Trend Detection in Precipitation and River Discharge to Assess Climate Change in the Upper Tana Basin

Abstract

With Climate Change being a point of focus in modern times, this research study was done to detect the existence of a statistically significant trend in the hydro-meteorological characteristics of the upper Tana catchment to deduce whether Climate Change had occurred. The methodology involved subjecting mean monthly and mean annual precipitation and river discharge data to the Mann-Kendall trend detection tool from the XLSTAT software. Further, the one-tail trend tests (upper-bound test and lower-bound test) were done to confirm whether the trend detected was increasing or decreasing. In order for Climate Change to be reasonably implied, at least 50% of the mean monthly data needed to exhibit a statistically significant trend and secondly, have mean annual data return a statistically significant trend as well. The results suggested that for mean monthly rainfall, there was a statistically significant decreasing (negative) trend detected in the months of August and September (16.67%). With regard to river discharge, a statistically significant positive trend was observed in the months of January, February, March and November (33.33%). However, for mean annual discharge, there was an overall positive statistically significant trend for the years spanning 1966-2006. These results for mean annual discharge were inclusive of data that had perceived outliers (2001-2003). When the same Mann-Kendall trend test was done on mean annual discharge data between 1966-2000, no statistically significant trend was observed. In conclusion, the results obtained from this study could not conclusively imply Climate Change for two reasons: first, only 16.67% and 33.33% of mean monthly rainfall and mean monthly discharge respectively detected a statistically significant trend; both results were below the (50%) threshold set in the objectives. The mean annual rainfall did not detect any statistically significant trend and although the mean annual discharge data for the period 1966-2006 detected a statistically significant positive trend, when the Mann Kendall tool was run on the mean annual discharge omitting the 2001-2003 data, no statistically significant trend was detected. The second reason why Climate Change could not be reasonably implied is due to the fact that the rainfall data obtained had a relatively short span (1980-1994) and only six rainfall stations were analyzed in the vast upper Tana catchment. This sample dataset could be viewed as disproportionate to the size of the catchment and therefore relatively inadequate to give a proper representation of the catchment characteristics. It was recommended that a study be done on a comparable catchment with a longer record of data to check trend; advanced methods to check data accuracy be deployed to validate rainfall data for years 2001-2003; an independent study of catchment degradation be done to determine its contribution to increased runoff in the rivers and finally a trend detection study be carried out on extreme rainfall and river flow values.