A ranking–correction–fusion framework (HARMONY) for climate-zone-aware multi-source daily precipitation over China at 0.1°

Precipitation data is critical for hydrological modelling and for assessing floods and droughts. However, China's complex topography and pronounced climatic variations mean that data from a single source often exhibits significant regional and seasonal biases, making it challenging to produce stable, reliable nationwide products. This study presents a multi-source fusion framework-HARMONY for generating nationwide daily precipitation data. The core methodology involves objective selection, bias correction, regional adaptive fusion and hierarchical evaluation and interpretation. The structure initially picks out additional input datasets from possible products within the limits of various indicators for diagnosis. It then applies consistency bias control to reduce systematic error propagation, performing fusion at climate zoning scales to generate seamless national results. Multi-level evaluations incorporate continuous statistics, event detection and extreme indicators.降水数据对于水文模拟以及洪涝、干旱灾害评估至关重要。然而，中国复杂的地形与显著的气候差异，导致单一来源的降水数据往往存在显著的区域与季节偏差，难以生成稳定可靠的全国范围产品。本研究提出了一种用于生成全国逐日降水数据的多源融合框架——HARMONY。其核心方法包括客观遴选、偏差校正、区域自适应融合以及分层评估与解译。该框架首先基于各类诊断指标，从潜在候选产品中筛选出额外的输入数据集；随后通过一致性偏差控制以降低系统误差传播，并以气候分区为尺度开展融合，最终生成无缝衔接的全国范围降水结果。多层次评估涵盖连续统计量、事件检测以及极端值指标。