全球12公里沙尘数据集

全球12公里沙尘数据（TJ-Dust）是中科天机基于全球‑区域一体化数值天气模式，将中科院大气物理研究所 iDust 沙尘方案与自主研发的 SD3 动力内核深度耦合生成的高分辨率沙尘数据集。其综合考虑风速、地形地貌、土壤湿度、降水等关键因子，完整刻画沙尘排放、输送、干湿沉降等全物理过程，并依托 SD3 精准的输送模拟与真实的非静力动力过程有效降低模拟不确定性；同时通过沙尘物理过程与模式动力框架的深度耦合协同计算，避免了传统独立运算带来的系统开销与数据交互损耗，使沙尘模块可直接复用 SD3 动力内核的计算优化成果，在国产异构算力平台上实现高效并行加速与整体性能提升；此外，相较于依赖外部风场驱动的离线模式或依托物理参数化接口的传统模型，天机模式在沙尘动力更新方面具备显著更高的时间精度，降低了对边界层混合过程的依赖，生成更贴合实际的沙尘排放与沉降数据。
全球12公里沙尘数据（TJ-Dust）可提供地表沙尘浓度、沙尘光学厚度、累计沉降量、地表扬沙通量四个核心沙尘要素，支持长达30天的沙尘过程连续模拟与精准输出，能够完整还原沙尘起沙、传输、扩散、沉降的全生命周期演变特征，为全球沙尘过程模拟、环

The Global 12-km Dust Dataset (TJ-Dust) is a high-resolution dust dataset developed by Zhongke Tianji, which is deeply coupled with the iDust dust emission scheme from the Institute of Atmospheric Physics, Chinese Academy of Sciences and the independently developed SD3 dynamic core based on the global-regional integrated numerical weather prediction model. It comprehensively considers key factors such as wind speed, topography, soil moisture and precipitation, and fully depicts the entire physical processes including dust emission, transportation, dry and wet deposition. Leveraging SD3's precise transportation simulation and realistic non-hydrostatic dynamic processes, it effectively reduces simulation uncertainties. Meanwhile, through the deep coupling and collaborative computation of dust physical processes and the model's dynamic framework, it avoids the system overhead and data interaction losses caused by traditional separate computations, enabling the dust module to directly reuse the computational optimization achievements of the SD3 dynamic core and achieve efficient parallel acceleration and overall performance improvement on domestic heterogeneous computing platforms. Compared to traditional offline models driven by external wind fields or models relying on physical parameterization interfaces, the Tianji model exhibits significantly higher temporal accuracy in dust dynamic updates, reducing reliance on boundary layer mixing processes and generating dust emission and deposition data that better align with real-world observations.

The Global 12-km Dust Dataset (TJ-Dust) offers four core dust variables: surface dust concentration, dust optical depth, cumulative deposition, and surface dust emission flux. It supports continuous simulation and accurate output of dust events for up to 30 days, and can fully reproduce the full lifecycle evolution characteristics of dust emission, transportation, diffusion and deposition, providing support for global dust process simulation, environmental

搜集汇总

背景与挑战

背景概述

全球12公里沙尘数据集（TJ-Dust）由中科天机气象科技有限公司基于全球-区域一体化数值天气模式生成，通过深度耦合iDust沙尘方案与SD3动力内核，综合考虑风速、地形地貌等多因子完整模拟沙尘全物理过程。该数据集提供地表沙尘浓度、沙尘光学厚度等四个核心要素，支持长达30天的连续模拟，适用于全球沙尘过程模拟与环境评估等应用场景。