Broad scale monitoring of inland waters is essential to research on carbon and water cycles, and for application in the monitoring of disasters including floods and droughts on various spatial and temporal scales. Satellite remote sensing using spatiotemporal data fusion (STF) has recently attracted attention as a way of simultaneously describing spatial heterogeneity and tracking the temporal variability of inland waters. However, existing STF approaches have limitations in describing abrupt temporal changes, integrating “dissimilar” datasets (i.e., fusions between microwave and optical data), and compiling long-term, frequent STF datasets. To overcome these limitations, in this study we developed and evaluated a lookup table (LUT)-based STF, termed database unmixing (DBUX), using multiple types of satellite data (AMSR series, MODIS, and Landsat), and applied it to semi-arid seasonal wetlands in Namibia. The results show that DBUX is: 1) flexible in integrating optical data (MODIS or Landsat) with microwave (AMSR series) and seasonal (day of year) information; 2) able to generate long-term, frequent Landsat-like datasets; and 3) more reliable than an existing approach (spatial and temporal adaptive reflectance fusion model; STARFM) for tracking dynamic temporal variations in seasonal wetlands. Water maps retrieved from the resulting STF dataset for the wetlands had a 30-m spatial resolution and a temporal frequency of 1 or 2 days, and the dataset covered from 2002 to 2015. The time series water maps accurately described both seasonal and interannual changes in the wetlands, and could act as a basis for understanding the hydrological features of the region. Further studies are required to enable application of DBUX in other regions, and for other landscapes with different satellite sensor combinations.
All Science Journal Classification (ASJC) codes