High-resolution spatiotemporal satellite data are useful in a variety of fields such as ecology, hydrology, meteorology and epidemiology for monitoring changes over a large region of the earth. However, the usability of most satellite data products are often hampered by large amounts of missing values caused by clouds, cloud shadows, and other atmospheric conditions. This paper introduces a general spatiotemporal satellite data imputation method based on sparse functional data analysis techniques by treating the satellite data observed in different years as repeated measurements of a latent spatiotemporal process. To address the computational issue associated with large images, we introduce a hierarchical multi-resolution imputation algorithm which allow efficient imputation. The methodology is demonstrated using data from Landsat satellites and the Moderate Resolution Imaging Spectroradiometer (MODIS) daily land surface temperature (LST) data.