Workflow to model GEDI PAI with Google Earth Engine and R
Scritps to reproduce the analysis can be found in the scripts directory. All data needed for the analysis are freely available and downloading those is part of the scripts.
- 000_R_presettings.R ** presettings that are useful in several R scripts and can be "source"ed
- 010_R_gedi_granules.R ** creates a "list" of GEDI orbits: orbit filename ** Output list has to be manually copied and imported in Google Earth Engine (GEE) as input for 020_GEE_gedi_looper.js
- 020_GEE_gedi_looper.js ** script to be run in Google Earth Engine ** loads the specified orbit FeatureCollections and extracts Sentinel 1 and 2 information (currently at the point level) ** it produces several export task with roughly 150 orbits to keep export size at a manageable size ** manually adjust path of granule_list.txt input ** manually copy code into new GEE Script ** ...and back if you change anything. Version control with R and GEE in one github feels too cumbersome to make it efficient. ** run in GEE and export all different output parts manually via Google Drive to current workstation into data/GEE/
- 030_R_join_geee_tables.R ** Import data table (csv) created in GEE, homogenize, put into one table and reproject
- 040_R_corine_and_filter.R ** extracting corine for gedi footprints ** filtering data
- 050_R_prepare_data_for_models.R ** some restructining and cleaning of data before the actual model calculations
- 060_R_run_model.R ** actual model comoutation on server
- 060_slurm_runner.slurm ** scheduling script for running 060_R_run_model.R on server
- 070_R_validation.Rmd ** Markdown script with evaluation and analysis of results. ** Plotting of results.
- 080_GEE_monthly_composite.R ** run in Google Earth engine to create monthly composites
- 090_R_spatial_prediction.R ** spatial prediction on monthly composites
- 095_R_analyze_spatial_predictions.R ** analysing temporal variablility of spatial predictions
- 100_R_plot_spatial_predictions.R ** plotting spatial predictions as maps