Global WaterPack – A 250 m resolution dataset revealing the daily dynamics of global inland water bodies

Abstract

Knowledge about inland surface water distribution and its short- to long-term variability is of high importance. Recently many studies have presented interesting results at regional, continental and global scale with 14.25 m to 25 km spatial and 10 day to one year temporal resolution. However, surface inland water bodies can feature temporally rapid spatial changes caused by extreme events, seasonal variability as well as human activity. Such changes can occur on temporal scales which are below the currently existing dynamic water body products. While the daily temporal resolution of available sensors has not been exploited yet. In this study we present an approach which uses the full temporal resolution of the Moderate Resolution Imaging Spectroradiometer (MODIS) to generate a 250 m daily global dataset of inland water bodies based on spectral information and dynamic thresholding. Based on a combination of MODIS Terra and Aqua daily classifications, auxiliary mask layers and temporal interpolation, a global cloud and gap free time series of inland water bodies is produced. The presented results are validated with 321 Landsat images across the globe. The executed validation shows an overall accuracy of 96.3% with 7.8% omission and 0.5% commission error, and a kappa coefficient of 93.3% for pure water pixels. The presented results demonstrate the high potential for different applications requiring information of inland water body dynamics at high temporal resolution. Examples demonstrate that e.g. the filling and emptying of water reservoirs, changes and inundation cycles of natural water bodies as well as freezing and thawing of lakes can be analyzed at a highly detailed temporal scale