First National-level Tree Canopy Cover: Integrating AlphaEarth Embeddings with Landsat for Forest Monitoring in a Tropical Region: Dataset for 2017, 2020, and 2024

The files contain 30m tree canopy cover for Nigeria in 2017, 2020, and 2024. The dataset is the first of its kind in Nigeria, offering various stakeholders the opportunity to assess and monitor tree canopy cover across different states and regions of the country. The datasets were developed using a combination of Landsat 8 bands and indices, with the newly released AlphaEarth Embeddings as predictor variables. We modeled over 3,000 tree canopy cover reference points using random forests on the cloud computing platform Google Earth Engine. The 2024 model had a root-mean-square error of 15% on the validation data, and the waterways were masked. Based on the TCC, we used a national forest definition (areas with canopy cover ≥15% and a minimum area of 0.54ha [six 30 × 30m pixels]) to develop a binary forest/non-forest map for the three years. We also included the classified forest and non-forest layers for 2017, 2020, and 2024. With access to high-resolution imagery, we intend to improve the dataset, and a newer version will be published.     File structure: For TCC, we named each file with "Year"_TCC_masked. For forest and non-forest layers, we named each file with "Year"_FNF_masked.