Digital Mapping of Soil Drainage Classes Using Multitemporal RADARSAT-1 and ASTER Images and Soil Survey Data

Table 6

Variables included in the drainage classification model using a forward stepwise discriminant analysis for different land use types (LUT).

Variables in the model	Wilks’ Lambda	F to remove	P value	1−Toler (1−R−Sqr.)
LUT = N1 (bare soils); n training = 59
NDVI	0.61	2.97	0.03	0.05
05/05/03 S2a (t1)	0.55	1.62	0.18	0.21
05/05/10 S1a (t2)	0.55	1.47	0.23	0.16
05/06/10 S2d (t5)	0.53	1.13	0.35	0.10
Total	0.49	2.00	0.01
LUT = N2 (bare soil with sparse vegetation); n training = 24
NDII	0.45	7.68	0.01	0.55
05/05/03 S2a (t1)	0.24	0.65	0.54	0.29
05/06/10 S2d (t5)	0.41	6.37	0.01	0.66
05/05/10 S1a (t2)	0.37	5.11	0.02	0.54
05/05/24 S3d (t3)	0.32	3.32	0.06	0.26
NDVI	0.26	1.31	0.30	0.23
Total	0.22	2.81	0.01
LUT = P (pasture and forage); n training = 96
NDVI	0.93	8.39	0.00	0.00
05/05/27 S2a (t4)	0.79	2.29	0.08	0.00
Total	0.73	5.17	<0.01
LUT = C (emerged crops); n training = 22
05/05/27 S2a (t4)	0.57	2.93	0.08	0.42
05/05/24 S3d (t3)	0.48	1.21	0.33	0.07
NDWI	0.51	1.77	0.20	0.00
05/05/10 S1a (t2)	0.48	1.19	0.33	0.45
Total	0.41	2.10	0.07
LUT = F (forest); n training = 86
NDVI	0.84	6.74	0.00	0.11
05/05/03 S2a (t1)	0.68	1.60	0.18	0.08
05/05/27 S2a (t4)	0.67	1.41	0.24	0.10
05/05/10 S1a (t2)	0.66	1.05	0.39	0.08
Total	0.62	2.47	<0.01
LUT = H (wetland); n training = 40
05/06/10 S2d (t5)	0.66	4.17	0.02	0.03
NDVI	0.65	3.69	0.04	0.03
05/05/24 S3d (t3)	0.57	1.29	0.29	0.04
05/05/27 S2a (t4)	0.57	1.11	0.34	0.02
Total	0.53	2.15	<0.01