PKU GIMMS NDVI: Global Vegetation Index Data from 1982 to 2020

Global biweekly Normalized Difference Vegetation Index (NDVI) data from 1982 to 2020, generated by researchers at Peking University. The dataset uses a machine learning model trained on 4.9 million Landsat NDVI samples and consolidates AVHRR and MODIS data. It is designed to provide a spatiotemporally consistent record for studying vegetation dynamics and climate feedbacks.

Use Cases

• Assessing long-term vegetation dynamics and trends based on the 39-year global NDVI record.
• Modeling climate change impacts and feedbacks on ecosystems using the spatiotemporally consistent data.
• Validating other remote sensing products by comparing with the high-accuracy PKU GIMMS NDVI product.
• Studying seasonal and interannual vegetation changes in tropical areas where orbital drift effects have been mitigated.

Strengths

• High accuracy compared to its predecessor, with an R² of 0.975, mean absolute error of 0.033, and mean absolute percentage error of 9%.
• Based on a large training set of 4.9 million globally distributed Landsat NDVI samples.
• Provides a long-term, consistent time series from 1982 to 2020, addressing limitations like orbital drift in earlier versions.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Last update date is unknown; freshness unverified.
• Row count and file size are unknown, which may limit suitability assessment.

Provenance

Source

Peking University (Author: Muyi Li)

Collection Method

Generated using a machine learning model trained on Landsat samples to consolidate AVHRR and MODIS data.

Time Range

1982 to 2020

Geography

Global