Data from paper: Large carbon sink potential of Secondary Forests in Brazilian Amazon to mitigate climate change (public)

Title: Large carbon sink potential of Secondary Forests in the Brazilian Amazon to mitigate climate change

Contact: Viola Heinrich (viola.heinrich@bristol.ac.uk)

This repository contains:

1. Zipped folder: Fig1_data_input.zip - all the files needed to produce Figure 1a-e of the main paper. Set the working directory to folder containing the file and use the script "Fig1a_f_plot.R" to run (see below). The folder contains the input files of the 6 driving variables used to build regrowth models seen in Figure 1 - these files are in the format "<driver>_assessment_v2.csv". The columns in the files are: A: age of secondary forest; B: 50th percentile (median) of the modal Aboveground Biomass (AGB) value for the given age (note, units are in biomass not carbon: Mg/ha/yr); C: The bias-corrected AGB value, calculated by subtracting the lowest AGB value in column B such that the AGB data starts at or near 0Mg/ha/yr at age 1. D: the number of secondary forest pixels observed to have the given age, E: "Threshold" : the threshold limits of the given driver e.g.  0 Fires in fire_assessmentv2.csv implies the corresponding secondary forest pixels experienced 0 fires throughout the analysis period.  The folder also contains the output regrowth models seen in Figure 1 in the format "regrowth_model_<driver_threshold>.RData" where driver_threshold refers to the driving variable name and the associated threshold limit for the given driver.
2. Zipped folder: Fig2_regions_outline.zip - contains the boundaries of the 4 regions identified in Figure 2a of the main paper in a shapefile (.shp) format and the corresponding file formats needed to produce and load a shapefile. 
3. Zipped folder: Fig1g_2b_e_variable_importance.zip - contains the output files of the random forest analysis assessing the variable importance for the whole Amazon ("whole_Amazon" subfolder) and for the different regions identified in Figure2a. Files are given as .RDS files that can be loaded in R and the corresponding figures produced using the script "Fig1g_2b_e_plot.R". Files start with the region of interest e.g. "whole_Amazon" or "NE_sector". Middle part of the filename - importance_conditionalTrue/False - this determines whether the importance was calculated using the conditional permutation (True) or not (False). The end of the file name - seed<NUM> - denotes the number of the random seed that was set to extract the sample data. e.g. whole_Amazon_2500_cforest_important_conditionalTrue_seed200.RDS - shows the conditional permutation importance assessment using a sample size of 2500 when the setseed parameter was set to 200 to extract a random sample representing the whole Amazon. The remaining files are the random forest output - as .RDS file. Please note the code to produce the random forest model and the importance assessment has not been included here - this code takes multiple days to run, so only the input and outputs have been included here. Please contact the corresponding author (see end) for more information on this. 
4. Zipped folder: Fig3_data_input.zip -  all the files needed to produce Figure 3a-d of the main paper. Set the working directory to folder containing the file and use the script "Fig3_plot.R" to run (see below). The folder contains the input files of the 6 driving variables used to build regrowth models seen in Figure 3 - these files are in the format "<REGION>-Group.csv". See bullet point 1 for explanations for the columns in the file. Again column E -"threshold" denotes the code used to identify the the 4 subclasses of regrowth seen in the Figure. Where 11 = No disturbance; 12 = Only burning; 21 = Only (multiple) deforestations; 22 = Both burning and multiple deforestations as disturbance. The code takes data in AGB and converts to AGC.  The folder also contains the output regrowth models seen in Figure 3 in the format "regrowth_model_<region_disturbance_type>.RData" where region_disturbance refers to the region and the type of disturbance experienced. 
5.  Zipped folder: Fig4_5_carbon_sink_2017.zip - Contains two subfolders: a) Map_aggre_0.1deg -this folder contains .tiff files (and associated files) of the losses, gains and net change in AGC between 2016 - 2017 in secondary forests in Amazonia - this has been aggregated to 0.1 degree grid cells so each cell contains the total sum of the losses/gains experienced by secondary forests in that 0.1degree grid cell. b) secondary_forest_by_region_and_disturbance - this folder contains .tiff files (and associated files) of the secondary forest data at the original resolution (30m) for 2016 and 2017 split up according to the regions identified in Figure 2, and the type of disturbance (if any). The associated files include a .dbf file which includes additional data [read "README.txt" file in folder] - upon loading the data in a GIS software - the age of the secondary forest pixel will be displayed - open the attribute table to see more data associated with that given pixel e.g. modelled associated AGB for a given pixel. Files in this folder can be used to make Figure 4d and Figure 5 - see script "Fig4_Fig5_plot.R" in the code repository (see below). 

Code: The corresponding code mentioned here can be access here: heinrichTrees/secondary-forest-regrowth-amazon-public (github.com)

Data usage: When using any code or data in this repository or another related to this study please cite Heinrich et al.2021 and the original paper as well as the DOI of this repository. 

If you need anything else, please contact the corresponding author: Viola Heinrich (viola.heinrich@bristol.ac.uk)