Projecting climate change impacts on streamflow extremes using nonstationary generalized extreme value analysis

English Abstract:

Projecting nonstationarity of the streamflow extremes is crucial for managing river flooding in a changing climate. The objective of this study is to develop a statistical modelling framework for relating climate variables (i.e., precipitation and temperature) with streamflow extremes. The non-stationarity in streamflow extremes is represented by the variable parameter Generalized Extreme Value (GEV) distribution, with the Conditional Density Network (CDN) model used to evaluate the GEV parameters as a function of seasonal precipitation and temperature. The model was set up using an ensemble of 23 CMIP3 climate models derived precipitation and temperature, and simulated annual maximum streamflow from the Variable Infiltration Capacity (VIC) hydrologic model for the Fraser River Basin, Canada. Based on the model setup for the CMIP3 generation of GCMs, the CMIP5 precipitation and temperature will be used to derive the GEV distribution of annual maximum streamflow under CMIP5 climate change scenarios. Preliminary results indicate the flexibility of the GEV-CDN model to describe the non-stationarity of streamflow extremes. Future changes in streamflow extremes indicate increasing magnitude for a given return period. Such changes in the return periods could have major implications on flood risk, such as adequacy of the existing dikes in the lower Fraser region to offer long-term flood protection.