全球变化知识资源中心

Information on spatial and temporal variation in rainfall and temperature pattern is of prime importance for understanding the climate dynamics and prerequisite for preparation of strategies to mitigate or adapt the ill impacts of potential future climate change on water resources, crop production and enhance resilience of vulnerable communities. In this study, five trend detection methods considering parametric tests [linear regression, Prais-Winsten AR, Pearson product-moment correlation coefficient (r)] and non-parametric tests [Spearman rank correlation coefficient (Spearman's rho) and Mann-Kendall test with Theil-Sen's slope] were considered. In addition, a new hybrid approach (Gaussian-linear trend detection test) to investigate the trends in historical rainfall and auto-segmented linear function for detecting trend in temperature of Upper Kharun Catchment (UKC), India, are introduced. Both parametric (except Gaussian-linear regression) and non-parametric test for trend analyses of annual rainfall (1961-2011) for the UKC shows non-significant trend. However, the Gaussian-linear trend analysis for peak (maximum) monthly rainfall amount in a year shows an increasing trend at the rate of 1.94mm per annum (p=0.05 level of statistical significance). At individual stations only Bhilai rainfall station shows a significant increase in monthly rainfall amount at the rate of 2.43mm (p=0.1) in August. Concerning trend detection of temperature, the results show no statistically significant trend for mean annual maximum, mean annual minimum and mean annual average temperature. However, a statistically significant slight increasing trend in temperature was detected in specific months.