Enhancing Remote Sensing Imagery Temporal Resolution Using STARFM Data Fusion Approach for Improved Land Surface Monitoring

High-resolution remote sensing imagery plays a critical role in various domains, such as farm-level agricultural operations, environmental monitoring, and natural resource management. However, data with high spatial resolution typically have low temporal resolution, and those with high temporal resolution often lack spatial detail. For example, Landsat 8 and 9 satellites deliver high spatial resolution images with a 30-meter pixel size but suffer from low temporal resolution, with a 16-day revisit cycle. In contrast, satellites like MODIS and VIIRS provide daily images but with a much coarser spatial resolution (375 meters or more), reducing spatial details. Additionally, there is a lack of an intuitive open-source tool, which has been noted by researchers and farmers who require high-resolution images in their work. To address these issues, we have developed an open-source tool that leverages the STARFM algorithm and extends its capabilities to integrate high spatial resolution imagery of Landsat 8 and 9 with high temporal resolution imagery of VIIRS to generate Landsat-like images at a 30-meter resolution for any specified date. While the STARFM algorithm is publicly available, it lacks support for essential pre- and post-processing steps. Our tool addresses this gap by automating these processes and providing a userfriendly interface. Users can specify parameters such as dates, individual Landsat path/row combinations, or entire regions. The system automatically downloads and prepares the required input data and processes the output, delivering high-quality results with minimal user effort. We have evaluated our approach on agricultural fields and validated its performance in regions of Nebraska and Kansas, which have dense agricultural activity. The results demonstrate the effectiveness of our approach over existing methods in generating high-resolution imagery, making it a valuable resource for various users. Advisor: Hongfeng Yu