High-frequency radar (HFR)-based observations are extremely useful for studying coastal process due to their fine spatio-temporal resolution. However, these observations often suffer from gaps, caused by a range of operational and external factors, which pose major challenges during their analysis. Herein, we propose the use of spatio-temporal kriging as a means to effectively fill such gaps. This method has been applied to HFR derived surface-current measurements made from the Tamil Nadu High-Frequency Radar station along the eastern coast of India. The proposed method was able to restore surface-current fields and reconstruct the mesoscale and submesoscale circulation features from sparse observations. The proposed method was cross-validated using the leave-one-out cross-validation and the leave-one-day-out cross-validation schemes which, when compared against the accuracy bounds of HFR-based surface-current observations, demonstrated the ability of the method to reconstruct surface-current fields with reasonable fidelity.