The Impact of outlier detection to estimate groundwater fluctuations using GRACE satellite data; Case Study: Khuzestan Province, Iran

Abstract

Groundwater aquifers are one of the most significant freshwater resources in the world. Hence, the monitoring of these resources is particularly important for available water resources planning. Piezometric wells have traditionally been used to monitor groundwater. This approach is costly and pointwise, which is not feasible for places with steep topography and mountainous areas. Nowadays, remote sensing techniques are widely used in various fields of engineering as appropriate alternatives to traditional methods. In water resources management, the Gravity Recovery and Climate Experiment (GRACE) satellites can monitor groundwater changes with acceptable accuracies. This paper applied the GRACE satellite data for a 40-month period to assess the variation of the groundwater level in Khuzestan province. The Global Land Data Assimilation System (GLDAS) model was used to counteract the soil moisture effect in final results. The observed data from piezometric wells were pre-processed to detect outliers using the Mahalanobis algorithm in Khuzestan province. At last, the outputs of GRACE were compared with these processed observed data. Despite the relatively small size of the area in question, the results indicated the efficiency of GRACE data (RMSE = 0.8, NRMSE = 0.2) for monitoring the groundwater level changes.

Keywords

• groundwater monitoring
• grace data
• gldas model
• outlier detection
• mahalanobis algorithm