CDI_Dataset

This dataset provides a high-resolution (1 km) Composite Drought Index (CDI) for Tunisia covering the period 2000–2025, developed to support drought monitoring and forecasting in arid and semi-arid environments. The CDI was generated by integrating multi-source remote sensing and reanalysis products, including CHIRPS precipitation, ERA5-Land climate variables, and MODIS vegetation and temperature data. Five drought-related indicators were derived: the Standardized Precipitation Index (SPI), Standardized Precipitation Evapotranspiration Index (SPEI), Soil Moisture Anomaly (SMA), NDVI anomaly (NDVI-A), and Land Surface Temperature anomaly (LST-A). These indicators were combined through a logic-based cause–effect framework to capture drought interactions across meteorological, agricultural, and thermal conditions. The dataset reveals strong spatio-temporal drought variability and highlights major drought episodes such as 2002–2003, 2016–2018, and 2020–2024, particularly affecting central and southern Tunisia. Validation against in-situ rainfall observations shows strong agreement between station-based SPI and CHIRPS-derived estimates. Wavelet analysis indicates a significant relationship between CDI variability and the North Atlantic Oscillation (NAO), emphasizing the influence of large-scale atmospheric circulation on drought dynamics. In addition, the dataset was used to benchmark twelve deep learning forecasting models, where attention-based architectures such as TimeFormer achieved the best performance. This CDI dataset constitutes a valuable resource for operational drought early warning, climate resilience studies, and sustainable water resource management in Tunisia.本数据集面向突尼斯地区，提供了2000—2025年时段的1公里分辨率综合干旱指数（Composite Drought Index, CDI），旨在支撑干旱半干旱环境下的干旱监测与预报工作。该综合干旱指数通过融合多源遥感与再分析产品生成，涵盖CHIRPS降水数据、ERA5-Land气候变量数据以及MODIS植被与温度数据。研究共衍生了5项干旱相关指标：标准化降水指数（Standardized Precipitation Index, SPI）、标准化降水蒸散指数（Standardized Precipitation Evapotranspiration Index, SPEI）、土壤湿度距平（Soil Moisture Anomaly, SMA）、归一化植被指数距平（NDVI anomaly, NDVI-A）以及地表温度距平（Land Surface Temperature anomaly, LST-A）。上述指标通过基于逻辑的因果框架进行融合，以捕捉气象、农业与热力条件下的干旱交互特征。本数据集揭示了显著的干旱时空变异特征，并凸显了2002—2003年、2016—2018年以及2020—2024年等主要干旱事件，这些事件对突尼斯中部与南部地区影响尤为严重。通过原位降雨观测数据进行验证的结果表明，台站观测标准化降水指数与CHIRPS反演估算值之间具有高度一致性。小波分析结果显示，CDI变异与北大西洋涛动（North Atlantic Oscillation, NAO）之间存在显著关联，凸显了大尺度大气环流对干旱动态过程的影响。此外，本数据集被用于12款深度学习预报模型的基准测试，其中基于注意力机制的架构（如TimeFormer）取得了最优性能。本综合干旱指数数据集是突尼斯地区开展业务化干旱预警、气候韧性研究以及可持续水资源管理的宝贵资源。