An event-based approach to understanding the hydrological impacts of different land uses in semi-arid catchments

Abstract

Summary In semi-arid catchments around the world re-vegetation is often implemented to reduce quick surface runoff, combat severe soil erosion, restore degraded ecosystem functionality, and, ultimately, improve ecosystem productivity. However, to date, in these water stressed regions, the event-based hydrological impact of different land uses induced by re-vegetation activities is not fully understood at the watershed scale. Traditional hillslope plot experiments and paired watershed experiments have proved difficult to up-scale to a watershed level. In 2006 and 2007, we used broad-crested weirs to measure event streamflow from six catchments within the Caijiachuan watershed (area = 40.1 km2), located in the Loess Plateau, a semi-arid region in China. The six catchments have different land use compositions with functional combinations of crop, grassland, shrubland, secondary forest, and plantations. Over the same period, event rainfall was measured by a network of rainfall gauges deployed over the study site. We examined the difference in hydrological properties between the catchments using the non-parametric Firedman test, and differentiated the role of each land use in governing watershed hydrology using a numerical analysis technique. Our results showed important differences between the six catchments with respect to event runoff coefficients, normalized peak flow, and event duration. Each land use played a different role in catchment hydrology, as shown by the different mean runoff coefficients (rc) and mean representative surface flow velocities (V). Compared to secondary forest (rc = 0.017 and V = 0.07 m s−1), plantations (rc = 0.001 and V = 0.18 m s−1) provide a greater potential for increasing shearing force and had a larger impact on runoff reduction. Although shrubland (rc = 0.096 and V = 0.20 m s−1) and grassland (rc = 0.127 and V = 0.02 m s−1) have similar magnitude of mean runoff coefficients, grassland seems more capable of trapping sediment due to its lower surface runoff velocity. Cropland (rc = 0.008 and V = 0.05 m s−1) exerted an important effect on runoff reduction and a moderate effect on flow retardation. We concluded that, to combat severe soil erosion while minimizing water use, re-vegetation in the semi-arid Loess Plateau should not overly, or even solely, rely on plantations. Alternatively, to produce the desired ecosystem functionality, preservation and establishment of grassland during re-vegetation processes should be encouraged, at least, in the early stage of ecological restoration within a “successional re-vegetation” framework.