Assessing Variabilities of Extreme Precipitation and Snow Depth Using Climate and Stochastic Models

Floods are natural disasters with a significant impact on regions worldwide. They cause extensive damage to infrastructure, disrupt transportation and communication networks, and lead to the displacement of populations. Moreover, floods have long-term consequences on ecosystems, agriculture, and economies. In recent years, Canada has experienced several devastating flood events, highlighting the nation’s vulnerability to such disasters. Climate change, with its associated extreme weather patterns, has exacerbated the frequency and intensity of these events. Specifically, heavy rainfall and rapid snowmelt have triggered extensive flooding in multiple provinces. As global temperatures rise and weather patterns change, the world must remain vigilant and adapt approaches to address the evolving threat of floods. To address this issue, we present an extensive investigation of climate models’ performance in reproducing annual maxima of daily precipitation (AMP) globally and daily snow depth (SD) in Canadian catchments. We analyze projections for extreme precipitation, emphasizing the importance of adopting non-stationary models. Additionally, we introduce a stochastic model replicating SD time series with the same observed statistical properties to overcome limited observed SD data. These studies employ advanced and novel statistical methods, including bivariate analyses, L-moment metrics, Monte Carlo analysis, and autoregressive models. To accurately assess climate models, we use numerous unique observational datasets, along with the latest generation of climate models, the Coupled Model Intercomparison Project Phase 6 (CMIP6), to reflect recent advances in climate change impacts. First, the results show that 70% of CMIP6 models exhibit a percentage difference of ±10% in annual maxima mean and variation. However, CMIP6 simulations generally overestimate daily SD by at least 10%, with some regions challenging to simulate due to their complex atmospheric and land interactions, such as the Arctic and tropical regions. Second, extreme precipitation projections indicate that the return period of 100-year historical events will decrease by approximately 50% and 70% in the northern and southern hemispheres, respectively. Under the highest emission scenario, the projected 100-year levels are expected to increase by 7.5% to 21% over historical levels. Using stationary models to estimate the 100-year return level for AMP projections with trends leads to an average underestimation of 3.4%. Third, the developed stochastic model can reproduce daily distributions, temporal clustering and correlation, daily probability of zero, and annual seasonal patterns. This model can provide a reliable synthetic time series of SD, minimizing the scarcity of observed data for SD. This thesis provides engineers with essential information about climate change impacts, climate model performance, statistical behaviour of various models, and necessary datasets related to AMP and SD, which contribute to severe floods. Therefore, the findings are essential for hydrological, hydrodynamical, ecological, and water resources applications, helping society adapt to extreme climate conditions.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.