Investigating the hemispheric asymmetry in global ocean warming and the links between sea surface salinity and Australian precipitation

Ocean temperature and salinity are essential variables for understanding variability and changes in the coupled ocean and atmosphere system. By tracking changes in temperature and salinity, one can assess the changes in the Earth's energy budget and changes in the global and regional hydrological cycle. The first two research chapters of this thesis investigates the role of internal climate variability in the hemispheric asymmetry of global ocean heat content (OHC) change and derives a fingerprint of this internal climate variability that is responsible for the asymmetry. Hemispheric asymmetry in OHC change corresponds to the high heat gain by the southern hemisphere and reduced heat gain by the northern hemisphere. The next two chapters of this thesis investigates the links between sea surface salinity (SSS) and Australian rainfall during El Ni√±o Southern Oscillation (ENSO) and Indian Ocean Dipole (IOD) events, and the prediction of rainfall over northeastern Australia using SSS. The second chapter of this thesis investigates the hemispheric asymmetry in which 90% of the net global ocean heat gain during 2005‚-2015 was confined to the southern hemisphere with little corresponding heat gain in the northern hemisphere ocean. We propose that this heating pattern of the ocean is driven by anthropogenic climate change and an asymmetric climate variation between the two hemispheres. This asymmetric climate variation is found in the pre-industrial control simulations from 11 climate models. While both layers (0‚-700 m and 700‚-2000 m) experience steady anthropogenic warming, the 0‚-700 m layer experiences large internal variability, which primarily drives the observed hemispheric asymmetry of global ocean heat gain in the 0‚-2000 m layer. We infer that the rate of global ocean warming is consistent with the climate simulations for this period. This chapter reveals the observed hemispheric asymmetry in heat gain can be explained by this asymmetric mode in Earth's internal climate variability without invoking alternate hypotheses, such as asymmetric aerosol loading in the atmosphere. The third chapter shows that this observed asymmetric mode is independent of other known modes of internal climate variability such as ENSO, Pacific Decadal Oscillation (PDO), etc. Composite events (102 events) from the pre-industrial control simulations of 20 CMIP5 models are consistent and well correlated with the observed asymmetric pattern. This asymmetric mode is distinct from the anthropogenic forcing simulations under different Representative Concentration Pathways (RCP). Furthermore, the variance explained by this asymmetric mode in the observed OHC changes is similar in magnitude to the underlying anthropogenic warming. Indeed, unlike the other known modes of climate variability, this asymmetric mode is a new mode of variation that has major influence on the redistribution of the global ocean heat gain and ocean heat transport predominantly through the meridional overturning circulation. Our study also suggests that this asymmetric mode can plausibly play a crucial role in the near-term (~10-15 years) prediction of sea-level rise, heat and carbon uptake by the ocean and the cross-equatorial ocean heat transport via the Atlantic Meridional Overturning Circulation. The fourth chapter of this thesis demonstrates that SSS variability is well correlated with variability in Australian precipitation on inter-seasonal to inter-annual timescales and provides a tool to locate the source of moisture. We have constructed seasonal composites during ENSO and IOD events to understand variations in atmospheric moisture transport and rainfall over Australia, and its association with SSS variability. We show that as ENSO and IOD events evolve, patterns of positive and negative SSS emerge in the Indo-Pacific warm pool region and are accompanied by atmospheric moisture transport anomalies towards (or away from) Australia. During co-occurring La Ni√±a and negative IOD events, salty anomalies around the Maritime Continent (north of Australia) indicate freshwater export. These anomalies are associated with a significant moisture transport that converges over Australia to create anomalous wet conditions. In contrast, during co-occurring El Ni√±o and positive IOD events, there is a moisture transport divergence anomaly over Australia that results in anomalously dry conditions. A case study of the extreme hydroclimatic events of Australia (e.g., 2010-11 Brisbane flood) demonstrates that the changes in SSS occurs before the peak of ENSO/IOD events. This lead time in SSS variations raises the prospect that tracking SSS variations could aid the prediction of Australian rainfall in tropical regions. The following chapter uses this relationship to demonstrate an improvement in the prediction of Australian rainfall using SSS. In the fifth chapter of this thesis, we combine singular value decomposition (SVD) and composite analysis with a machine learning algorithm (random forest regression) to use SSS as an additional precursor to improve Australian rainfall prediction. From SVD analysis, we show that the SSS in the prior seasons (July-September and September-November) in two regions (tropical western Pacific ‚- SSSP, and tropical eastern Indian Oceans - SSSI) covary with the December-February rainfall over northeastern Australia. The composite analysis is based on years of high and low SSS anomaly in the SSSP and SSSI regions. These SSS anomalies show the signature of co-occurring ENSO and IOD events. The use of random forest regression shows that incorporating the prior season's SSS anomaly as an additional precursor, in addition to temperature-based indices (ENSO and IOD), improves the prediction of rainfall over northeastern Australia. We also show that SSS in the SSSP region is the second most important precursor after ENSO for predicting rainfall over northeastern Australia. The variance explained by incorporating the SSS (r\\(^2\\) = 0.49) is more than the variance explained by the SST-based precursors alone, i.e., ENSO and IOD indices (r\\(^2\\) = 0.38) for the prediction of rainfall over northeastern Australia. Our study suggests that monitoring the SSS variations can provide a metric for Australian precipitation and can aid in improving its prediction. In conclusion, this thesis investigates the strong influence of the asymmetric mode of internal climate variability that can effectively redistribute anthropogenic heat gain by the global ocean. This asymmetric mode of internal variability is found in the pre-industrial control simulation, so it is additive to anthropogenic forcing. This thesis also demonstrates the physical link between SSS variability and Australian rainfall through atmospheric moisture transport and suggests that monitoring SSS variability could aid the prediction of Australian rainfall.