FAPAR monthly time-series (250 m): Long-term trend (2000-2021)

• Main Authors: Julia Hackländer, Leandro Parente, Yu-Feng Ho, Davide Consoli, Tomislav Hengl
• Format: Dataset
• Language: English
• Published: 2023
• Subjects: FAPAR; global; OEMC; Antarc*; Antarctica
• Online Access: https://zenodo.org/record/8399173; https://doi.org/10.5281/zenodo.8399173

General Description The monthly aggregated Fraction of Absorbed Photosynthetically Active Radiation (FAPAR) dataset is derived from 250m 8d GLASS V6 FAPAR. The data set is derived from Moderate Resolution Imaging Spectroradiometer (MODIS) reflectance and LAI data using several other FAPAR products (MODIS Collection 6, GLASS FAPAR V5, and PROBA-V1 FAPAR) to generate a bidirectional long-short-term memory (Bi-LSTM) model to estimate FAPAR. The dataset time spans from March 2000 to December 2021 and provides data that covers the entire globe. The dataset can be used in many applications like land degradation modeling, land productivity mapping, and land potential mapping. The dataset includes: Long-term: Derived from monthly time-series. This dataset provides linear trend model for the p95 variable: (1) slope beta mean (p95.beta_m), p-value for beta (p95.beta_pv), intercept alpha mean (p95.alpha_m), p-value for alpha (p95.alpha_pv), and coefficient of determination R2 (p95.r2_m). Monthly time-series: Monthly aggregation with three standard statistics: (1) 5th percentile (p05), median (p50), and 95th percentile (p95). For each month, we aggregate images inside the months and one image before and after, about 5 to 6 images for a single month depending on the number of images inside the month. Data Details Time period: March 2000 – December 2021 Type of data: Fraction of Absorbed Photosynthetically Active Radiation (FAPAR) How the data was collected or derived: Derived from 250m 8 d GLASS V6 FAPAR using Python running in a local HPC. Cloudy pixels were removed and only positive values of water vapor were considered to compute the statistics. The time-series gap-filling and time-series analysis were computed using the Scikit-map Python package. Statistical methods used: for the long-term, trend analysis of p95 monthly variable; for the monthly time-series, percentiles 05, 50, and 95. Limitations or exclusions in the data: The dataset does not include data for Antarctica. Coordinate reference system: EPSG:4326 Bounding ...