Geostatistical inverse modeling with large atmospheric data: data files for a case study from OCO-2

The files in this data repository provide the inputs required to run an inverse modeling case study. This case study will estimate CO2 fluxes across North America for July 2015 using synthetic observations that have been created to resemble observations from NASA's Orbiting Carbon Observatory 2 (OCO-2) satellite. This data repository is specifically linked to a GitHub code repository (http://doi.org/10.5281/zenodo.3241524 or https://github.com/greenhousegaslab/geostatistical_inverse_modeling). That GitHub repository provides scripts for constructing a geostatistical inverse model that will estimate greenhouse gas fluxes or air pollution emissions using atmospheric observations. The GitHub repository includes a case study that can be run out-of-the-box; the case study provides users an opportunity to test out and explore the inverse modeling code. All of the input data files for that case study are provided for download here. Here is a brief explanation of the different files included in this data repository, but refer to the linked GitHub repository for greater details. All of these files are in a ".mat" file that can be read into Matlab using the load function or can be read into R using the R.matlab package. H.tar.gz: This tar file contains the H matrices or sensitivity matrices required by the inverse model. These inputs were generated using the Stochastic Time-Inverted Lagrangian Transport (STILT) model as part of NOAA's CarbonTracker-Lagrange program (https://www.esrl.noaa.gov/gmd/ccgg/carbontracker-lagrange/). The H matrix is too large to store in a single file. We have therefore split up the matrix into 328 different files (all contained within H.tar.gz). Each file contains a vertical strip of the H matrix that corresponds to a different time period of fluxes to be estimated as part of the inverse model. Z.mat: This file contains the synthetic OCO-2 observations used in the case study. areas_us.mat: This file lists the area of each model grid box used in the case study in units of meters2. This file only includes grid box area for model grid boxes that fall within the continental United States. We estimate CO2 fluxes across terrestrial North America on a 1 degree latitude by 1 degree longitude grid as part of the case study. Each of these model grid boxes will have a different area, depending upon the latitude of that model grid box. distmat.mat: This file contains a matrix that lists the distance (in kilometers) between the center of each model grid box used in the case study. land_mask.mat: We only estimate CO2 fluxes for terrestrial regions of North America as part of the case study. This land mask is used to convert the fluxes estimated by the inverse model to a latitude-longitude grid that can then be plotted. H_all_OCO2.mat: This file contains the H matrices summed across differnt time periods. I.e., this file is the sum of all the individual H files contained within H.tar.gz. Xvar.tar.gz: This file contains different environmental variables from ERA5 meteorology that have been reformatted to match the H footprint matrices. These different variables can be used as predictors of CO2 fluxes in an inverse model. The different variables included in this file are as follows: Xvar_e.mat Evaporation Xvar_msdwswrf.mat Mean surface downward short-wave radiation flux Xvar_q.mat Specific humidity Xvar_stl1.mat Soil temperature level 1 Xvar_stl3.mat Soil temperature level 3 Xvar_swvl1.mat Volumetric soil water layer 1 Xvar_t2m.mat 2 metre temperature Xvar_tp.mat Total precipitation Xvar_mer.mat Mean evaporation rate Xvar_pev.mat Potential evaporation Xvar_r.mat Relative humidity Xvar_swvl3.mat Volumetric soil water layer 1 Xvar_tcc.mat Total cloud cover本数据集仓库内的文件为开展一项反演建模（inverse modeling）案例研究提供了所需输入数据。本案例将借助模拟生成的观测数据——旨在模拟美国国家航空航天局（National Aeronautics and Space Administration, NASA）轨道碳观测卫星2号（Orbiting Carbon Observatory 2, OCO-2）的实测结果——估算2015年7月北美全域的二氧化碳通量（CO2 fluxes）。 本数据集仓库与指定GitHub代码仓库（http://doi.org/10.5281/zenodo.3241524 或 https://github.com/greenhousegaslab/geostatistical_inverse_modeling）关联。该GitHub仓库提供了构建地质统计反演模型的脚本，可借助大气观测数据估算温室气体通量或大气污染物排放量；其中包含一个开箱即用的案例研究，供用户测试并探索反演建模代码。本仓库已提供该案例所需的全部输入数据文件供下载。 以下为本数据集仓库内各文件的简要说明，更多细节请参阅关联的GitHub仓库。所有文件均为".mat"格式文件，可通过MATLAB的load函数读取，或借助R语言的R.matlab包进行读取。 H.tar.gz：该压缩包包含反演模型所需的H矩阵（或称敏感性矩阵）。此类输入数据由美国国家海洋和大气管理局（National Oceanic and Atmospheric Administration, NOAA）CarbonTracker-Lagrange项目（https://www.esrl.noaa.gov/gmd/ccgg/carbontracker-lagrange/）中的随机时间反转拉格朗日输送（Stochastic Time-Inverted Lagrangian Transport, STILT）模型生成。由于H矩阵体积过大，无法存储于单个文件中，因此我们将其拆分为328个独立文件（均包含于H.tar.gz内），每个文件对应反演建模中待估算通量的一个不同时间段的H矩阵垂直切片。 Z.mat：本文件包含本案例研究中使用的模拟OCO-2观测数据。 areas_us.mat：本文件列出了本案例研究中使用的各模型网格单元的面积，单位为平方米。本文件仅包含美国本土范围内的模型网格单元面积。本案例中，我们将以1°纬度×1°经度的网格为基础，估算北美陆地全域的二氧化碳通量，各网格单元的面积会随其所在纬度发生变化。 distmat.mat：本文件包含一个矩阵，列出了本案例研究中各模型网格单元中心点之间的距离（单位：千米）。 land_mask.mat：本案例中我们仅估算北美陆地区域的二氧化碳通量，该陆地掩膜文件用于将反演模型得到的通量结果转换为可绘图的经纬度网格数据。 H_all_OCO2.mat：本文件包含所有不同时间段合并后的H矩阵，即H.tar.gz内所有单个H文件的总和。 Xvar.tar.gz：该压缩包包含经格式重构以匹配H足迹矩阵的ERA5气象环境变量，此类变量可作为反演模型中二氧化碳通量的预测因子。本压缩包包含的变量如下： Xvar_e.mat：蒸发量 Xvar_msdwswrf.mat：地表向下平均短波辐射通量 Xvar_q.mat：比湿 Xvar_stl1.mat：1层土壤温度 Xvar_stl3.mat：3层土壤温度 Xvar_swvl1.mat：1层土壤体积含水量 Xvar_t2m.mat：2米高度气温 Xvar_tp.mat：总降水量 Xvar_mer.mat：平均蒸发速率 Xvar_pev.mat：潜在蒸发量 Xvar_r.mat：相对湿度 Xvar_swvl3.mat：第一层土壤体积含水量 Xvar_tcc.mat：总云量