Trends in Low Flows of the Selected Rivers in Lorestan Province

Background and Objectives
Detection of trends in streamflow characteristics such as low flows is so important in optimal water resources management. This is especially true in arid and semi-arid climates that water is vital for human being and all other living things. Lorestan province, located in the west part of Iran was considered as the study area. Daily mean streamflow data of hydrometric stations during the time period provided by Lorestan Regional Water Company located in Khorramabad city. Eight hydrometric stations selected for this purpose. The altitude of the chosen stations varied between 770 and 2050 m above the mean sea level. Literature review on this subject indicated that low flow trends in Lorestan province river has not been before studied. On the other hand, such a study is so important for better management of fresh water in the region, therefore, conducting this study seems to be necessary.
Methodology
In the first step of this study, flow duration curves (FDC) plotted for the stations. Two indices including the Q_0.05 and Q_0.10 were considered here as measures of low flows. The Q_0.05 index value extracted from the FDC as a five percent low flow quantile. This index shows that the streamflow discharge is less than that in five percent of the days in a year. In addition to Q_0.05 the second measure namely Q_0.10 is read from FDC of the selected sites. Therefore, for each site, these two indices were gathered during the used time period. Then, trends of the mentioned indices are analyzed using the Mann-Kendall method. In this regard the effect of serial correlations removed from the time series. Moreover, the slope of trend lines estimated using the Sen’s estimator approach. Then the selected quantiles fitted for suitable statistical distributions. In this regard the well-known method namely Kolmogorov—Smirnov was used.The parameters of the selected distribution estimated using the maximum likelihood method. Finally, at each site the values of low flows corresponding to different return periods i.e. 2, 5, 10, 25 and 50 years were estimated. It is worthy to mention that the missing data are not reconstructed here, because the used approach (Mann-Kendall) is a non-parametric method and no need for reconstruction of the missed values.
Findings
Results showed that two stations namely Sokaneh-Nahaee and Cham-Chit had significant first-lag serial correlation for Q₀₁₀ time series. The other six sites had no significant serial correlation for this quantile. Furthermore, in the case of Q_0.05, three sites showed significant first-lag serial correlations. So, the modified Mann-Kendall method was used for these time series. This analysis indicated that the mentioned series auto-correlation was significant in 5% level. The conventional MK method was used to detect trends in other time series which their serial correlations are not significant. Results showed that trends of Q_0.10 quantile series in four out of the eight hydrometric stations were downward and two of them were statistically significant at 1% level. The two others had no significant trends. At the same time the other sites showed upward trends in Q_0.10 series. However, among these series one station namely Cham-Chit had statistically upward trend at 1% level. Trends in the other three sites are not statistically significant.  In the case of Q_0.05 series, the six out of the eight stations showed negative trends, in which two sites had statistically significant trends at 1% level. The names of these sites are Sokaneh-Nahaee and Vanaee. In contrast, the two other time series showed positive trends, in which only one of them (namely Cham-Chit station) had statistically significant trend at 1% level. The trend line slopes of Q_0.10 quantile time series are ranged between -0.0844 (in Sokaneh-Nahaee) and +0.12 (in Cham-Chit). However, in the case of Q_0.05, this range was between -0.0885 (in Sokaneh-Nahaee) and +0.008 (in Cham-Chit station).
Conclusion
Some of the stations showed upward trends in low flows in Lorestan province rivers and some others showed downward trends. It is worthy to note that the time periods of the used data of the stations are not same. Although Masih et al. (2011) reported that the mean daily stream flows (i.e. Q_0.50 or 50^th quantile also called median) of Zagros Mountain rivers had downward trends. Investigation of sites location indicated that no obvious pattern in trends of low flows existed in the area under studied. This is true for both two low flow indices (i.e. Q_0.05 and Q_0.10) used for hydrometric stations. The findings of this study would be helpful in better management of surface water in Lorestan province.