18.06.2015 Views

Flood Risk and Vulnerability Analysis Project - Atlantic Climate ...

Flood Risk and Vulnerability Analysis Project - Atlantic Climate ...

Flood Risk and Vulnerability Analysis Project - Atlantic Climate ...

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

Development of <strong>Project</strong>ed Intensity-Duration-Frequency Curves<br />

for Corner Brook <strong>and</strong> Goulds/Petty Harbour, Newfoundl<strong>and</strong> May 16, 2012<br />

would affect the ―weight‖ of the tail, which is of the most interest in modeling extreme<br />

values. The strength of predictors was evaluated to condition the shape parameter of<br />

the Frechet distribution explicitly, but no variables showed appreciable strength. This<br />

may be because there are not a sufficiently large number of annual values to derive a<br />

significant relationship. Accordingly, the Gumbel distribution was adopted.<br />

The GEV models were fitted using the extRemes package (http://cran.rproject.org/web/packages/extRemes/extRemes.pdf)<br />

in the statistical package ―R‖<br />

(http://www.r-project.org/). To allow for possible seasonal shifts in extreme event<br />

occurrence, seasonal models were fit for each storm duration. In the seasonal model<br />

fitting process, all predictor data across May to October were fit against all the data for<br />

each individual storm duration. For each duration, models for three principal predictors<br />

were evaluated: average precipitation, average temperature <strong>and</strong> the product of average<br />

temperature <strong>and</strong> average precipitation. For each duration, models based on different<br />

predictors were accepted based on statistical significance at a criterion of 0.05 <strong>and</strong> the<br />

best model was selected based on the Akaike Information Criterion (AIC) (explained in<br />

Towler, et al., 2010; referenced from Akaike 1974).<br />

For the 9 different durations, statistically significant models were found for each station.<br />

Of these, the best predictor for 5 of the models was total monthly precipitation, for 11<br />

models the best predictor was the product of monthly precipitation <strong>and</strong> temperature, <strong>and</strong><br />

for the remaining 2 models the best predictor was average monthly temperature. The<br />

final model predictors are provided in Appendix A. The parameter selected as the best<br />

predictor for each of the month/duration combinations is shown in Table 4-3.<br />

AMEC Environment & Infrastructure 22

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!