Calibration of a large-scale groundwater flow model using GRACE data: a case study in the Qaidam Basin, China [Calage d’un modèle d’écoulement d’eau souterraine à grande échelle en utilisant les données GRACE: cas du Bassin de Qaidam, Chine] [Calibração de modelo de fluxo subterrâneo usando dados do GRACE: estudo de caso na bacia do Qaidam, China] [Calibración de un modelo de flujo de agua subterránea a gran escala usando datos de GRACE: un caso de estudio en la cuenca de Qaidam, China]

Traditional numerical models usually use extensive observed hydraulic-head data as calibration targets. However, this calibration process is not applicable in remote areas with limited or no monitoring data. This study presents an approach to calibrate a large-scale groundwater flow model using the monthly Gravity Recovery and Climate Experiment (GRACE) satellite data, which have been available globally on a spatial grid of 1° in the geographic coordinate system since 2002. A groundwater storage anomaly isolated from the terrestrial water storage (TWS) anomaly is converted into hydraulic head at the center of the grid, which is then used as observed data to calibrate a numerical model to estimate aquifer hydraulic conductivity. The aquifer system in the remote and hyperarid Qaidam Basin, China, is used as a case study to demonstrate the applicability of this approach. A groundwater model using FEFLOW is constructed for the Qaidam Basin and the GRACE-derived groundwater storage anomaly over the period 2003–2012 is included to calibrate the model, which is done using an automatic estimation method (PEST). The calibrated model is then run to output hydraulic heads at three sites where long-term hydraulic head data are available. The reasonably good fit between the calculated and observed hydraulic heads, together with the very similar groundwater storage anomalies from the numerical model and GRACE data, demonstrate that this approach is generally applicable in regions of groundwater data scarcity. © 2015, Springer-Verlag Berlin Heidelberg.