OEMC Hackathon 2023: Global FAPAR Modeling Dataset

Dataset organized by the Open-Earth-Monitor (OEMC) project within the context of Hackathon 2023. The dataset contains monthly mean FAPAR values aggregated by each ground station. FAPAR represents the fraction of the incoming (photosynthetic active) radiation that is absorbed by vegetation, and is given in the range 0-1. It is a measure of vegetation health and ecosystem functioning, and a key parameter in light use efficiency models that model primary productivity. For each monthly FAPAR value, a set of covariates / features were extracted from 32 raster spatial layers, including including satellite (spectral bands and indices) and temperature images (land surface temperature), climate images (precipitation) and digital terrain model (slope and elevation). The features are organized by columns, unique data points in time are identified by the sample_id column, and data points points belonging to the same location are identified by station_number. Column names: sample_id: unique identifier of datapoint station: ground station number fapar: monthly mean FAPAR month: month of measurement modis_{..}: NDVI, EVI, reflectance bands 1 (red), 2 (near-infrared), 3 (blue), and 7 (mid-infrared) based on MOD13Q1 modis_lst_day_p{..}: Land surface temperatures daytime of percentiles 5th, 50th and 95th based on MOD11A2 modis_lst_night_p{..}: Land surface temperatures nighttime of percentiles 5th, 50th and 95th based on MOD11A2 wv_yearly_p{..}: Water vapour aggregated yearly by percentiles 25th, 50th and 75th based on derived from MCD19A2 wv_monthly_lt_p{..}: Water vapour aggregated long-term monthly by percentiles 25th, 50th and 75th based on MCD19A2 wv_monthly_lt_sd: Water vapour aggregated long-term monthly standard deviation based on MCD19A2 wv_monthly_ts_raw: Water vapour monthly time series based on MCD19A2 wv_monthly_ts_smooth: Water vapour monthly time series smoothed using the Whittaker method based on MCD19A2 accum_pr_monthly: Monthly accumulated precipitation based on CHELSA timeseries dtm_{..}: Several DTM derivatives (Elevation, Slope, aspect (sine, cosine), curvature (up- and downslope), openness (negative, positive), compound topographic index (cti), valley bottom flatness (vbf)) based on MERIT DEM Files train.csv: Training set with 3,461 rows and 36 columns, including sample id (sample_id - index column), ground station (station), reference month (month), measured FAPAR (fapar), and 32 features / covariates test.csv: Test set with 4,939 rows and 34 columns, including sample id (sample_id - index column), ground station (station), reference month (month) and 32 features / covariates sample_submission.csv: a sample submission file with 4,939 rows and 2 columns, including sample id (sample_id - index column) and measured FAPAR (fapar)