ABSTRACT
Remote sensing data with high spatial and temporal resolution are desired to study land surface dynamics. Unfortunately, due to the tradeoff between acquisition frequency and spatial resolution, there is no single sensor could provide this type of data. To address this issue, this chapter introduces two spatiotemporal data fusion techniques, namely NDVI Linear Mixing Growth Model (NDVI-LMGM) and Flexible Spatiotemporal DAta Fusion method (FSDAF), to take advantage of different spatial and temporal characteristics of multi-sensor data to generate synthetic data with rich spatial details and frequent temporal information. NDVI-LMGM is designed to construct high spatial- and temporal-resolution NDVI time series while FSDAF focus on producing high spatial resolution reflectance data by leveraging multi-sensor data of different spatial and temporal resolution. NDVI-LMGM using spatial moving window to address local variability issue, while FSDAF employ thin-plate-spline interpolation method to deal with abrupt land cover type change. Both methods have demonstrated superior performances than existing methods. These spatiotemporal data fusion methods could significantly improve the capability of monitoring rapid surface changes over heterogeneous landscape.
