A Model for Nighttime Minimum Temperatures

作者：Dupuis Debbie J^*

来源：Journal of Climate, 2014, 27(19): 7207-7229.

DOI：10.1175/JCLI-D-14-00101.1

• North America
• Extreme events
• Climate change
• Statistical techniques
• Time series

The southwestern United States has experienced some of the most important increases in nighttime minimum temperatures over the last 60 yr, and climate models are projecting more of the same to the end of the century. As climate, geography, and population density vary considerably over the area, very diverse extreme temperature levels and dynamics are observed. It is shown how nighttime minimum temperatures over the 1950-2009 period exhibit more complex dynamics than daytime maximum temperatures. The author studies nighttime minimum temperature series from 12 locations and presents one model capable of capturing all the features of the data at each location. The time series preprocessing model normalizes seasonal shocks by daily and yearly volatility components before modeling the residual volatility as an exponential generalized autoregressive conditional heteroskedasticity [EGARCH(1, 1)] process with seasonal autoregressive structure to account for the presence of nonlinear and seasonal linear dependence, respectively, in the residual series. An exceedance over high thresholds approach is then used to model the tail of the distribution of scaled residuals from the preprocessing model. The resulting marginal distribution of nighttime minimum temperature at each location is then examined to see how it has changed in mean, scale, and shape, respectively, over the 60-yr period. Changes at the 12 locations vary considerably: many locations have seen considerable change in some or all of the three parameters, while two locations have experienced little or no change.

• 出版日期2014-10