Prediction of the Reference Evapotranspiration Using a Chaotic Approach

Table 4

Prediction results for daily reference evapotranspiration of the stations selected in China by chaotic model, AR model, and BP neural network model.

	Location	Mean absolute error (MAE)	Root mean square error (RMSE)	Correlation coefficient (CC)	Modified coefficient of efficiency	Optimal embedding dimension (m)	Optimal number of neighbors
Chaotic Model	Baotou	0.1378	0.1999	0.9965	0.913	4	72
	Zhangbei	0.2023	0.2799	0.9887	0.8709	4	76
	Kaifeng	0.1962	0.3095	0.9957	0.8679	4	80
	Shaoguan	0.2358	0.3054	0.9784	0.8078	4	68
BP Model	Baotou	0.203	0.098	0.992	0.872
	Zhangbei	0.189	0.089	0.989	0.894
	Kaifeng	0.198	0.076	0.997	0.871
	Shaoguan	0.209	0.082	0.993	0.834
AR Model	Baotou	0.573	0.665	0.876	0.652
	Zhangbei	0.612	0.672	0.889	0.673
	Kaifeng	0.553	0.589	0.897	0.709
	Shaoguan	0.557	0.699	0.903	0.655