases in seasonal forecasting encouraged to study of the different responses of ISMR to distinct El Niño and La Niña decay phases. Previous studies suggested the seasonal mean ISMR forecasts rely on the skillful representation of interannual modes mainly ENSO, Indian Ocean Dipole (IOD) and co-occurrence of both in the tropical ocean and their associated teleconnection patterns. In order to ensure the consistency between large-scale driving fields and model solutions, spectral nudging has been used, particularly in the case of seasonal simulations. Therefore, the present thesis is aimed to demonstrate (i) the performance of global and seasonal models in capturing different phases of the El-Niño/LaNiña and IOD/ENSO and their effects on the simulation of ISMR and its variability across the temporal scales and (ii) the advantage of high-resolution in capturing regional features of ISMR and its variability. The thesis is structured accordingly, beginning from the introduction chapter to the concluding chapter along with four working chapters. The first working chapter investigates the reliability and accuracy of high-resolution satellite rainfall products such as TRMM, GPM, IMR, and HEM are compared against rain gauge observation of IMD. GPM has captured the all-India seasonal rainfall (6.8 mm day –1) close to the IMD rainfall (6.7 mm day –1). While the seasonal all-India rainfall is overestimated in TRMM (7.6 mm day –1), HEM (8.4 mm day –1), and IMR (14.2 mm day –1) products. The GPM has reasonably captured All-India rainfall at daily, monthly, and seasonal scales than the others. The second working chapter is focused on understanding the variations in the El Niño decay phases and their diverse impacts on ISM rainfall using IMD observation and CMIP6 model outputs. The El Niño/La Niña decays are classified into three categories based on their rapid and/or slow transition with respect to the following boreal summer season. When El Niño early-decay, El Niño mid-decay, and La Niña no-decay occurrences occur in the summer, widespread rainfall occurs throughout India; when El Niño early-decay, La Niña mid-decay, and La Niña no-decay occur, the rainfall is suppressed. The MMM of top ~ 20% of CMIP6 models could capture the ISM rainfall anomaly patterns in the summer of El Niño early-decay, La Niña early-decay and El Niño mid-decay; however, it failed in the other three phases (La Niña early-decay, La Niña mid-decay, and El Niño no-decay in summer) because of the unrealistic representation of IOD and TIO SST anomaly. It is noted that most of the models have a tendency to show strong IOD during ENSO decay summers, which may influence the large-scale circulation features and regional rainfall anomalies. The seasonal mean ISM rainfall forecasts directly depend on the representation of large-scale features associated with ENSO developing as well as the ENSO decaying phases and their associated responses. The third working chapter addresses the credibility of high resolution in capturing interannual modes such as ENSO, IOD and their co-occurrences in relation to ISM rainfall. For this purpose, the high-resolution (38 km grid-spacing) seasonal model (CFSv2) and coarser resolution (~100 km grid-spacing) ensemble means of North American MultiModel Ensemble (NMME) hindcasts are evaluated against IMD observations. The overall mean AISMR (standard deviation) in CFSv2 is 6.3 mm day-1 (0.42 mm day-1), which shows good agreement with an observational value of 6.9 mm day-1 (0.59 mm day-1). The standard deviation (0.25 mm day-1) is inadequately represented despite the close representation of MMM (6.1 mm day-1) with observation. Even when SST anomalies are positioned precisely, the ISM rainfall patterns show significant changes in spatial and temporal variability. Accurate SST patterns during the Pure El Niño (P-EN), Pure La Niña (P-LN), Pure Positive IOD (P-pIOD), Pure Negative IOD (P-nIOD), Co-occurrence of El Niño with positive IOD (Co_EN_pIOD), and Co-occurrence of La Niña with negative IOD (Co_LN_nIOD) must be captured for better representation of circulations needed for the better prediction of the different categories of extremes. The fourth working chapter examines the performance of spectral nudging in the dynamical downscaling of ISMR in the WRF model framework. Two experiments were conducted i) simulation using spectral nudging (SN), in which the zonal, meridional and temperature fields are nudged, and ii) simulation with no nudging (NSN). The spectral nudging accurately simulated ISMR characteristics at seasonal, sub-seasonal, and across homogenous regions. The area-averaged RMSE is less for SN in all the homogeneous regions compared to NSN. Significant improvements are seen in the first 60 days’ simulation of SN over India and its homogeneous regions. At the same time, the NSN experiment showed relatively poor performance. Thus, the study highlights the importance of spectral nudging to high-resolution downscaling for improved long-range simulations. All the conclusions are briefed in the summary and conclusion chapter. The overall outcomes of the present dissertation are (i) better satellite product capturing realistic ISM rainfall characteristics and variability, (ii) insights on the unrevealing processes associated with ENSO decay phases that influence the interannual variability of ISMR (iii) the necessity of high-resolution for improved ISM rainfall variability as studied from seasonal as well as mesoscale WRF model. These efforts are intended to help state-of-the-art dynamical models identify the systematic biases and their removal for realistic representations of ISM rainfall and variability.