2009-2020年基于GOSAT卫星的全球中低纬二氧化碳柱浓度数据集  

A dataset of global CO2 concentration in global middle and low latitudes based on GOSAT satellite data during 2009–2020

作　　者：许静静 龚威 张劲 张豪伟 马昕[3] 韩舸 XU Jingjing;GONG Wei;ZHANG Jin;ZHANG Haowei;MA Xin;HAN Ge(School of Electronic Information,Wuhan University,Wuhan 473072,P.R.China;Wuhan Geomatics Institute,Wuhan 430079,P.R.China;State Key Laboratory of Information Engineering in Surveying,Mapping,and Remote Sensing,Wuhan University,Wuhan 430079,P.R.China;School of Remote Sensing and Information Engineering,Wuhan University,Wuhan 430079,P.R.China)

机构地区：[1]武汉大学,电子信息学院,武汉473072 [2]武汉测绘研究院,武汉430079 [3]武汉大学,测绘遥感信息工程国家重点实验室,武汉430079 [4]武汉大学,遥感信息工程学院,武汉430079

出　　处：《中国科学数据（中英文网络版）》2023年第3期462-471,共10页China Scientific Data

基　　金：国家自然科学基金(42171464、41971283、41801261、41827801、41801282);国家重点研发计划(2017YFC0212600);国家对地观测数据中心开放基金(NODAOP2021005);LIESMARS特别研究基金。

摘　　要：为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。本研究利用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°。本数据集由2009–2020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。In response to the increasingly intensifying greenhouse effect,Countries around the world jointly signed the Paris Agreement,and China also made plans and policies of peak carbon dioxide emissions,carbon neutral plans.Carbon dioxide is the focus of international concern as the most important greenhouse gas.So,it’s crucial to know how to obtain carbon dioxide concentration of high precision,high resolution temporal and spatial distribution for advancing the top-down assessment of carbon source,carbon sink and carbon neutral research.This paper creatively proposes a method to obtain high-precision and high-resolution temporal and spatial distribution map of global carbon dioxide concentration by using satellite data.We constructed a new prior time curve parameter library for fitting time domain information.In this paper,we used the transfer learning theory to integrate the time information as a prior profile into the spatial information based on the global data of GOSAT satellite.The spatial prediction information was adjusted to obtain more accurate spatio-temporal prediction of carbon dioxide concentration.The spatiotemporal resolution of the product database is 0.25°.Finally,the database has been compared with TCCON data in middle and low latitudes,which shows the correlation coefficient R and RMSE is 0.98 and 1.38 ppm of the monthly average carbon dioxide concentration respectively.The recommended database can be applied to the calculation of carbon sources and carbon sinks on a large scale.

关 键 词：二氧化碳柱浓度 GOSAT 迁移学习 全球 

分 类 号：X51[环境科学与工程—环境工程]