The effects of changes in agricultural land use, management practices and landscape features on catchment flow and sediment generation

Agricultural intensification is related to changes in agricultural practices, land use management and alterations to landscape features. The use of a model to represent real-based scenarios of agricultural practices, land use management and landscape features alterations providing quantitative effects on flow and sediments was not found in previous studies. Therefore, this contribution examines flow and sediment generation under agricultural change scenarios by applying a physically-based, spatially-distributed model, SHETRAN, in two agricultural catchments in southwest England. Model calibration was conducted on the Blackwater catchment (18 km2) with evaluation on the Kit Brook catchment (22 km2) using 15-minute flow and suspended sediment data from September 2010 to October 2014. Calibration simulations were performed at a 50 x 50 m grid resolution using a 2010 land cover map where winter barley was the dominant arable crop. Parameter sensitivity was evaluated by one-at-a-time and multi-criteria methods, in addition to model sensitivity to changes in DEM resolution. Simulated catchment scenarios included: a) historic land cover maps; b) land cover end-member; c) crop types; d) soil compaction effects; e) arable fields varying in slope range; f) grass and woodland riparian buffers; and g) randomized spatial arrangement of cropped fields. In three subcatchments, alterations of landscape features included: a) roads, b) sunken lanes, c) urban areas and d-f) the removal of them (substituted by grass). Model calibration presented a Nash-Sutcliffe coefficient of 0.8 and 0.6 in model evaluation. The parameters to which the model is most sensitive were saturated hydraulic conductivity, Stickler coefficient, and the erodibility coefficients. Sensitivity analysis on grid resolution showed increases in flow and sediment flux peaks and decreases in event flow/sediment duration with decreasing grid size. Comparison of past land covers indicate small flow volume (Qv) variations. However, sediment yield (Sy) increased linearly with cropland area. The woodland cover end-member presented Qv (16%) and Sy (35%) reductions compared with the grass cover end-member. The use of grass as a cover crop after harvesting showed sediment reduction by up to 62% for maize, 49% for spring cereals and 28% for winter cereals. Removing a soil compaction effect from arable land produced minor changes in Qv and Sy. Arable fields in low slope areas reduced Sy by 30%; whereas crops located on slope >7° increased exports by 22% (Blackwater). Riparian buffer showed Sy reduction with a higher efficiency in woodland (14% Blackwater, 22% Kit Brook) than grass buffer (11% Blackwater, 19% Kit Brook). Altering cropped field spatial arrangements varied Sy 7-13%. Subcatchment scenarios showed contrasting behaviour, especially in Sy, which was attributed to differences in topography and land cover. Road representation (on-level and sunken) contributed to faster flow arrival and longer event duration. Sunken lanes had a slightly higher impact on 4-year flow compared with roads, but were highly variable in Sy. Urban areas showed flow spatial distribution changes that provoked a 3% Qv increase; whereas Sy were reduced 60% as paved areas are less prone to soil detachment compared with grass. The results show higher impact in Sy than Qv at catchment and sub-catchments scale, suggesting SHETRAN use as a tool for soil erosion prevention and land use management