Background Construction activities such as transport projects may promote social development.However, it could also lead to serious soil erosion and environmental problems. In recent years, intensive engineering and construction activities have been conducted in the middle Yarlung Zangbo River basin in southeastern Tibet, China. Consequently, the local fragile ecological environment has been seriously disturbed. Quantitative analysis and theoretical support for erosion control and engineering disturbance evaluation is urgently needed. A composite fingerprint study about the contribution of soil loss from construction disturbance and its spatial variation can be helpful in engineering planning and assist the optimal deployment of conservation measures.

Methods Soil samples were taken from potential erosion sources including woodland, grassland, slag, road cutting and unpaved roads. Sediment samples were taken from riverbeds, and totally 18 physical and chemical parameters were measured. The best fingerprint combination of K₂O, Na₂O, CaO and Cr was selected using Kruskal-Wallis H test and multivariate stepwise discriminant analysis. A multivariate linear mixed model was used to analyze sediment contribution rate of each erosion source. The quantitative relationship between sediment transport distance and contribution rate was established to reveal its spatial variation.

Results The best fingerprint factors combination reached an overall correctly classified percentage of 94%for the total source soil samples. The average sediment contribution of the three main soil loss sources was ranked as: Road and side slope (61.61%) > slag site (21.15%) > natural woodland and grassland (17.24%). The average value of goodness of fit (GOF) was calculated as 0.867. The results showed that soil erosion from engineering disturbance was the main source of sediment that deposited in riverbeds of the small watershed. The sediment contribution rates of the two engineering disturbance sources showed different spatial variability. Sediment from the spoil site followed a linear decline trend of contribution rate along the river, with a critical impact distance of approximately 2 800 m. On the other hand, road cutting which generally parallel to the river continuously contributed to deposited sediment along the river. The road-related sediment contribution rates were high within a range of 220 m from the road. Outside this range, the contribution rate would decay exponentially as the distance between the road and the river increases. The effective road influence distance on deposited river sediment could be estimated as approximately 400 m.

Conclusions The results of this study show the profound impact of engineering disturbance on erosion and sediment production in a small watershed of Southeast Tibet. It revealed the difference in sediment contribution variation of linear road and slag site. Empirical equations were established to predict the potential impact range of engineering disturbance on local watershed sedimentation.