A unified spatial-spectral-temporal fusion model using Landsat and MODIS imagery

Due to the tradeoff between spatial, spectral, and temporal resolution, there is no such a unified sensor that can produce images with fine spatial-, temporal-, and spectral-resolution simultaneously. However, facing an emerging need of fine spatial details, frequent coverage, and multi-spectral remotely sensed data for global change detection, image fusion technique that blends multi-sensors' characteristics to generate synthetic data with fine resolutions has aroused great interest within the remote sensing community. Currently image fusion can be generally divided into two major categories: spatial and spectral fusion, and spatial and temporal fusion. During the past decades, although there is much achievement made for each category, there has been limited study addressing integration simultaneously. This article proposes a unified spatial-temporal-spectral blending model using the Landsat LTM+ and MODIS images to predict a synthetic "daily" Landsat-like data with 15m spatial resolution. This model is implemented in two stages. First, the spatial resolution of Landsat LTM+ data is enhanced based on an Improved Adaptive-IHS approach; second, the MODIS and enhanced Landsat LTM+ data are fused by Spatial and Temporal Adaptive Reflectance Fusion Model (STARFM) to generate final synthetic data. This model provides a practical application aiming at fusing synthetic spatial, temporal, and spectral information. The results of tests with both simulated and real experiments show that the model can accurately capture the general trend of changes for the predicted period, and enhance spatial resolution of the data while preserving the original spectral information at the same time. Potential applications using synthetic fusion data with fine-resolutions are addressed.