Introduction
Zhang et al. (2020) report results from an experimental design that was modified from the design reported by Zhang (2019). The experimental design involved the use of two flumes (1.8 x 0.5 x 0.1 m) set side by side with a 2-5 cm gap between them. The flumes were packed with soil having 20.8 % sand, 58.6 % silt, and 20.6 % clay. Rain was applied for one hour using a nozzle type simulator whose intensity was controlled by the frequency of nozzle oscillation. 3 rainfall intensities (60, 90, 120 mm h-1) and 3 slope gradients ( 9, 18, and 27%) were used. Splashed material was collected in the slot while sediment discharged in runoff was collected at the bottom of the flumes. The soil was the same as used by Zhang (2019) but packed to a bulk density of 1.4 g cm-3 as opposed to 1.2 g cm-3 used previously. The 1.4 g cm-3bulk density was chosen after trying several alternatives because the authors perceived that it created a detachment-limited regime under high intensities and slopes but a transport-limited regime under lower intensities and slopes.
Two treatments were used in association with the variations 3 variations in rainfall intensity and 3 variations in slope gradients. In one treatment, a screen was suspended 5 cm above the soil surface. In the other, a tarp was placed over the surface to prevent upslope runoff from flowing over the exposed surface. In the first 15 minutes of experiments lasting 1 hour, the bottom quarter of the flumes was exposed to rainfall produced by the rainfall simulator. In the 2nd 15-minute period, erosion by the rainfall occurred on the bottom half of the flumes. Three-quarters of the flume area was exposed in the 3rd 15-minute period with the whole area being exposed in the 4th 15-minute period. In the first hour, 60 mm hr-1 rainfall intensity was used. In the 2nd hour, the exposure sequence was repeated with 90 mm hr-1 rainfall intensity, and 120 mm hr-1 rainfall intensity in the 3rdhour. The screen experiments were designed so that surface water flow conditions did not vary in the bottom section of the 1.8 m long surface as the number of exposed sections varied. Consequently, the transport capacity of the rain-impacted flows in the bottom section remained constant during each 1 hour rainfall event. In contrast, the tarp experiments were designed so that the transport capacity of rain-impacted flows in the bottom section varied as the number of exposed sections varied. The two treatments were applied on 9 %, 18 % and 27 % slopes. Newly prepared surfaces were used for a sequence of 3 one hour rainfalls varying in intensity from 60 mm hr-1 to 120 mm hr-1 and totalling 270 mm on for each slope gradient. Sediment-laden runoff was collected every 3 minutes
In analysing the data for splash in these experiments, Zhang et al assumed that following Zhang (2019), splashed material per unit area of the slot was equal to the detached and/or previously detached soil material per unit area of the plot. In order to make direct comparisons between event total wash loss and splash, the total splash amounts were upscaled according to ratios of respective area of the flumes and the slot and those values reflected the total loose material available on the soil surface during the entire event. Given that, Zhang at al assumed that if these measured splash amounts exceeded the amounts washed, then transport-limiting conditions existed. Conversely, if the washed amounts exceeded the measured splash amounts, then flow detachment had occurred. Based on this assumption, they concluded that flow detachment occurred in all the screen experiments and in all the tarp experiments on the 27 % slope and the tarp experiment with 120 mm h-1 rain on the 18 % slope.