Spatiotemporal Approaches for Quality Control and Error Correction of Atmospheric Data through Machine Learning

Table 4

Comparison of MLQC-HT results on raw data and interpolated data.

Method	Weather element	Raw data					Interpolated data
		Correlation coefficient	RMSE	3σ error		Time (sec)	Correlation coefficient	RMSE	3σ error		Time (sec)
				#err/#total	Rateerr(%)				#err/#total	Rateerr(%)
SVR	Temperature	1.000	0.0589	292/10050	2.91	1464	1.000	0.0556	287/9665	2.97	1402
	Humidity	0.9995	0.6456	148/10050	1.47	2378	0.9997	0.4712	129/6429	2.01	798
	UV-rays	0.9919	1.7307	4508/10050	44.86	2202	0.9919	1.7307	4508/10050	44.86	2202
	PM2.5	0.9209	8.1594	390/10050	3.88	888	0.9834	3.6107	388/10013	3.87	928
	Solar radiation	0.9062	0.0049	138/10050	1.37	1126	0.9062	0.0049	138/10050	1.37	1126
	u	0.7576	0.5815	432/10050	4.30	717	0.7739	0.3589	332/7254	4.58	245
		0.7639	0.7385	565/10050	5.62	704	0.7837	0.4251	436/7254	6.01	219
MLP	Temperature	0.9999	0.0662	1710/10050	17.01	53	0.9999	0.0648	493/9665	5.10	46
	Humidity	0.9993	0.7906	1585/10050	15.77	53	0.9996	0.5484	281/6429	4.37	31
	UV-rays	0.9902	1.9707	3688/10050	36.70	49	0.9902	1.9707	3688/10050	36.70	49
	PM2.5	0.9039	8.9938	772/10050	7.68	54	0.9805	3.9132	405/10013	4.04	50
	Solar radiation	0.9016	0.0051	88/10050	0.88	50	0.9016	0.0051	88/10050	0.88	50
	u	0.6273	0.7208	1006/10050	10.01	52	0.654	0.4535	419/7254	5.78	34
		0.7256	0.7901	1420/10050	14.13	53	0.7075	0.4935	478/7254	6.59	16
DT	Temperature	0.9943	0.6297	7680/10050	76.42	4	0.9945	0.6095	7716/9665	79.83	2
	Humidity	0.9947	2.1775	6752/10050	67.18	5	0.9937	2.2225	4315/6429	67.12	1
	UV-rays	0.9796	2.7371	7033/10050	69.98	2	0.9796	2.7371	7033/10050	69.98	2
	PM2.5	0.8968	9.2503	3466/10050	34.49	8	0.9704	4.7958	3398/10013	33.94	2
	Solar radiation	0.9181	0.0045	136/10050	1.35	2	0.9181	0.0045	136/10050	1.35	2
	u	0.6854	0.6418	794/10050	7.90	5	0.7202	0.391	501/7254	6.91	2
		0.7130	0.7959	1158/10050	11.52	4	0.7337	0.464	613/7254	8.45	2

SVR is support vector regression, MLP is multilayer perceptron, and DT is decision table.