2009-2020年基于GOSAT卫星的全球二氧化碳浓度数据集

应对日益加剧的温室效应问题，全球各国联合签署了《巴黎协定》，我国也制定了碳达峰、碳中和的计划和政策，二氧化碳作为最主要的温室气体是国际关注的重点。因此，获得高精度、高分辨率的二氧化碳柱浓度时空分布图对于推进“自上而下”评估碳源、碳汇、碳中和的研究至关重要。本文利用GOSAT卫星全球数据，通过迁移学习理论，将时间信息作为先验廓线融入空间信息，对空间预测信息进行调整，得到更高准确度的二氧化碳柱浓度时空预测结果。与中低纬的TCCON站点数据对比，本文算法最终得到的月均二氧化碳柱浓度图指标的综合结果R为0.98，RMSE为1.38 ppm，空间分辨率为0.25°。本文的数据产品由2009至2020年月均二氧化碳柱浓度文件组成，包含136个h5文件，可应用于长时间序列的碳源和碳汇计算。

To address the intensifying greenhouse effect, countries worldwide have jointly signed the Paris Agreement. China has also formulated plans and policies for carbon peak and carbon neutrality. As the most dominant greenhouse gas, carbon dioxide has been the focal point of international concern. Therefore, acquiring spatiotemporal distribution maps of carbon dioxide column concentration with high accuracy and high resolution is critically important for advancing "top-down" research on assessing carbon sources, carbon sinks, and carbon neutrality. This study utilizes global data from the GOSAT satellite, integrates temporal information as prior profiles into spatial information via transfer learning theory to adjust spatial prediction outputs, and thus obtains spatiotemporal prediction results of carbon dioxide column concentration with improved accuracy. Compared with TCCON site data in mid-low latitudes, the comprehensive performance metrics of the monthly mean carbon dioxide column concentration maps derived from the proposed algorithm exhibit a Pearson correlation coefficient R of 0.98, a root mean square error (RMSE) of 1.38 ppm, and a spatial resolution of 0.25°. The data product presented in this study comprises monthly mean carbon dioxide column concentration files from 2009 to 2020, including 136 h5 files, which can be applied to long-time series carbon source and sink calculations.

搜集汇总

背景与挑战

背景概述

该数据集是基于GOSAT卫星数据生成的2009-2020年全球二氧化碳柱浓度月平均产品，空间分辨率为0.25度，包含136个h5格式文件。它采用转移学习理论整合时空信息，实现了高精度（与TCCON站点数据相比，R为0.98，RMSE为1.38 ppm）的二氧化碳浓度时空分布图，适用于长期碳源碳汇计算和碳中和研究。