Quantifying Past and Future Climate Extremes: Characteristics, Uncertainties and Improvements

Abstract

In recent years, it has been widely recognized that a potential change in climate extreme events will occur under a warming climate. Understanding present and future changes in weather and climate extremes is critical to manage disaster risk and advance climate change adaptation. To date, certain aspects of extreme precipitation and temperature events have been analyzed in previous studies, such as increasing frequency of hot days with extreme high temperatures, reduced frequency of extreme cold days and increases in intense precipitation events. However, a large part remains unknown, especially over some sparsely observed regions, e.g. Africa due to complexity of the problem, making it difficult to interpret the spatiotemporal evolvement of extreme events. This dissertation aims to advance fundamental understandings on the depiction skills of past and future climate extremes in observations, reanalyses and climate projections as well as potential improvements via data assimilation techniques. In Chapter 2, five reanalyses precipitation datasets are assessed on their reconstruction of drought properties over Sub-Saharan Africa over the period 1979 to 2012. Skills in depicting the spatiotemporal characteristics of droughts are quantified for reanalyses products. Chapter 3 builds on the evaluation of reanalyses precipitation datasets and includes real-time satellite rainfall products. The depiction of global drought events by four reanalyses and three satellite precipitation estimates are compared against an observational reference dataset, Princeton Global Forcing (PGF) data. Chapter 4 analyzes projected changes in the characteristic of extreme temperature and precipitation events including frequency, seasonal timing, maximum spatial and temporal extent as well as severity based on a suite of 37 climate models archived in the Coupled Model Intercomparison Project Phase 5 (CMIP5). Chapter 5 presents a Monte Carlo based assimilation procedure to integrate soil moisture information into the Variable Infiltration Capacity (VIC) land surface model to improve real-time, satellite precipitation estimates. In the context of weather and climate extremes, the results of this dissertation show a path towards evaluating spatiotemporal characteristic of extreme events. This work also demonstrates potential improvements to real-time hydrologic states and fluxes estimates through data assimilation.