Stochastic precipitation generation for climate change studies

点击翻译按钮获取中文摘要

Canada is already experiencing the effect of climate change in the form of shrinking Arctic sea ice, shorter windows for operation of winter roads, and the apparent increase in the frequency of extreme weather events, to name a few examples. This trend is likely to continue into the future, and there is an interest in better understanding the magnitude and the nature of changes. Global climate models (GCMs) are used to predict the response of the Earth system to changing levels of greenhouse gasses. However, GCMs are known to produce biased estimates of precipitation which makes it difficult to use model data as input to hydrologic models and other management models. In practice, it is necessary to post-process GCM precipitation output using statistical methods. The proposed research program addresses the general issue of how precipitation information from climate models (global and regional) can be post-processed in a way that makes the data useful for input to different models. The methods currently in use have various short-comings which, depending on the specific application of the data, could be important. Specific research objectives include the development of new bias-correction techniques that will address some of the short-comings of existing methods and will result in more realistic precipitation scenarios. A second objective will focus on the development of stochastic weather generators that can be used to downscale precipitation to scales of practical interest. The proposed research will extend classical single-site and multi-site weather generators to fully gridded weather generators that can be used in climate change studies of large watersheds. A third objective will address the issue of climate change impact on the distribution of extreme precipitation. The design of urban drainage infrastructure is based on this type of information. Because climate model predictions of extreme precipitation of short duration are known to be unreliable, we will develop procedures to extract relevant information from precipitation information at other time scales using the principle of temporal scaling. The proposed research will address some of the gaps that exist between the climate modelling community and the end users, which include engineers who must design and plan the operation of water-related infrastructure. Climate change will have a large economic impact on countries around the world and Canada is no exception. Water is Canada's most precious natural resource and the proposed research will produce tools that will assist engineers and scientists in the practical use of climate model output. It is expected that this will lead to more sustainable and cost-effective design solutions for Canada's infrastructure.