Change, Variability and Trend Analysis of Hydro-Climatic Time Series

Abstract

Hydro-meteorological time series are no longer stationary, hence the assumptions of stationarity in water resources assessment are no longer valid. Consequently, hydrometeorological trends are expected to either increase or decrease due to changes in land use and climate, which alter means and extremes of, for example, precipitation and streamflow. However, stationary trends have been observed in other disciplines. Trend analysis is therefore crucial for water resources planning and future projections of climate change impacts. The Mann-Kendall test and Sen's slope estimator are selected in order to detect trends in rainfall and streamflow in Mbuluzi catchment as well as to estimate the magnitude of the trends. The results indicate an increasing and statistically non-significant trend in rainfall and a decreasing and statistically non-significant trend in streamflow in the catchment. Due to non-uniformity and data scarcity in trend analysis, three levels of trend detection (short-term, medium and long-term) were proposed and it is suggested that trend analysis be undertaken for observed and simulated time series data in order to promote consistency and to consider data availability issues. It is also proposed that trend analysis be categorised into climate-driven and development-driven trends. Furthermore, the difference between change and trends in relation to stationarity is emphasised. It is concluded that trend analysis can be conducted using simulated data in cases of data scarcity and to establish the influence of climate-driven change on streamflow/water resources systems.