中科天机全球12公里沙尘数据-样例

# 中科天机全球12公里沙尘数据

## 1. 数据集简介
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、环境评估及相关科学研究提供高质量、高可靠的数据支撑。

如需获取实时更新的模拟数据集，欢迎访问 [天机官网](https://www.tjweather.com/info/Download)。

## 2. 数据格式与规范
* **文件格式**：NetCDF (.nc)
* **空间范围**：全球 (网格大小: 3600 × 1800,
经度范围：grid_xt : 0.05 to 359.95°E，
纬度范围：grid_yt : -89.95 to 89.95 °N)
* **空间分辨率**：0.1° (~12km)
* **时间分辨率**：逐小时 (Hourly)
* **路径规则**：/tjweather/tjweather_dust_sample/TJdust/{变量名}/{起报时间}_变量名_f{起报后的小时数}.nc
*示例：2026年3月3日00z起报的预报第1小时的地表沙尘浓度数据路径：/tjweather/tjweather_dust_sample/TJdust/dust_conc/2026030300_dust_conc_f001.nc*

## 3. 数据要素清单

| 变量名 | 物理含义 | 单位 |
| :--- | :--- | :--- |
| **dust_conc** | 地表沙尘浓度 | $kg/kg$ |
| **dust_ddep** | 累计沉降 | $kg/m^2/s$ |
| **dod** | 沙尘光学厚度 | $-$ |
| **dust_emis** | 地表扬沙通量 | $kg/m^2/s$ |

## 4. 下载方法及使用方法
数据集文件元信息以及数据文件，请浏览“数据集文件”页面获取。

下载方法请参照 [官方说明](https://www.modelscope.cn/docs/datasets/download)

netCDF4数据文件使用方法请参考 [天机netCDF4数据文件格式介绍](https://www.tjweather.com/info/doc/datause/netcdf4-readme.html)

# Global 12 km Dust Dataset (CAS Tianji)

## 1. Dataset Introduction
The Global 12 km Dust Dataset (TJ-Dust) is a high-resolution dust dataset developed by Zhongke Tianji, which is built by deeply coupling the iDust dust scheme from the Institute of Atmospheric Physics, Chinese Academy of Sciences (IAP CAS) with the self-developed SD3 dynamic core based on the global-regional integrated numerical weather model. It comprehensively considers key factors such as wind speed, topography, soil moisture and precipitation, and fully characterizes the entire physical processes of dust emission, transportation, dry and wet deposition. Relying on the accurate transportation simulation of SD3 and realistic non-hydrostatic dynamic processes, it effectively reduces simulation uncertainties. Meanwhile, the deep coupling and collaborative calculation between the dust physical processes and the model dynamic framework avoid the system overhead and data interaction loss caused by traditional independent operations, allowing the dust module to directly reuse the computational optimization achievements of the SD3 dynamic core, achieving efficient parallel acceleration and overall performance improvement on domestic heterogeneous computing platforms. Furthermore, compared with offline models driven by external wind fields or traditional models relying on physical parameterization interfaces, the Tianji model has significantly higher temporal accuracy in dust dynamic updates, reduces dependence on boundary layer mixing processes, and generates dust emission and deposition data that are more consistent with actual observations.

The Global 12 km Dust Dataset (TJ-Dust) provides 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 processes for up to 30 days, and can fully restore the entire life cycle evolution characteristics of dust initiation, transportation, diffusion and deposition, providing high-quality and highly reliable data support for global dust process simulation, environmental assessment and related scientific research.

For access to real-time updated simulation datasets, please visit the [official Tianji website](https://www.tjweather.com/info/Download).

## 2. Data Format and Specifications
* **File Format**: NetCDF (.nc)
* **Spatial Extent**: Global (Grid Size: 3600 × 1800, Longitude Range: grid_xt: 0.05°E to 359.95°E, Latitude Range: grid_yt: -89.95°N to 89.95°N)
* **Spatial Resolution**: 0.1° (~12 km)
* **Temporal Resolution**: Hourly
* **File Path Rule**: /tjweather/tjweather_dust_sample/TJdust/{variable_name}/{initialization_time}_{variable_name}_f{forecast_hour}.nc
*Example*: The path of surface dust concentration data for the 1st forecast hour initialized at 00Z on March 3, 2026: /tjweather/tjweather_dust_sample/TJdust/dust_conc/2026030300_dust_conc_f001.nc

## 3. Data Variable List

| Variable Name | Physical Meaning | Unit |
| :--- | :--- | :--- |
| **dust_conc** | Surface Dust Concentration | $kg/kg$ |
| **dust_ddep** | Cumulative Deposition | $kg/m^2/s$ |
| **dod** | Dust Optical Depth | $-$ |
| **dust_emis** | Surface Dust Emission Flux | $kg/m^2/s$ |

## 4. Download and Usage Instructions
Metadata and data files of the dataset can be obtained via the "Dataset Files" page.
For download methods, please refer to the [official instructions](https://www.modelscope.cn/docs/datasets/download).
For usage instructions of netCDF4 data files, please refer to the [Tianji netCDF4 Data Format Introduction](https://www.tjweather.com/info/doc/datause/netcdf4-readme.html).

搜集汇总

背景与挑战

背景概述

该数据集是一个全球沙尘样例数据，由CAS-TJ开发，通过耦合iDust沙尘方案与SD3动力核心，模拟沙尘排放、传输和沉积的全过程。它提供表面沙尘浓度、沙尘光学深度等四个核心变量，数据格式为NetCDF，空间分辨率为12公里，时间分辨率为每小时。