Analysing the influence of human activity on runoff in the Weihe River basin

Changing runoff patterns can have profound effects on the economic development of river basins. To assess the impact of human activity on runoff in the Weihe River basin, principal component analysis (PCA) was applied to a set of 17 widely used indicators of economic development to construct general combined indicators reflecting different types of human activity. Grey relational analysis suggested that the combined indicator associated with agricultural activity was most likely to have influenced the changes in runoff observed within the river basin during 1994–2011. Curve fitting was then performed to characterize the relationship between the general agricultural indicator and the measured runoff, revealing a reasonably high correlation (R2 = 0.393) and an exponential relationship. Finally, a sensitivity analysis was performed to assess the influence of the 17 individual indicators on the measured runoff, confirming that indicators associated with agricultural activity had profound effects whereas those associated with urbanization had relatively little impact.


INTRODUCTION
Runoff volumes are sensitive to both climate change and changing land use patterns (He et al. 2012).Consequently, there is great interest in determining how global warming interacts with other environmental changes and human activities to influence runoff behaviour and the water cycle (Hou et al. 2011).The measured runoff volumes in China's six largest river basins have declined significantly over the last 50 years; a particularly pronounced decline has occurred in the middle and lower reaches of the Yellow River, especially in the basins of the Jinghe and Weihe rivers (Zhang et al. 2007(Zhang et al. , 2009)).The Weihe River basin is located in the eastern part of the fragile ecological environment of Northwest China, which is prone to natural disasters and is heavily affected by human activity, especially in the vicinity of Guanzhong.Because alterations in the region's runoff patterns could adversely affect regional development and hinder attempts to mitigate climate change, we investigated the relationship between runoff in the Weihe River basin and changes in socioeconomic activity within the basin over time.
The Weihe River flows over China's Loess Plateau, originating north of Niaoshu mountain in Weiyuan county of Gansu Province, and emptying into the Yellow River in Tongguan county of Shaanxi Province.It has a drainage area of 135 000 km 2 , is 818 km long, and flows in an eastward direction through 84 counties and three provinces across the Guanzhong basin (Zhang et al. 2009).The Weihe River basin spans longitude and latitude ranges of 103.5°-110.5°Eand 5°-37.5°N,respectively.Its two main tributaries are the rivers Jinghe and Beiluohe (Jiang et al. 2013).The southeastern part of the Weihe River basin is located in the continental monsoon zone, while the northwest is in the transitional zone between arid and humid regions, which experiences droughts in spring and alternating periods of rainy heat and drought in summer due to the west Pacific Subtropical High.The region is dry and cold in the autumn and winter due to the Mongolia High (Zhang 2002).Around 65% of the basin's annual precipitation occurs between July and October (Wang and Wang 2000).The region's average temperature is -1 to -3°C in the coldest month (January) and 23-26°C in the warmest (July) (He and Xu 2006).

Methodology
Principal component analysis (PCA) was used to extract principal components from a set of indicators of social and economic activity in the Guanzhong area of the Weihe River basin.Grey relational analysis was then used to determine the correlation between these principal components and the runoff data.The curve estimation feature of SPSS was used to obtain establish a regression equation relating the principal components to the measured runoff values.Finally, a sensitivity analysis was performed to determine the influence of each individual socio-economic activity indicator on the runoff volumes of the Weihe River basin.

Principal component analysis PCA was introduced by
Pearson in 1901 and extended to random vectors in 1933 by Hotelling (Yu and Fu 2005).It is mainly used to reduce the dimensionality of large datasets by converting many observations of correlated variables into a set of values of linearly uncorrelated principal components, thereby simplifying subsequent data handling and interpretation (Liu and Zhang 2011).PCA was performed using SPSS as described by Lin and Zhang (2005) to construct generalized indicators reflecting different types of social and economic activity within the basin.

Grey relational analysis
Grey relational analysis (GRA) is a component of the grey system theory developed by Prof. Deng.It aims to quantitatively describe the evolution of a studied system in comparative terms.Essentially, it is used to determine the extent to which a given sequence of observations resembles a reference sequence and to determine the correlation between the two (Zhou and Li 2007).There are two kinds of GRA, absolute and relative.Relative GRA is used to characterize the dynamic similarity of the investigated sequences and only provides information of the rate of change of the values in the sequence.However, it avoids some problems that have been identified with absolute GRA (Chen 2007).In this work, a relative GRA was performed to assess the similarity of the trends observed in the runoff data to the variation observed in the generalized socio-economic activity indicators obtained from the principal component analysis.

Sensitivity analysis
Sensitivity analysis is used to determine how sensitive the output of a model or system is to uncertainty in its various inputs.In this work, sensitivity analysis was performed on a model relating the runoff due to human activity within the Weihe River basin to variation in one of the generalized socio-economic activity indicators derived by principal component analysis.The sensitivity analysis was performed to determine how changes in the individual socio-economic activity indicators that contribute to the generalized indicator influence the expected runoff.

Grey relational analysis of PCs and runoff
To assess the impact of human activity on runoff in the Weihe River basin, we examined runoff measurements made at the Huaxian monitoring station.The difference between the unimpaired runoff QH,T and the measured runoff QH,S was computed to determine the runoff due to human activity, QH,R, in the years 1994-2011, and equations ( 1) and ( 2) were used to compute values for PC1 and PC2 over the same period.GRA was then used to assess the relationship between QH,R and the two principal components.The γ values for PC1 and PC2 were γ1 = 0.50 and γ2 = 0.889, respectively, indicating that the variation in QH,R most closely resembles that of PC2.

Curve fitting
The curve estimation module of the SPSS software package was used to analyse the relationship between QH,R and PC2.The best fit was achieved with an exponential relationship (Table 3).The regression analysis yielded the following relationship: Figure 2 depicts an increasing function, suggesting that as the agricultural development of the Weihe River basin proceeds, human activity will cause progressively more pronounced changes in the basin's runoff.influential factor with a negative impact on QH,R was X5.Six indicators (X9, X17, X10, X16, X1) had modest impacts, causing QH,R to change by 0.05-0.1%.The other indicators all changed QH,R by less than 0.05%.The indicators associated with the most pronounced changes in QH,R were all linked to agricultural activity, demonstrating that agriculture continues to account for most of the water use in the Weihe River basin and has profound effects on the basin's runoff as well as the demand for water resources.Indicators associated with urbanization have relatively little impact on runoff.

CONCLUSIONS
Principal component analysis (PCA) has been used to construct combined indicators of human social and economic activity in the vicinity of Guangzhong in the Weihe River basin from a set of 17 individual indicators.Relative GRA was then used to determine which of these combined indicators had the greatest effect on the runoff due to human activity (QH,R).Finally, curve fitting and sensitivity analysis were used to determine which of the original 17 indicators had the greatest impact on the impact of these combined indicators on QH,R.It was found that most of the variation in the 17 socioeconomic activity indicators could be summarized using only two principal components: PC1, which was interpreted as a general economic development indicator, and the general agricultural development indicator PC2.Relative GRA revealed that the variation in QH,R resembled that of PC2 (γ2 = 0.889) more closely than that of PC1.Curve fitting using SPSS was used to derive an exponential expression relating QH,R and PC2, and sensitivity analysis was then performed to identify the individual socio-economic activity indicators with the greatest effect on QH,R.Overall, our results indicate that agriculture has profound effects on runoff in the Weihe River basin, but the impact of urbanization on runoff is less clear.

Fig. 2
Fig. 2 The relationship between PC2 and QH,R.The solid line shows the fitted curve.

Fig. 3
Fig. 3 Tornado diagram showing the sensitivity of QH,R to a 1% increase in each of the 17 individual socioeconomic activity indicators considered in this work.

Table 3
Results of the curve fitting analysis.