No abstract

A number of recent studies have examined trends in sea ice cover using ordinary least squares regression. In this study, quantile regression is applied to analyse other aspects of the distribution of sea ice extent. More specifically, trends in the mean, maximum, and minimum sea ice extent in the Arctic and Antarctic are investigated. While there is a significant decreasing trend in mean Arctic sea ice extent of −4.5% per decade from 1979 through 2010, the Antarctic results show a small positive trend of 2.3% per decade. In some cases such as the Antarctic minimum ice cover, selected quantile regressions yield slope estimates that differ from trends in the mean. It was also found that the variability in Antarctic sea ice extent is higher than that in the Arctic.

Large flood events in recent years in the state of Wisconsin, USA raised a question of whether high precipitation events are on the rise. The objective of the research was to examine temporal and spatial patterns of extreme precipitation in Wisconsin during 1950–2006. The daily precipitation data used in the study were created by researchers at the University of Wisconsin‐Madison using spatial interpolation of weather stations data across the state to a grid mesh with a spatial resolution of 8 km. For extreme precipitation indices, we calculated 99th, 95th, 90th, 85th, and 80th percentiles of daily total precipitation (>1 mm) in a year and the number of days per year with daily precipitation exceeding 10 mm, 20 mm, and 50 mm. We also conducted the Mann–Kendall test for trend, examined how geographical heterogeneity varied over time, and built quantile regression models for annual summer precipitation. Main findings include the following: the temporal trend of extreme precipitation varied widely across the state; the highest percentile index showed an increasing trend over the largest area, whereas indices of less extreme precipitation tended to generally decrease; extreme precipitation tended to show more dispersed and skewed spatial patterns than annual total precipitation. Overall, indices related to frequency showed more similar spatial and temporal trends to total precipitation than magnitude indices.

The effects of parameters such as weir length, height of weir, water depth at weir entrance and bed slope on the discharge coefficient of a rectangular sharp-crested side weir were analysed using standard linear regression and the adaptive neuro-fuzzy inference system (Anfis). The results showed that the Anfis model performed better than the regression model in all different scenarios defined in this study. Anfis can simulate the discharge coefficient better than the regression model and always works better. Without considering the channel slope, both models represent satisfactory results, but the Anfis model shows more sensitivity to channel slope than does the regression model. Based on the results of this study, the Anfis model is suggested for estimation of side weir discharge coefficient. NotationFroude number at upstream end of side weir dQ/ds discharge per unit length of side weir ((m 3 /s)/m) g acceleration due to gravity (m/s 2 ) L length of side weir (m) p height of weir crest (m) Q discharge in main channel (m 3 /s) Q 1 discharge in main channel at upstream end of side weir (m 3 /s) Q 2 discharge in main channel at downstream end of side weir (m 3 /s) q discharge per unit length over side weir ((m 3 /s)/m) S 0 channel slope s distance along side weir measured from upstream end of side weir (m) V 1 mean velocity of flow at upstream end of side weir (m/s) y flow depth measured from the channel bottom (m) y 1 flow depth at upstream end of side weir at channel centre (m) y 2 flow depth at downstream end of side weir at channel centre (m)