Divergent global land change scenarios from 2001 to 2019

Global land use and land cover (LULC) change plays a critical role in environmental change, biodiversity conservation, climate regulation, and ecosystem services. Over the past few decades, multiple global LULC datasets—such as MCD12Q1, ESA CCI-LC, GLC-FCS30D, HYDE 3.3, HILDA+ , and LUH2—have been developed using remote sensing or historical reconstruction. In this study, we comprehensively analyse the similarities and discrepancies among six global LULC datasets, revealing land-cover changes and transitions from 2001 to 2019. The most significant global LULC changes during this period include forest loss, cropland gain, and settlement expansion. The mutual conversion between forest and cropland was particularly prominent, accounting for around 20 % of all land-use transitions. However, notable discrepancies exist among the datasets in capturing global LULC changes. For example, estimates of net forest area change vary from an increase of 1 . 10 × 10 5 km 2 to a decrease of 1 . 55 × 10 6 km 2 , while cropland area changes vary from a rise of 1 . 59 × 10 6 km 2 to a decline of 4 . 11 × 10 4 km 2 . Settlement areas saw the most significant increase in the GLC-FCS30D dataset. These differences were most pronounced in Africa and South America for forest loss and cropland gain, and in Asia for settlement expansion. Comparisons with FAOSTAT show that GLC-FCS30D and HYDE 3.3 align closely with cropland changes, while HILDA+ closely matches forest change data. These findings highlight the importance of harmonizing LULC datasets to improve the accuracy of global land change assessments and support better environmental management strategies. • Analyzed six global LULC datasets for land-use changes and transitions from 2001 to 2019. • Notable Dataset discrepancies exist among the datasets in capturing global LULC changes. These differences were most pronounced for forest loss and cropland gain, and settlement expansion. • Notable forest-cropland conversion (around 20 %). • Emphasizes need for harmonized LULC datasets for accurate assessment.