Trend in frequency of extreme precipitation events over Ontario from ensembles of multiple GCMs

Abstract

As one of the most important extreme weather event types, extreme precipitation events have significant impacts on human and natural environment. This study assesses the projected long term trends in frequency of occurrence of extreme precipitation events represented by heavy precipitation days, very heavy precipitation days, very wet days and extreme wet days over Ontario, based on results of 21 CMIP3 GCM runs. To achieve this goal, first, all model data are linearly interpolated onto 682 grid points (0.45° × 0.45°) in Ontario; Next, biases in model daily precipitation amount are corrected with a local intensity scaling method to make the total wet days and total wet day precipitation from each of the GCMs are consistent with that from the climate forecast system reanalysis data, and then the four indices are estimated for each of the 21 GCM runs for 1968–2000, 2046–2065 and 2081–2100. After that, with the assumption that the rate parameter of the Poisson process for the occurrence of extreme precipitation events may vary with time as climate changes, the Poisson regression model which expresses the log rate as a linear function of time is used to detect the trend in frequency of extreme events in the GCMs simulations; Finally, the trends and their uncertainty are estimated. The result shows that in the twenty-first century annual heavy precipitation days, very heavy precipitation days and very wet days and extreme wet days are likely to significantly increase over major parts of Ontario and particularly heavy precipitation days, very wet days are very likely to significantly increase in some sub-regions in eastern Ontario. However, trends of seasonal indices are not significant.