Table 3.

Predictions of SVM models throughout multiple growth stages

Year	Bands	Sample	Level 0	Level 1	Level 2	Level 3	Level 4	True
2019	All	Level 0	16	0	0	0	0	16
		Level 1	0	11	3	0	0	14
		Level 2	0	0	27	2	0	29
		Level 3	0	0	5	15	1	21
		Level 4	0	0	0	2	17	19
		Prediction	16	11	35	19	18	99
	491 nm	Level 0	15	1	0	0	0	16
	671 nm	Level 1	1	11	2	0	0	14
		Level 2	0	1	24	4	0	29
		Level 3	0	0	5	16	0	21
		Level 4	0	0	0	4	15	19
		Prediction	16	13	31	24	15	99
	491 nm	Level 0	15	1	0	0	0	16
	668 nm	Level 1	3	8	3	0	0	14
		Level 2	0	0	24	5	0	29
		Level 3	0	0	5	16	0	21
		Level 4	0	0	0	2	17	19
		Prediction	18	9	32	23	17	99
2021	All	Level 0	13	0	0	0	0	13
		Level 1	0	34	0	0	0	34
		Level 2	0	10	3	2	0	15
		Level 3	0	0	1	19	1	21
		Level 4	0	0	0	3	10	13
		Prediction	13	44	4	24	11	96
	491 nm	Level 0	13	0	0	0	0	13
	671 nm	Level 1	2	32	0	0	0	34
		Level 2	0	9	0	6	0	15
		Level 3	0	0	0	21	0	21
		Level 4	0	0	0	2	11	13
		Prediction	15	41	0	29	11	96
	491 nm	Level 0	12	1	0	0	0	13
	668 nm	Level 1	1	33	0	0	0	34
		Level 2	0	8	2	5	0	15
		Level 3	0	0	1	20	0	21
		Level 4	0	0	0	2	11	13
		Prediction	13	42	3	27	11	96