DustSCAN：基于SEVIRI的2022年沙尘暴逐小时数据集

空气中矿物粉尘对空气质量、人类健康和全球气候具有重要影响。由于地面监测站点（尤其在沙尘源区）分布稀疏，沙尘监测主要依赖极轨卫星光学仪器的气溶胶光学厚度反演遥感技术。这类数据虽具价值，但时间分辨率不足，难以实现精确的沙尘羽流追踪和源区特征分析。本研究推出的DustSCAN数据集，基于静止轨道气象卫星SEVIRI影像，构建了五年期逐小时沙尘羽流数据库。通过多通道红外影像技术，我们成功探测大气沙尘并实现逐小时沙尘影响像元追踪，继而运用基于密度的噪声空间聚类算法将受影响像元聚类为独立沙尘羽流。Mineral dust in the atmosphere exerts significant impacts on air quality, human health, and global climate. Due to the sparse distribution of ground-based monitoring stations, especially in dust source regions, dust monitoring primarily relies on remote sensing technology for aerosol optical depth (AOD) inversion using optical instruments aboard polar-orbiting satellites. While such data hold considerable value, their insufficient temporal resolution impedes accurate dust plume tracking and source region characteristic analysis. The DustSCAN dataset introduced in this study is built upon imagery from the geostationary meteorological satellite SEVIRI, establishing a five-year hourly dust plume database. Using multi-channel infrared imaging technology, we successfully detected atmospheric dust and achieved hourly tracking of dust-affected pixels. Subsequently, we clustered these affected pixels into individual dust plumes via the Density-Based Spatial Clustering of Applications with Noise (DBSCAN) algorithm.

搜集汇总

背景与挑战

背景概述

DustSCAN数据集是基于SEVIRI静止气象卫星的2022年沙尘暴逐小时监测数据，覆盖撒哈拉沙漠至中亚的主要沙尘源区，空间分辨率为0.25度。它利用多通道红外成像和DBSCAN聚类算法，实现了对沙尘羽流的高时间分辨率（逐小时）跟踪和识别，包含9950个独立沙尘羽流观测，旨在弥补地面监测站稀疏和极轨卫星数据时间分辨率不足的局限。