Homogeneity analysis of streamflow records in arid and semi-arid regions of northwestern Iran

Homogeneity analysis of the streamflow time series is essential to hydrological modeling, water resources management and climate change studies. In this study, five absolute homogeneity tests and one clustering approach were used to determine the homogeneity status of the streamflow time series （over the period 1960-2010） in 14 hydrometric stations of three important basins （i.e., Aras River Basin, Urmia Lake Basin and Sefid-Roud Basin） in northwestern Iran. Results of the Buishand range test, yon Neumann ratio test, cumulative deviation test, standard normal homogeneity test and Pettitt test for monthly streamflow time series detected that about 42.26%, 38.09%, 33.33%, 39.28% and 68.45% of the streamflow time series were inhomogeneous at the 0.01 significance level, respectively. Streamflow time series of the stations located in the eastern parts of the study area or within the Urmia Lake Basin were mostly homogeneous. In contrast, streamflow time series in the stations of the Aras River Basin and Sefied-Roud Basin showed inhomogeneity at annual scales. Based on the overall classification for the monthly and annual streamflow series, we determined that about 45.60%, 11.53% and 42.85% of the time series were categorized into the ＇useful＇, ＇doubtful＇ and ＇suspect＇ classes according to the five absolute homogeneity tests. We also found the homogeneity patterns of the streamflow time series by using the clustering approach. The results suggested the effectiveness of the clustering approach for homogeneity analysis of the streamflow time series in addition to the absolute homogeneity tests. Moreover, results of the absolute homogeneity tests and clustering approach indicated obvious decreasing change points of the streamflow time series in the 1990s over the three basins, which were mostly related to the hydrological droughts.

Implicit Hypotheses Are Hidden Power Droppers in Family-Based Association Studies of Secondary Outcomes

Family-based tests of association between a genetic marker and a disease constitute a common design to dissect the genetic architecture of complex traits. The FBAT software is one of the most popular tools to perform such studies. However, researchers are also often interested in the genetic contribution to a more specific manifestation of the phenotype (e.g. severe vs. non-severe form) known as a secondary outcome. Here, what we demonstrate is the limited power of the classical formulation of the FBAT statistic to detect the effect of genetic variants that influence a secondary outcome, in particular when these variants also impact on the onset of the disease, the primary outcome. We prove that this loss of power is driven by an implicit hypothesis, and we propose a derivation of the original FBAT statistic, free from this implicit hypothesis. Finally, we demonstrate analytically that our new statistic is robust and more powerful than FBAT for the detection of association between a genetic variant and a secondary outcome.

Score Tests for Zero-inflated Double Poisson Regression Models

Count data with excess zeros encountered in many applications often exhibit extra variation. There- fore, zero-inflated Poisson （ZIP） model may fail to fit such data. In this paper, a zero-inflated double Poisson model （ZIDP）, which is generalization of the ZIP model, is studied and the score tests for the significance of dis- persion and zero-inflation in ZIDP model are developed. Meanwhile, this work also develops homogeneous tests for dispersion and/or zero-inflation parameter, and corresponding score test statistics are obtained. One numer- ical example is given to illustrate our methodology and the properties of score test statistics are investigated through Monte Carlo simulations.