The overall goal of this study is to identify the teleconnection mechanisms that underlie ENSO-Antarctic climate links. Initially time series plots and cross correlation analysis of Antarctic surface and upper air climate data are used to search for high latitude atmospheric signals during Pacific Ocean warm (ENSO) and cold (La Nina) events. Consistent increases (decreases) in pressure were found during warm (cold) events, with a variable response in temperature. Upper air data demonstrate consistent changes in windspeed, cloud cover geopotential height" wind speeds and direction, temperature and relative humidity during ENSO events. Spatial anomaly plots, Hovmoller, harmonic and cluster analyses are used to identify ENSO related climate anomaly wavetrains, teleconnections and propagation mechanisms that link Southern Hemisphere low and high latitudes. Although inter-warm and cold event variability is a characteristic, strong meridional anomaly contrasts, equatorward and poleward anomaly propagation, and distinct jetstream behaviour were apparent in all events studied. In warm (cold) events subtropical jet strength increases (decreases) and polar jet strength decreases (increases) resulting in a decrease (increase) in poleward moving cyclonicity. The jetstreams are considered to play a major role in ENSO related climate anomaly propagations.

Paired kitchen-living room dust samples from 30 UK houses were collected for the analysis of BFRs, including PBDEs, HBCDDs and PBEB, EH-TBB, BTBPE, BEH-TEBP, DBDPE. Ninety-six plastic kitchen utensils were collected, screened for Br concentration by X-ray fluorescence spectrometer, with 30 of these samples analysed for BFR concentrations. A simulated cooking experiment was conducted to evaluate BFR exposure. Temporal and geographical differences in concentrations of BFRs in indoor dust samples were investigated via comparing BFR concentrations in UK samples in 2006-07 and 2015 and comparing 116 indoor house dust samples collected between 2014-15 from 6 countries (Finland, Greece, Spain, Jordan, US and Mexico) respectively. Concentrations of BDE-209 in living room dust were significantly lower and those of DBDPE significantly higher (p < 0.05) compared to concentrations in 2006-07 in UK dust. All target BFRs were present at higher concentrations in living rooms than kitchens. Considerable BFR transfer from kitchen utensils to cooking oils was observed and estimated exposure via cooking was 60 ng/day. US dust showed the highest Penta-BDE concentrations, followed by Mexico. Jordanian dust samples contained the highest concentrations of Octa-BDE. US and Mexican samples were found to display a similar composition to that found in the FireMaster® 550 formulation (EH-TBB:BEH-TEBP=4:1).

Data assimilation (DA) is an emerging research area in palaeoclimatology. Here, ensemble-based DA schemes are implemented and evaluated for the reconstruction of the climate of some of the key periods from the past millennium. The study is among the first to employ a General Circulation Model for palaeoclimate DA. An off-line and an on-line DA method are first compared, assimilating continental proxy-based temperature reconstructions and using the 17th century as testing period. Both schemes provide simulations that follow the assimilated targets on large scales better than without DA. The on-line scheme has the advantage of temporal consistency of the analysis, and is subsequently used to reconstruct the climate for 1750-1850 AD. The assimilation performs well on large-scale temperatures, but there is no agreement between the DA analysis and reconstructions for regional temperature patterns. Evidence is presented to suggest that this lack of information propagation to smaller spatial scales is likely due to the fact that the Northern Hemisphere continental mean temperatures are not the best predictors for large-scale circulation anomalies, or that the assimilated reconstructions include noise. The lack of regional skill is again found when instrumental data for 1850-1949 AD are assimilated. Based on these results, it is argued that a potential way of improving the performance of DA is the assimilation of temperature reconstructions with higher spatial resolution.

This research has examined the application of machine learning and statistical methods for developing roadside particle (number/mass concentrations) prediction models that can be used for air quality management. Data collected from continuous monitoring stations including pollutants, traffic and meteorological variables were used for training the models. A hybrid feature selection method involving Genetic Algorithms and Random Forests was successfully used in selecting the most relevant predictor variables for the models from the variables selected based on their correlation with the PM\(_+\), PM\(_{2.5}\) and PNC concentrations. The study found that the hybrid feature selection can be used with both statistical and machine learning methods to produce less expensive and more efficient air quality prediction models. Among the machine learning models studied the Boosted Regression Trees (BRT), Random Forests (RF), Extreme Learning Machines (ELM) and Deep Learning Algorithms were found to be the most suitable for the predictions of roadside PM\(_+\), PM\(_{2.5}\), and PNC concentrations. The machine learning models performed better than the ADMS-road model in spatiotemporal predictions involving monitoring sites locations. Moreover, they performed much better in predicting the concentrations in street Canyons. The ANN and BRT were found to be suitable for air quality management applications involving traffic management scenarios.