Seamless surface water dynamic database (15-days and 10m) for East Africa in 2017-2023

Citation: Wang, Z., Hao, Z., Yang, Q., Mapfumo, P., Nyakudya, E., Du, Y., Yan, X. & Ling, F. (2025). Mapping seamless surface water dynamics over East Africa semimonthly at a 10-meter resolution in 2017–2023 by integrating Sentinel-1/2 data. ISPRS Journal of Photogrammetry and Remote Sensing, 225, 440-460.This dataset provides a seamless, semimonthly 10-meter resolution surface water dataset over East Africa (EA), covering countries such as Tanzania, Uganda, Ethiopia, and others. It was generated using a novel surface water mapping framework specifically designed for Sentinel-1 and Sentinel-2 data.! The extent of original 1-degree grid system (refer to the GridForData file) is incorrect. Please refer to the raster data we provided.eaYear_Month_Day_idYear: Year of acquisition, ranging from 2017 to 2023.Month: Month of acquisition, from 1 to 12.Day: Indicates whether the data was acquired in the first or second half of the month. "01" denotes the first half, and "15" denotes the second half.id: Refers to the grid ID specified in the GridForData file.This dataset offers accurate surface water delineation across East Africa. However, some uncertainties remain:Speckle noise from Sentinel-1 imagery can lead to misclassifications, even after applying filters. These misclassifications are more frequent in the northern parts of EA.A very small amount of data loss occurred during the complex processing workflow.To address these issues, we manually inspected the dataset. Surface water maps with excessive misclassifications or significant data loss were removed. Nevertheless, we recommend using this dataset primarily for analyzing local or national-scale surface water dynamics, rather than as a single-phase surface water reference map.P.S. The Eritrea data will be uploaded later, as we have splited these data into two parts.Don't forget to cite the paper if using this dataset!

引用说明：论文正在撰写中，未来将上传DOI。

本数据集提供了东非地区（East Africa, EA）无缝、半月度的10米分辨率地表水数据集，覆盖坦桑尼亚、乌干达、埃塞俄比亚等国家。该数据集通过专为Sentinel-1和Sentinel-2数据设计的新型地表水制图框架生成。

数据集基于1度网格系统（参见GridForData文件）组织，并随后按国家边界聚合。文件命名遵循以下规则：eaYear_Month_Day_id
年份：数据获取年份，范围为2017年至2023年。
月份：数据获取月份，范围为1至12。
日期：指示数据获取于当月上半月还是下半月，"01"代表上半月，"15"代表下半月。
id：指GridForData文件中指定的网格ID。

该数据集可实现东非地区准确的地表水边界提取，但仍存在一些不确定性：
1. Sentinel-1影像的斑点噪声（speckle noise）即使在应用滤波后仍可能导致错误分类，这类错误在东非北部地区更为常见。
2. 在复杂的处理流程中发生了少量数据丢失。

为解决这些问题，我们对数据集进行了人工检查，移除了存在过多错误分类或显著数据丢失的地表水地图。尽管如此，我们建议该数据集主要用于分析地方或国家尺度的地表水动态，而非作为单期地表水参考地图。

补充说明：厄立特里亚的数据将后续上传，因我们已将这些数据分为两部分。

使用本数据集时请务必引用相关论文！