Background Quantifying the impacts of precipitation variation and vegetation restoration on runoff is essential not only for understanding the mechanism of hydrological response in watershed, but also for local water resources management as well as preventing floods and droughts. The Pengchongjian small watershed has experienced forest deforestation by local inhabitants in early 1980s. Since then, its vegetation has been restoring without human disturbances, and it is a natural site for this study. This paper aims to calculate the exact contribution rates of precipitation variation and vegetation restoration to runoff.

Methods Based on observed daily precipitation and runoff data at Pengchongjian hydrological station from 1983 to 2014, we applied Mann-Kendall test and trend analysis methods to detect the inflection points of annual precipitation and runoff. Then, after dividing baseline period and changing period from 1983 to 2014, we calculated the exact contribution rates of precipitation and vegetation restoration on different time scales by using comparative methods of empirical statistics and hydrological modeling simulation.

Results 1) A consistent inflection point of 2003 was found both for annual precipitation and runoff. Thus taking 1983 to 2003 as baseline period, 2004 to 2014 as changing period, the linear relationship between annual precipitation and runoff in baseline period was fitted. 2) Compared to the baseline period, annual precipitation and runoff decreased by 8.7% and 29.2%, with average annual decrease of 12.7 and 22.1 mm. The average depth of runoff in Spring, in Summer and in whole year scale decreased by 100.2, 105.8 and 243.0 mm respectively. 3) The results of the empirical statistics suggested that the contribution rates of precipitation variation and vegetation restoration on runoff were 58.9%, 41.1% and 71.6%, 28.4% in Spring and in Summer respectively, while results from the hydrological model simulation were 61%, 39% and 81.6%, 18.4%. However, on annual scale, the contribution rates of precipitation variation and vegetation restoration were 57.1% and 42.9% by empirical statistics method while the rates were 70% and 30% by hydrological model simulation method.

Conclusions The results of the two methods are generally similar, indicating that the results are more credible. It is also known that the results are different on different time scales, and the contribution rate of precipitation variation is greater than that of vegetation restoration. That is to say, both precipitation and vegetation influence the change of runoff, but the main driving factor of runoff change is precipitation, which determines the trend of runoff change.