Thapa Narayan, Nepali Sujan, Shrestha Raman, Sanjel Suman
International Center for Integrated Mountain Development, Nepal.
Nepal Open University, Nepal.
Data Brief. 2025 Aug 26;62:112010. doi: 10.1016/j.dib.2025.112010. eCollection 2025 Oct.
In mountainous countries like Nepal, floods are a major challenge due to complex topography, intense snowmelt, and highly variable monsoon rainfall that drive frequent flooding events. This study focuses on the Hilly and Himalayan regions of Nepal, where flood monitoring and risk management are increasingly important for safeguarding vulnerable communities and infrastructure. This study presents a high-resolution, time-series flood extent dataset derived from the Copernicus Sentinel-2 Level-2A imagery at a 10-meter spatial resolution, covering the years 2019 to 2023. Flood mapping was performed using the Normalized Difference Vegetation Index (NDVI) combined with region-specific thresholding. NDVI values below 0 represent open water, while values between 0 and 0.1 often indicate mud, bare soil. A threshold of NDVI <0.019 was applied to identify flood-affected areas in the hilly region to capture the debris flow type flood, whereas NDVI <0 was used for the Himalayan region, because of the presence of snow and water that complicated classification due to their spectral similarity with other features. Snow-covered areas were masked using the Copernicus Global Land Cover dataset to improve accuracy in the high altitude zones. Data processing was performed on the Google Earth Engine (GEE) platform. Monsoon-season image composites were generated after applying cloud masking using the Scene Classification Layer (SCL), and temporal cloud gaps were filled using post-monsoon imagery to ensure continuous temporal data. The resulting flood extent maps reveal consistent spatial patterns and provide critical data for flood forecasting, risk-sensitive land use planning, and interdisciplinary studies. Despite challenges with cloud interference and complex terrain, this dataset offers valuable insights into flood dynamics across Nepal's mountainous landscape.
在尼泊尔这样的多山国家，由于地形复杂、融雪强烈以及季风降雨变化极大，洪水成为一项重大挑战，这些因素导致频繁的洪水事件。本研究聚焦于尼泊尔的丘陵和喜马拉雅地区，在这些地区，洪水监测和风险管理对于保护脆弱社区和基础设施日益重要。本研究呈现了一个高分辨率的时间序列洪水范围数据集，该数据集源自哥白尼哨兵 - 2二级A图像，空间分辨率为10米，覆盖2019年至2023年。洪水制图采用归一化植被指数（NDVI）结合特定区域阈值法进行。NDVI值低于0表示开阔水域，而值在0到0.1之间通常表明是泥浆、裸土。应用NDVI <0.019的阈值来识别丘陵地区受洪水影响的区域，以捕捉泥石流型洪水，而在喜马拉雅地区则使用NDVI <0，因为雪和水的存在因其与其他特征的光谱相似性而使分类变得复杂。利用哥白尼全球土地覆盖数据集对积雪覆盖区域进行掩膜，以提高高海拔地区的精度。数据处理在谷歌地球引擎（GEE）平台上进行。在使用场景分类层（SCL）进行云掩膜后生成季风季节图像合成图，并使用季风后图像填补时间上的云间隙，以确保时间数据的连续性。生成的洪水范围地图揭示了一致的空间模式，并为洪水预报、风险敏感型土地利用规划和跨学科研究提供了关键数据。尽管存在云干扰和复杂地形等挑战，但该数据集为洞察尼泊尔山区的洪水动态提供了有价值的见解。
