VFSMOD-based runoff and sediment yield simulation on loess slope with vegetation filter strips

Abstract: The vegetation filter strip is an effective land management practice to reduce the runoff and sediment yield in the sloping orchards in Loess Plateau, China. In the present paper, the feasibility of using vegetative filter strip model (VFSMOD) as a tool to predict the runoff coefficient and sediment yield of vegetative filter strip was evaluated. The experiment was carried out in Yangling, Shannxi. The loess soil was silt loam with bulk density of 1.35-1.40 g/cm3. Two kinds of plots included jujube-grass plots (each was 1.4 m in length, 0.8 m in width and 0.8 m in height) and jujube plots (each was 0.8 m in length, width and height). In 2011, a jujube (Ziziphus jujube Mill.) tree was planted in the center at 0.8 m away from the upper boundary of the plots and grass was planted at the seeding rate of 15 g/m2 in the center at 0.6 m away from the lower boundary of the plots. Grass species were white clover (Triolium repens L.) and birdfoot trefoil (Lotus corniculatus L.) conjoined to source areas. Each type of vegetative filter strip treatments had 3 replications. Simulated rainfalls were applied in 3 development periods (early, middle and late period) of the grass. In each period, 3 rainfall events with different intensities (1.0, 1.2 and 1.5 mm/min) and identical rainfall duration of 60 minutes were designed. The initial soil water content was 0.26-0.29 for all the rainfalls. Grass was cut to 10 cm in height before rainfall. Runoff from each plot was collected for runoff coefficient (the ratio of runoff to rainfall) calculation and sediment yield measurement. Meanwhile, the VFSMOD model was used to simulate the runoff coefficient and sediment yield. The results showed that the bias of runoff coefficient simulation was between -10% and 10%. The Nash coefficient (NE) was higher than 0.9 and the normalized root mean square (NRMSE) was 6.2%. It indicated that the simulation was reliable for runoff. For the sediment yield, the model had the NRMSE of 41.5%. The bias analysis showed that the large error occurred to the birdfoot trefoil at the rainfall intensity of 1.2 and 1.5 mm/min. The VFSMOD model underestimated the sediment yield of birdfoot trefoil filter strip in late September by 28.8% at 1.2 mm/min rainfall intensity and 40.6% at 1.5 mm/min rainfall intensity. The large bias might be attributed to the poor growth status of birdfoot trefoil vegetation filter strip in late September (the stem spacing of 3.07 cm) and poor coverage on the ground. Soil erosion occurred inside the vegetation filter strip in this condition, which was contradicted with the assumption of VFSMOD on soil erosion inside the vegetation filter strip. In the practice, the poor growth status at the storm season should be avoided since it would lead to great sediment yield. Removing these values of birdfoot trefoil, the model had a better simulation with NRMSE of 10.9% and NE of 0.98. Thus, the VFSMOD could well simulate the runoff and sediment yield of soil in the white clover filter strips and the filter strips of birdfoot trefoil at the early and middle development periods. The simulated results showed that the runoff coefficient increased during the development periods of the 2 grass species. The runoff coefficient was 9.2%-29.6%. The runoff coefficient increased with increasing the rainfall intensity, indicating the large rainfall may lead to big runoff. At the same rainfall intensity and development period, the white clover had the lower runoff coefficient (P<0.05) than the birdfoot trefoil. It suggested that the former grass had the better runoff reduction effect after rainfall. The sediment yield of the 2 species was different. At the early period, both were similar in sediment yield. At the late period, the sediment yield of birdfoot trefoil was significantly higher than the white clover (P<0.05). Since the plot with the white clover filter strip had the small runoff and sediment yield in September when rainfall occurs frequently in the loess plateau. Thus, it rather than the birdfoot trefoil was probably suitable to use for filter strip in the loess plateau especially in September.