Global 30m 2000-2014 Surface Water Dynamics Map Derived from All Landsat 5, 7, and 8

Abstract:

Water is critical for human life, agriculture, and ecosystems. A better understanding of where it is and how it is changing will enable better management of this valuable resource and guide protection of sensitive ecological areas. Global water maps have typically been representations of surface water at one given time. However, there is both seasonal and interannual variability: rivers meander, lakes disappear, floods arise. To address this ephemeral nature of water, in this study University of Maryland has developed a method that analyzes every Landsat 5, 7, and 8 scene from 1999-2015 to produce global seasonal maps (Winter, Spring, Summer, Fall) of surface water dynamics from 2000-2014. Each Landsat scene is automatically classified into land, water, cloud, haze, shadow, and snow via a decision tree algorithm. The land and water observations are aggregated per pixel into percent occurrence of water in a 3 year moving window for each meteorological season. These annual water percentages form a curve for each season that is discretized into a continuous 3 band RGB map. Frequency of water observation and type of surface water change (loss, gain, peak, or dip) is clearly seen through brightness and hue respectively. Additional data layers include: the year the change began, peak year, minimum year, and the year the change process ended. Currently these maps have been created for 18 1°x1° test tiles scattered around the world, and a portion of the September-November map over Bangladesh is shown below. The entire Landsat archive from 1999-2015 will be processed through a partnership with Google Earth Engine to complete the global product in the coming months. In areas where there is sufficient satellite data density (e.g. the United States), this project could be expanded to 1984-2015. This study provides both scientific researchers and the public an understandable, temporally rich, and globally consistent map showing surface water changes over time.