Future changes in intensity and seasonal pattern of occurrence of daily and multi-day annual maximum precipitation over Canada

Summary

Daily and multi-day extreme precipitation events can cause important flooding. Assessment of the future evolution of heavy precipitation is therefore crucial in a context of climate change. Simulation results for Canada from the Canadian Global Climate Model (CGCM3) have been analyzed for 1 to 5-day annual maximum (AM) precipitation events over the 1850–2100 period using simulation series from five ensemble members. Trend analysis showed that daily and multi-day intense precipitation series were stationary over the 1850–1980 period while trends emerged during the period 1980–2005. Probabilities of occurrence of AM precipitation for the various months were also estimated. For the historical climate (1850–1980), a comparison with observed data suggested that the model adequately reproduced the observed regional patterns of seasonal occurrence of AM events. Future projections suggest that, for many Canadian regions, a shift will take place from summer to spring and/or autumn in the seasonal occurrence of AM precipitation events. Moreover, statistical frequency analysis of models series suggests that daily and multi-day events will be more intense and frequent in a future climate for all regions except the Prairies. In some regions (e.g. west coast of British Columbia), the return period associated with a given precipitation intensity in historical climate will decrease by a factor of five over the 2080–2100 period. No significant differences have been observed between daily and multi-day projections for intensity/frequency occurrence of AM events.

Introduction

It is anticipated that the intensity/frequency of extreme events will increase in a future climate. Numerous studies analyzing the output of Global Climate Models (GCMs) support this hypothesis (see IPCC, 2007 for a summary of actual projections) as well as physical arguments about climatic processes involved in the generation of intense rainfall events (see e.g. Emori and Brown, 2005). It is therefore important to get some insight about the future evolution of extreme precipitation events (e.g. for urban and rural drainage infrastructure design or flood management; see Mailhot and Duchesne, 2010).

Many studies have looked at the evolution of extreme precipitation for various regions of the world using either Global Climate Models or Regional Climate Models (see e.g. Beniston et al., 2007, Frei et al., 2006). Kharin et al. (2007) have shown, using a multimodel approach that combined simulation results of the Intergovernmental Panel on Climate Change (IPCC) coupled global models, that global averaged amplitude of 20-year return period values of annual precipitation extremes will increase by about 10% for the SRES B1 experiment, 16% for the A1B experiment, and 20% for the A2 experiment by the end of the 21st century. For North America, the averaged expected changes are of the order of 10–20%. Available projections for Canada support these conclusions (see e.g. Kharin and Zwiers, 2005, Zwiers and Kharin, 1998). For instance, Kharin and Zwiers (2005) have shown, using simulation results from the Canadian Global Climate Model (CGCM2), that return periods of daily extreme events could be halved by the end of the 21st century. Similarly Mailhot et al. (2007) have analyzed simulation results of the Canadian Regional Climate Model (Plummer et al., 2006) for the southern region of Quebec. Their results also suggest significant increase in the intensity of maximum annual daily and sub-daily rainfall events.

Knowing the frequency/intensity of multi-day precipitation events is also very important as these can influence the design of urban structures such as drainage systems, dams, spillways and flood control measures (Fowler and Kilsby, 2003). Moreover, multi-day events can be the cause of important flooding (Fowler and Kilsby, 2003). For instance, Pielke and Downton (2000) have showed that, of the precipitation measures examined in their study, the ones that most closely related to flood damage, according to available data within the United States, are the number of 2-day heavy rainfall events and the number of wet days. Mechanisms generating multi-day events can also be quite different from those lasting less than 24 h and available projections for multi-day events may be different from projections for daily events. Kyselý and Picek (2007), for example, mentioned that multi-day heavy precipitation in the Czech Republic were usually associated with slowly moving cyclones over central Europe. The same is true for tropical cyclones in the western North Atlantic, which can cause strong winds and extreme rainfall, and can have a large impact on the weather of eastern Canada (Milrad et al., 2009). Many studies have considered multi-day extreme rainfall events, analyzing trends in past records (e.g. Boroneant et al., 2006, New et al., 2006, Fowler and Kilsby, 2003) or analyzing available series from climate models (e.g. Kyselý and Picek, 2007).

The impact of climate change (CC) on seasonal patterns of occurrence of intense precipitation events is also important in order to assess their hydrological impacts as these may vary according to the time of the year when heavy precipitation occurs (Fowler and Kilsby, 2003). For example, more frequent intense spring events can have huge impacts and considerably increase the vulnerability of many Canadian regions to flooding (e.g. the Red River valley; see Simonovic and Li, 2003 and Wilson and Rashid, 2005). As pointed out by Wehner (2004), seasonal changes may have more impact on human and natural systems than annual changes.

This study analyzes simulation results of the CGCM3 model for 1 to 5-day annual maximum precipitation (AMP) over Canada. Simulation results from five ensemble members (EM) are available and three greenhouse gas (GHG) emission scenarios are considered (SRES A2, A1B and B1; see IPCC, 2000 for a description of these scenarios). This study addresses the following issues: (1) How will daily and multi-day precipitation intensity/frequency evolve in a future climate over Canada? (2) What is the impact of CC on the monthly and seasonal occurrence of extreme events? Comparison with available data is also made in order to assess the ability of climate models to mimic the observed trends for multi-day intense precipitation over the last century. The impact of internal variability on trend detection is also discussed. It is worth noting that many aspects of the statistical methodology used to analyze simulated series differs from previous studies.

The paper is organized as follows. Section 2 gives an overview of the available simulated and observed series. Trend analysis and effect of internal variability of CGCM3 model data are investigated in Section 3. The proposed statistical methodology is described in Sections 4 Statistical analysis of AM grid box series, 5 Statistical analysis of ensemble members grid box series. Section 6 is devoted to the analysis of the time of occurrence of intense precipitation. A comparison of simulated and observed trends in historical climate is presented in Section 7 while Section 8 provides a summary of the main conclusions with a discussion.