Dataset and code for "Quantifying River Persistence and Identifying Flow Cessation in Semi-arid Rivers Using Wetted River Length and Water Presence Duration"

This repository contains the dataset and code supporting the manuscript: "Quantifying River Persistence and Identifying Flow Cessation in Semi-arid Rivers Using Wetted River Length and Water Presence Duration". Repository Contents: Result_Rasters.zip: Contains the derived spatial datasets, including monthly surface-water extent masks, Water Presence Duration (WPD) continuous frequency rasters, and reclassified river connectivity status for the Xiliao and Tao'er Rivers during the 2021 non-frozen season (GeoTIFF format). Validation_Samples.zip: Contains the reference sample points (Shapefile format) used to train and validate the seasonally stratified random forest models. Source_Code.zip: Contains the complete Google Earth Engine (GEE) JavaScript codes for multidimensional feature construction, model training, and 10-day composite surface-water extraction. README.txt: Detailed data dictionary and instructions for reproducibility. Abstract of the associated manuscript: River flow cessation is an increasingly important hydroecological issue amid climate change and intensive human activities, particularly in semi-arid regions, where it threatens water security and ecosystem stability. Yet basin-scale characterization of flow cessation remains challenging because river persistence varies longitudinally and seasonally. Here, we developed a multidimensional feature-based, seasonally stratified remote-sensing framework to quantify river persistence and identify flow-cessation patterns in semi-arid river corridors. We further introduced two hydrologically interpretable indicators: wetted river length (WRL) for longitudinal connectivity and water presence duration (WPD) for temporal persistence. We applied the framework to the Xiliao and Tao'er Rivers in the semi-arid Songliao Basin, northeastern China, during the 2021 non-frozen season. Surface-water mapping achieved overall accuracies of 0.91 and 0.93, respectively. The Xiliao River showed severe seasonal contraction and frequent drying, with a non-frozen period WRL of 381.56 km; 35.19% of its main channel (153.03 km) had WPD < 25%, forming a pronounced connectivity barrier in the middle reaches. In contrast, the Tao'er River maintained stronger persistence, with a WRL of 560.59 km and only 4.61% of its main channel (25.84 km) showing WPD < 25%. These contrasting patterns suggest that semi-arid river flow cessation may occur as clustered disconnected reaches that fragment longitudinal connectivity, particularly where intensive irrigation and groundwater depletion promote channel water loss. Despite uncertainties in very narrow channels and spectrally confused wetted surfaces, the framework provides a transferable basis for quantifying river persistence and identifying cessation hotspots.本仓库包含支撑以下学术论文的数据集与代码：《利用河道过水长度（Wetted River Length, WRL）与水体存在时长（Water Presence Duration, WPD）量化半干旱河流的径流持续性并识别断流现象》。 仓库内容： Result_Rasters.zip：包含衍生空间数据集，涵盖2021年非冰封期西辽河与洮儿河的逐月地表水范围掩膜、水体存在时长（WPD）连续频率栅格，以及重分类后的河道连通状态数据，格式均为GeoTIFF。 Validation_Samples.zip：包含用于训练与验证季节分层随机森林模型的参考采样点，格式为Shapefile。 Source_Code.zip：包含用于多维特征构建、模型训练以及10日合成地表水提取的完整Google Earth Engine（GEE）JavaScript代码。 README.txt：包含详细的数据字典与可复现性操作指南。 相关学术论文摘要： 河流断流是气候变化与人类活动加剧背景下愈发严峻的水文生态问题，在半干旱地区尤为突出，严重威胁当地水安全与生态系统稳定性。然而，由于河流径流持续性随空间纵向与季节动态变化，流域尺度的断流特征刻画仍极具挑战。本研究构建了基于多维特征的季节分层遥感框架，用于量化半干旱河流廊道的径流持续性并识别断流模式；同时提出两个可水文解释的指标：用于表征纵向连通性的河道过水长度（WRL），以及用于表征时间持续性的水体存在时长（WPD）。我们将该框架应用于中国东北松辽盆地半干旱区域的西辽河与洮儿河2021年非冰封期。地表水提取的总体精度分别达到0.91与0.93。西辽河表现出显著的季节性河道收缩与频繁断流，非冰封期河道过水长度为381.56 km；其主河道中有35.19%（153.03 km）的水体存在时长低于25%，在中游形成了明显的连通性障碍。与之相对，洮儿河的径流持续性更强，河道过水长度达560.59 km，仅4.61%的主河道（25.84 km）水体存在时长低于25%。上述对比结果表明，半干旱河流的断流可能以簇状分散河段的形式出现，割裂纵向连通性，在高强度灌溉与地下水超采导致河道水体流失的区域尤为显著。尽管在极窄河道与光谱混淆的过水区域仍存在不确定性，本框架仍为量化河流径流持续性、识别断流热点区域提供了可迁移的研究基础。