Water & Wetlands

Water & Wetlands is terraPulse's surface-water data product. The terraPulse Surface Water dataset maps inundation of the ground surface by liquid, unfrozen water, either categorically at any point in time or proportionally over longer intervals, globally from 1984 to the present. Its standard data products are Surface Water Cover (percent of area, up to 30-m resolution and up to monthly time steps), Hydroperiod & inundation frequency (percent of observations detected as inundated, annual) and Inundation year & age (years since the most recent inundation, annual), each with a latency of the previous year. Through the terraPulse Land Cover Layer, surface water is also available as a binary 1-metre layer where high-resolution imagery exists. Certainty is reported as root mean square error for continuous products and probability of class membership for categorical ones.

The layers are mined from satellite measurements with machine learning; a peer-reviewed study describes the dataset as Landsat-derived. terraPulse was founded in 2014 by the University of Maryland Global Land Cover Facility team that produced the first high-resolution global maps of tree and water cover.

The data are delivered as gridded datasets, in the terraView web dashboard, and through the terraServe API (CSV, JSON and GeoTIFF, plus XYZ map tiles for ArcGIS, QGIS, Leaflet and OpenLayers). terraPulse names flood-risk exposure, aquaculture and paddy-agriculture detection, and wetland habitat identification as applications. Its published case studies cover the Chinese Academy of Sciences studying shrinking water bodies in Central Asia, the UN Green Climate Fund studying wetland urbanization in Laos, and University of Michigan scientists studying drought in the western US.