Validation data and analysis scripts — Tree-cover loss and associated uncertainty estimates in Nigeria (2001–2020)
Published by MUBARAK MAHMUD
About this dataset
This dataset accompanies the study "Quantifying Tree-Cover Loss and Associated Uncertainty in Satellite-Derived Estimates in Nigeria" (Mahmud et al., 2026). It includes the reference data used to assess the accuracy of the Hansen Global Forest Change product (v1.11) across Nigeria between 2001 and 2020, along with the analysis scripts used to reproduce the bias-corrected estimates. The dataset includes: * 783 validation points selected via a stratified random sampling design (ecological zone × mapped loss status); * reference labels derived from visual interpretation of high-resolution satellite imagery (Google Earth, Maxar, Airbus); * Python scripts (Jupyter Notebook) used to compute the confusion matrix, accuracy metrics (overall, user's, and producer's accuracy), and the bias-adjusted area estimate with 95% confidence intervals. These data support a bias-adjusted tree-cover loss estimate of 3.37 Mha (95% CI: 2.39–4.36 Mha) for Nigeria over 2001–2020, compared to 2.84 Mha from the raw mapped product.