Research on Monitoring Methods for the Appropriate Rice Harvest Period Based on Multispectral Remote Sensing

Table 4

Regression test results of single-band spectral reflectance and physiological indexes of rice.

Prediction method	Variety	Spectral band	Moisture content		Thousand-grain weight
			Determination coefficient (R^2)	Root mean square error (RMSE)	Determination coefficient (R^2)	Root mean square error (RMSE)
BP neural network	South Japonica 46	550	0.40	0.24	0.42	0.95
		660	0.36	0.25	0.78	0.88
		735	0.65	0.23	0.84	0.84
		790	0.49	0.24	0.77	0.88
	South Japonica 5055	550	0.69	0.20	0.80	0.57
		660	0.33	0.23	0.47	0.66
		735	0.65	0.20	0.87	0.51
		790	0.88	0.19	0.65	0.64
SVM	South Japonica 46	550	0.38	0.39	0.36	1.63
		660	0.28	0.39	0.35	1.57
		735	0.44	0.35	0.33	1.43
		790	0.54	0.28	0.59	1.30
	South Japonica 5055	550	0.43	0.39	0.63	1.13
		660	0.40	0.22	0.42	1.10
		735	0.53	0.33	0.78	0.88
		790	0.68	0.33	0.73	0.85
Decision tree	South Japonica 46	550	0.98	0.35	0.97	3.30
		660	0.80	0.36	0.90	3.67
		735	0.84	0.36	0.97	3.76
		790	0.91	0.35	0.90	3.72
	South Japonica 5055	550	0.98	0.35	0.91	2.61
		660	0.80	0.36	0.79	2.66
		735	0.84	0.36	0.91	2.72
		790	0.91	0.35	0.94	2.61