Peer Review #1 of "Assessment of crusting effects on interrill erosion by laser scanning (v0.2)"

Background. Crust formation affects soil erosion by raindrop impacted flow through changing particle size and cohesion between particles on the soil surface, as well as surface microtopography. Therefore, changes in soil microtopography can, in theory, be employed as a proxy to reflect the complex and dynamic interactions between crust formation and erosion caused by raindrop-impacted flow. However, it is unclear whether minor variations of soil microtopography can actually be detected with tools mapping the crust surface, often leaving the interpretation of interrill runoff and erosion dynamics qualitative or even speculative.


208
(1) = { 1 , 2 , 3 ,⋯, 500 } -( × 4.8619 + 0.1491) 209 Where, is the height of each data point (mm); is the distance between the laser and soil 210 surface of each data point; is the longest distance between the laser and { 1 , 2 , 3 ,⋯, 500 } 211 soil surface of all the points (namely the zero-level elevation); is in sequence from the 1 st to the 212 500 th scanning point; the constants 4.8619 and 0.1491 are the regression coefficients. 213 To exclude the distorted points around the opening in the center, as well these immediately 214 nearby the outer ring (Figure 1b), only two subsections on each side of the two transects were 215 analyzed: 91≤ ≤ 200 on the left half and 311≤ ≤ 420 on the right half of the X axis. 216 Furthermore, to eliminate the bias introduced by the original slope steepness and to better reflect 217 the relative surface height changes at local scale, the heights of each data point along the two 218 transects were standardized by the slope steepness and the distance from the lowest edge to the 219 targeted point (Eq. 2, 3). The sample protocol was also applied for the subsections of the upper 220 and lower half of the Y axis.

221
(2) ℎ = -(max { 91 , 92 , 93 ,⋯, 200 } -) × 10% 222 or, 223 (3) ℎ = -(min { 311 , 312 , 313 ,⋯, 420 }) × 10% 224 As there were no center opening or edge effects on the four subplots (Figure 1e), all the 500 data 225 points of each subplot were analyzed. Since the bias possibly introduced by the original slope 226 steepness of 10% was systematic and limited to the four subplots with small areas (5 cm  18 227 cm), the of each scanning point inside the subplots was not standardized to slope steepness in 228 this study. Moreover, to quantitatively compare height distributions in the four subplots, all the 229 measured heights were then classified into eight height classes: < 3 mm, 3-4 mm, 4-5 mm, 5-6 230 mm, 6-7 mm, 7-8 mm, 8-9 mm and > 9 mm. To visualize the changes of surface roughness after 231 erosion events, the variogram analysis of the four subplots were conducted using GS+ 232 (Geostatistics for the Environmental Sciences). Kriging regression was applied to give the best 233 linear unbiased prediction of the intermediate values, which were then employed to plot a 2-D 234 version of soil surface height distribution for each subplot. In addition, the height differences 235 between Before, Prewetted and After tests with the least and most eroded replicates were also 236 compared to detect whether the erosion processes were the same, but just operating at different 237 rates, or whether the soil surfaces developed in different ways and thus leading to different 238 erosion processes.   Table 2. 251 A further noteworthy result is the inter-replicate variability, which remained between 15 and 252 39% even after the maximum runoff and erosion were reached (Hu, Fister & Kuhn, 2016). Out 253 of the ten times repeated simulations, the least and most eroded replicate for the CS were CS-4 254 and CS-11, and replicate OS-9 and OS-12 for the OS (Table 3). Typically, the total runoff, soil 255 erosion and SOC loss of the most eroded replicates nearly doubled that on the least eroded 256 replicate, even though they received comparable rainfall amount (Table 3).

257
258 Changes of soil surface elevations across the two transects 259 Figure 2 shows the changes of the soil surface at different conditions (Before, Prewetted, After 260 and Post-dried). After 360 min prolonged rainfall, the CS surface was visibly smoother with 261 extended flat areas and few loose material (10.96 ± 3.01 g m -2 as listed in Table 3), whereas the 262 OS surface was covered to a greater extent by degraded aggregates (43.78 ± 11.40 g m -2 as listed 263 in Table 3). The surface elevation changes of the CS and OS are illustrated in Figure 3. For both 264 soils, the soil surface was lowered after the two rainfall events, with a reduction most evident 265 after the 360 min prolonged rainfall ( Figure 3). 266 Apart from the average changes of surface elevation, Figure 4 further compares the height 267 differences between the least and most eroded replicate under the three conditions (Before, 268 Prewetted and After). The height differences between the Prewetted and Before were more 269 closely clustered than that between the After and Before (Figure 4a vs. 4b, 4c vs. 4d). 270 Specifically, for the height differences between After and Before on the CS (Figure 4b), the most 271 eroded replicate CS-11 had more negative height differences (point clouds < 0) than increases 272 (point clouds > 0) when compared to the least eroded replicate CS-4. Similar, but more frequent 273 negative height differences, were observed on the most eroded replicate OS-12 than that of the 274 least eroded replicate OS-9. Moreover, the point clouds were also more concentrated under the 275 1:1 ratio for both the pair-comparison of CS-4 against CS-11, and that of OS-9 against OS-12 276 (Figure 4b, 4d).

277
278 Changes of soil surface elevations on the four subplots 279 Figure 5 shows that the height distribution of all the four subplots on the CS was greater and 280 more variable than that on the OS. While the surface of both soils progressively approached the 281 lower height classes over time ( Figure 5), this transition was much more skewed on the CS, 282 especially after the 360 min prolonged rain ( Figure 5d). The differences of surface elevation 283 between the two soils are also illustrated by the 2D classification in Figure 6. The surface 284 elevation of the two soils was quite similar when the soils were dry before the rainfall. After the 285 30 min of prewetting, the surface of the OS still showed a strong contrast of high and low 286 elevation (Figure 6d, 6e), whilst the surface of the CS was noticeably flattened (Figure 6a, 6b). 287 After the 360 min prolonged rain, the surface height of the CS became lowered to less than 4 mm 288 (Figure 6c), that of the OS remained rougher between 4.5-and 6.5-mm (Figure 6f). 294 that the differences in erosion and sediment properties observed between the two soils tested in 295 this study were associated with crust formation. They hypothesized that the different erosional 296 responses of the similarly textured CS and OS (Table 2) reflect the influence of aggregate 297 stability (Table 1) (2013): 302 greater aggregate stability of the OS slowed aggregate breakdown, maintaining roughness and 303 sediment size for longer, thus also resisting raindrop impact for longer than on the CS. Such 304 potential effects of aggregate stability on surface deformation are reflected by the more 305 pronounced skewing toward the smaller height classes on the CS surface than on the OS surface 306 ( Figure 5). Consequently, after the prolonged 360 min rainfall, the CS surface height was 307 noticeably flattened to be less than 4 mm (Figure 5d, 6c), whereas the soil surface of the OS was 308 much rougher between 4.5 mm and 6.5 mm ( Figure 6f) and still interspersed by more loose 309 material (Figure 2g, 2h, Table 3). The declining soil erosion rates on the CS after its runoff rate 310 exceeded 12.9 mm h -1 (Table 2) indicate that runoff had overcome transport limitation and 311 reached a supply-limited process after fine, light and loose particles had been selectively eroded 312 (flattened surface in Figure 5d, 6c). Judging from the abundant loose materials remaining on the 313 OS plots (Figure 2g, 2h, Table 2, as well in Hu, Fister & Kuhn (2013)), it would also eventually 314 reach a supply-limited condition as runoff grew more competent over time by removing loose 315 particles and exposing cohesive crust. This deduction is also supported by the delayed decline of 316 the ERsoc on the OS once runoff rates had stabilized (Table 2).  Figure 4b indicate that the scanned surface was rougher after 360 min of rainfall 329 (After) than before, and such roughening was more pronounced in the least eroded replicate CS-330 4 than the most eroded replicate CS-11. (2) The concentrated cloud in section II of Figure 4b 331 displays that certain parts of the least eroded replicate CS-4 became rougher after the prolonged 332 rainfall than Before (positive height differences from 0 mm to 6 mm), whereas some sections of 333 the most eroded replicate CS-11 were flattened after the prolonged rainfall events (negative 334 height differences from -6 mm to 0 mm). A similar, but even more obvious concentration of the 335 point cloud can be observed in section II of Figure 4d, illustrating the divergent development of 336 soil surface elevation between the least and most eroded replicates of the OS.
(3) The negative 337 height differences in section III under the 1:1 ratio line of Figure 4b suggest that those areas were 338 smoother after the 360 min rainfall, and the most eroded CS-11 was more smoothened than the 339 least eroded CS-4. All the three scenarios of more pronounced flattening (Figure 4b, 4d) on the 340 most eroded replicates CS-11 and OS-12 are consistent with their nearly doubled runoff and soil 341 loss as opposed to the least eroded replicates CS-4 and OS-9 (Table 3). 342 The covariance between the observed erosional response and the laser data illustrates the 343 effectiveness of millimeter-resolution laser scanning to detect the minor topographic changes. On 344 the one hand, this not only confirms the decisive role of soil properties such as organic matter 345 content and aggregate stability in crust formation, soil erosional responses and sediment 346 properties (Table 1, 2) . On the other, the results clearly illustrate that 347 laser data also help to uncover the causes of the 15%-39% inter-replicate variability among the 348 ten simulated rainfall events (Hu, Fister & Kuhn, 2016) by effectively distinguishing the rates of 349 surface feature development between the most and least eroded replicates ( Figure 4,