A Novel Method for Sea Surface Temperature Prediction Based on Deep Learning

Table 1

The best performance results of the proposed model on the two dataset with different predicting lengths (H = 30 days).

Predicting length	East China Sea dataset				Yellow Sea dataset
	No. of neurons	No. of epochs	RMSE	ACC	No. of neurons	No. of epochs	RMSE	ACC
3 days	[14, 12, 13]	89	0.4503	0.9895	[15, 14, 13]	90	0.4425	0.9888
5 days	[15, 16, 16]	92	0.4573	0.9842	[14, 15, 16]	95	0.4558	0.9875
1 week	[16, 17, 18]	90	0.5501	0.9828	[16, 16, 17]	100	0.4593	0.9865
2 weeks	[18, 17, 17]	101	0.6631	0.9815	[17, 17, 18]	99	0.4612	0.9836
1 month	[19, 19, 19]	99	0.6729	0.9803	[19, 20, 19]	105	0.5681	0.9817