A meta-analysis of mangrove soil CO2, CH4, and N2O fluxes: literature compilation and unpublished data

Please refer to the README_Workflow_full.html file for instructions on which .rmd scripts go with each section of the manuscript and subsequent analyses. Each .rmd file has detailed instructions on the required inputs. The README_Uncertainty_Propagation.html should be viewed if evaluating the uncertainty propagation workflow.</p><p dir="ltr">Of main interest here is:</p><ol><li>Global mangrove soil CO2 layer: Mangrove_soil_global_co2_rf_Final.tiff</li><li>Global mangrove soil N2O flux layer: Mangrove_soil_global_n2o_svmR_Final.tiff</li><li>Full mangrove soil GHG flux dataset: MetaFluxes_CH4_CO2_N2O_Mangroves_FINAL_DailyFlux.csv</li><li>Site-averaged, outliers removed soil GHG flux dataset: MetaFluxes_CH4_CO2_N2O_Mangroves_FINAL_DailyFlux_siteMean_noOut.csv</li></ol><p dir="ltr">Dataset files:</p><ol><li>MetaFluxes_CH4_CO2_N2O_Mangroves_FINAL_DailyFlux.csv: this dataset contains the full database of measurements.

All fluxes were scaled to the daily scale (mmol m-2 d-1 for CO2, CH4, and umol m-2 d-1 for N2O).</li><li>MetaFluxes_CH4_CO2_N2O_Mangroves_FINAL_DailyFlux_siteMean_noOut.csv: this dataset contains the site-averaged, outlier removed dataset for use as input to the "Mangrove_GHG_RasterExtract.rmd" script.</li><li>MetaFluxes_CH4_CO2_N2O_Mangroves_FINAL_DailyFlux_siteMean_noOut_FINAL.csv: this dataset contains the site-averaged, outlier removed dataset with additional columns added with extracted data from each geospatial predictor layer.

It is the output of the "Mangrove_GHG_RasterExtract.rmd" script and serves as an input for the "Mangrove_GHG_MachineLearning.rmd", "Mangrove_GHG_Shap_Ordination.rmd", and "Mangrove_GHG_Uncertainty_Propagation.rmd" scripts.</li></ol><p dir="ltr">Global soil flux geospatial layers:</p><ol><li>Mangrove_soil_global_co2_rf_Final.tif: global mangrove soil CO2 flux geospatial layer</li><li>Mangrove_soil_global_n2o_svmR_Final.tif: global mangrove soil N2O flux geospatial layer</li><li>soilFluxLayers folder: contains the final global geospatial data layers for mangrove soil CO2 and N2O fluxes.</li></ol><p dir="ltr">Geospatial predictors used for scaling machine learning models of soil flux:</p><ol><li>predictors_500m Folder: contains all the resampled 500 m spatial resolution geospatial predictor layers used to scale soil GHG flux models across the global mangrove coastline.

They are required inputs for the "Mangrove_GHG_GlobalScaling.rmd" script.</li></ol><p dir="ltr">Miscellaneous inputs:</p><ol><li>Mangrove_global_soil_upscale_Summary_Observed.csv: upscaled estimates of soil CO2, CH4, and N2O fluxes from observed data (database) at the global-scale, and for each geomorphic setting. It is an output of the "Mangrove_GHG_GlobalGridExtract.rmd" script.</li><li>final_hyperparameters.csv: hyperparameters for trained random forest and radial SVM machine learning models for CO2 and N2O, respectively.

Final is created in the "Mangrove_GHG_MachineLearning.rmd" script and used as input to the "Mangrove_GHG_UncertaintyPropagation.rmd" script.</li><li>all_ghg_models_list.rds: This is an R data object that contains all machine-learning models trained to predict soil CO2, CH4, and N2O fluxes, including models for the sensitivity analysis conducted on SOC and AGB data. It is a required input for the "Mangrove_GHG_Shap_Ordination.rmd" and "Mangrove_GHG_GlobalScaling.rmd" scripts.</li><li>MangroveArea_Pixel_Crop.rds: required shapefiles to chunk-process global grid for pixel-level predictor and mangrove area extracts.

A required input to "Mangrove_GHG_GlobalGridExtract.rmd" script.</li><li>Typology_v2_World_Join_Region.rds: required shapefile to extract geomorphic and sedimentary settings, and country and continent for each pixel in the global flux map grid. A required input to "Mangrove_GHG_GlobalGridExtract.rmd" script.</li><li>global_prediction_grid_500m_full.rds: R data object containing the data.frame or data.table with rows corresponding to pixels in the global soil flux grids and columns containing extracted geospatial predictor data for each pixel.

Created in the "Mangrove_GHG_GlobalGridExtract.rmd".</li><li>global_prediction_grid_500m.rds: R data object containing the data.frame or data.table with rows corresponding to pixels in the global soil flux grids and columns containing extracted geospatial predictor data for each pixel. Rows with 0 km2 mangrove area have been removed. This is the the input to the "Mangrove_GHG_UncertaintyPropagation.rmd" script.</li></ol><p><br></p><p><br></p>