Neural Networks Technique for Filling Gaps in Satellite Measurements: Application to Ocean Color Observations

Table 2

Accuracy of approximation on independent test set.

	Type of NN (inputs : neurons : outputs)	Bias	RMSE	CC
1	1 : 30 : 1, ()	4.e − 5	2.e − 3	1.00
2	23 : 30 : 1	−4.e − 2	1.76e − 1	0.67
3	Ensemble 6 NNs (23 : 30 : 1)	−2.e − 2	1.72e − 1	0.67
4	24 : 30 : 1	−4.e − 2	1.73e − 1	0.68
5	23 : 30 : 1 (Chl-a ≤ 1)	−2.e − 2	1.11e − 1	0.72
6	Ensemble (23 : 30 : 1) (Chl-a ≤ 1)	−2.e − 2	9.1e − 2	0.79
7	24 : 30 : 1 (Chl-a ≤ 1)	−3.e − 2	1.02e − 1	0.77