초록 열기/닫기 버튼
This study proposes the conjunction models of artificial neural network and residual kriging (ANNRK) for estimating the spatial distribution of groundwater level in the Nakdong River Basin, South Korea. The models’ performances are evaluated based on statistical performance indices and graphical comparison and compared with those of inverse distance weighting (IDW), Spline, simple kriging (SK), ordinary kriging (OK), universal kriging (UK) and regression kriging (RK). As a result, kriging and ANNRK models yield better results for spatial groundwater prediction than IDW and Spline models. The RK and ANNRK models with ground surface elevation as the auxiliary information produce significantly better performance compared with other models. Among all models, ANNRK-EXP model achieved the best performance in terms of model performance indices and graphical comparison. Therefore, these results indicate that introducing ground surface elevation as a topographical factor to spatial prediction models can improve the performance of spatial prediction significantly. Furthermore, ANNRK model coupling ANN and residual kriging can be an effective alternative for estimating the spatial distribution of groundwater level accurately.
키워드열기/닫기 버튼
Groundwater level, Spatial prediction, Artificial neural network, Kriging, Topographical auxiliary information
