Improving the prediction performance of leaf water content by coupling multi-source data with machine learning in rice (Oryza sativa L.)

Table 4 The ND_(1287,1673) models predicting LWC after integrating with multi-source data

Growth stage	Model	Multiple linear regression equation	Model Precision	Prediction Precision	RMSE
Booting	ND + SPAD	LWC = 1.91ND + 0.02SPAD − 0.19	0.60*	0.53*	0.04
Booting	ND + Fv/Fm	LWC = 2.75ND + 1.99Fv/Fm − 1.16	0.61*	0.62**	0.03
Booting	ND + CWSI	LWC = 1.22ND − 0.34CWSI + 0.66	0.69**	0.65**	0.03
Flowering	ND + SPAD	LWC = 1.48ND + 0.01SPAD + 0.19	0.71**	0.60*	0.02
Flowering	ND + Fv/Fm	LWC = 2.10ND + 0.37Fv/Fm + 0.18	0.74**	0.64*	0.02
Flowering	ND + CWSI	LWC = 1.49ND − 0.08CWSI + 0.57	0.75**	0.69**	0.01
Initial grain filling	ND + SPAD	LWC = 1.15ND + 0.01SPAD + 0.16	0.67**	0.64**	0.01
Initial grain filling	ND + Fv/Fm	LWC = 1.58ND + 0.88Fv/Fm − 0.16	0.68**	0.66**	0.01
Initial grain filling	ND + CWSI	LWC = 1.11ND − 0.20CWSI + 0.73	0.61*	0.55*	0.01
Middle grain filling	ND + SPAD	LWC = 0.60ND + 0.002SPAD + 0.56	0.56*	0.49*	0.01
Middle grain filling	ND + Fv/Fm	LWC = 0.70ND + 1.19Fv/Fm − 0.31	0.59*	0.56*	0.01
Middle grain filling	ND + CWSI	LWC = 0.59ND − 0.10CWSI + 0.67	0.57*	0.44	0.01

LWC: leaf water content; ND_(1287,1673): normalized difference index; SPAD: chlorophyll content; Fv/Fm: maximum photochemical efficiency; CWSI: crop water stress index; * and ** indicate significant correlation at 5% and 1% probability level, respectively

ISSN: 1746-4811