Table 6. A brief summary of RF modelling process for hard70 data using various FS methods and predictive variables.

1) models 1–25 based on the AVI using 20 variables; 2) models 26–38 based on the AVI using 41 variables; 3) models 39–49 based on KIAVI using 41 variables; and 4) models 50–52 based on the Boruta with the maximal number of importance source runs of 2000, 100 and 5000, and model 53 based on the RRF using 41 variables. The model fit is the predictive accuracy (ccr) of training samples by each RF model developed. The corresponding predictor for each number is listed in Table 1.

Model	Modelling.process	Predictors	No.predictors	Model.fit
1	All 20 predictive variables	All 20 variables	20	100
2	model 1: -relief	1–7, 9–20	19	100
3	model 2: -northing	1, 3–7, 9–20	18	100
4	model 3: -bs13	1, 3–7, 9–11, 13–20	17	100
5	model 4: -bs27	1, 3–7, 9–11, 13–14, 16–20	16	100
6	model 5: -slope	1, 3, 4–7, 10–11, 13–14, 16–20	15	100
7	model 6: -bs25	1, 3–7, 10–11, 13, 16–20	14	100
8	model 7: -bs21	1, 3–7, 10–11, 16–20	13	100
9	model 8: -bathy.moran	1, 3–4, 6–7, 10–11, 16–20	12	100
10	model 9: -surface	1, 3–4, 6–7, 11, 16–20	11	100
11	model 10: -bathy	1, 3, 6–7, 11, 16–20	10	100
12	model 11: -tpi	1, 3, 6–7, 16–20	9	100
13	model 12: -bs35	1, 3, 6–7, 16, 18–20	8	100
14	model 13: -variance	1, 3, 6–7, 16, 18, 20	7	100
15	model 14: -bs.moran	1, 3, 6–7, 16, 18	6	100
16	model 15: -bs32	1, 3, 6–7, 18	5	100
17	model 16: -easting	3, 6–7, 18	4	100
18	model 17: -profile.curv	3, 6, 18	3	96.43
19	model 18: -homogeneity	3, 6	2	95
20	model 19: -planar.curv	3	1	91.43
21	model 11: +bs21	1, 3, 6–7, 11, 13, 16–20	11	100
22	model 21: +bathy.moran	1, 3, 5–7, 11, 13, 16–20	12	100
23	model 21: -profile.curv	1, 3, 6, 11, 13, 16–20	10	100
24	model 21: -bs.moran	1, 3, 6–7, 11, 13, 16–19	10	100
25	model 24: -variance	1, 3, 6, 7, 11, 18, 13, 16, 17	9	100
26	All 41 predictors	1–11, 18–20, bs10-bs36	41	100
27	model 26:—bs25, bs10:bs14	1–11, 18–20, bs15-bs36	35	100
28	model 27:—bathy, relief, bs24,bs26, bs27	1–3, 5–7, 9–11, 18–20, bs15-bs23, bs28-bs36	30	100
29	model 28:—bs20, bs29, bs30, bs36	1–3, 5–7, 9–11, 18–20, bs15-bs19, bs21-bs23, bs28, bs31-bs35	26	100
30	model 29:—nothing and bs28	1, 3, 5–7, 9–11, 18–20, bs15-bs19, bs21-bs23, bs31-bs35	24	100
31	model 30:—slope and bs34	1, 3, 5–7, 10, 11, 18–20, bs15 to bs19, bs21-bs23, bs31-bs33, 17	22	100
32	model 31-—bs15, bs16, bs19, bs21, bs32, bs33 and bs 35	1, 3, 5–7, 10, 11, 18–20, bs17, bs18, bs22, bs23, bs31	15	100
33	model 32-—bathy.moran, surface, bs17 and bs22	1, 3, 6–7, 11, 18–20, bs18, bs23, bs31	11	100
34	model 33-—homogeneity	1, 3, 6–7, 11, 19–20, bs18, bs23, bs31	10	100
35	model 34-—tpi	1, 3, 6–7, 19–20, bs18, bs23, bs31	9	100
36	model 35-—bs23	1, 3, 6–7, 19–20, bs18, bs31	8	100
37	model 36-—bs18	1, 3, 6–7, 19–20, bs31	7	100
38	model 37-—bs.moran	1, 3, 6–7, 19, bs31	6	100
39	Variables for model 24, 33, and model 40 for hard90	1–3, 6–7, 10–11, 13, 18–20, bs10, bs18, bs23, bs28-bs36	23	100
40	model 39-—bs10	1–3, 6–7, 10–11, 13, 18–20,bs18, bs23, bs28-bs36	22	100
41	model 40-—bs28, bs30 abd bs36	1–3, 6–7, 10–11, 13, 18–20, bs18, bs23, bs29, bs31-17	19	100
42	model 41-—northing	1, 3, 6–7, 10–11, 13, 18–20, bs18, bs23, bs29, bs31-17	18	100
43	model 42-—bs35	1, 3, 6–7, 10–11, 13, 18–20, bs18, bs23, bs29, bs31-bs34	17	100
44	model 43-—bs29 and bs34	1, 3, 6–7, 10–11, 13, 18–20,bs18, bs23, bs31-bs33	15	100
45	model 44-—bs21	1, 3, 6–7, 10–11, 18–20,bs18, bs23, bs31-bs33	14	100
46	model 45-—bs33	1, 3, 6–7, 10–11, 16, 18–20,bs18, bs23, bs31	13	100
47	model 46-—surface	1, 3, 6–7, 11, 16, 18–20,bs18, bs23, bs31	12	100
48	model 47-—tpi	1, 3, 6–7, 16, 18–20,bs18, bs23, bs31	11	100
49	model 47-—tpi and bs31	1, 3, 6–7, 16, 18–20, bs18, bs23	10	100
50	27 predictors	1, 3, 7, 11, bs12, bs13, bs16-bs36	27	100
51	model 50-—easting, bs12	2–3, 7, 11, 20, bs13, bs16-bs36	25	100
52	model 50- +bs10, bs14	1, 3, 7, 11, bs10, bs12-bs14, bs16-bs36	29	100
53	31 predictors	1–3, 5, 7–11, 18–20, bs15, bs17-bs18, bs20-bs25, bs27-bs36	31	100