This data integrates multiple sources of data, including high-resolution remote sensing data, model simulation data, and historical agricultural census data. Using 1950 to 2020 as the time scale and a spatial resolution of 1 kilometer by 1 kilometer, it reconstructs the historical dynamics of four major economic crops in South America, namely soybean, corn, wheat, and rice. The research results indicate that over the past 70 years, the farmland area in South America has rapidly expanded by encroaching on other vegetation. Specifically, soybean is one of the most significantly expanding crops, with its planting area increasing from almost zero in 1950 to 48.8 million hectares in 2020, resulting in a total reduction of 23.92 million hectares in other vegetation (i.e. forests, pastures/grasslands, and unmanaged grasslands/shrubs). In addition, the planting area of corn increased by 2.1 times from 12.7 million hectares in 1950 to 26.9 million hectares in 2020, while the planting area of rice and wheat remained relatively stable. The newly developed crop type data provides important insights for evaluating the impact of agricultural land expansion in South America on crop yield, greenhouse gas emissions, and carbon and nitrogen cycling. In addition, these data are crucial for formulating national policies, sustainable trade, investment, and development strategies aimed at ensuring food supply in South America and other regions, as well as achieving other human and environmental goals
Including high-resolution remote sensing data, model simulation data, and historical agricultural census data. Specifically, as follows:
(1) Using remote sensing and model-based LULC and crop type 155-160 data, cultivated land density maps and crop type base maps of South America were generated;
(2) Copernicus Global Land Services Land Cover Map (CGLS-LC100): PROBA-V 100m time series data and high-quality land cover training samples are used to construct a land cover classification model with a first level accuracy of 80%;
(3) Global 30 meter Land Cover Dynamic Monitoring Dataset FCS30D;
(4) Global Environmental History Database (HYDE version 3.2);
(5) Historical Land Dynamics Assessment+(HILDA+);
(6) Spatial Production Allocation Model (SPAM 2010);
(7) Global Agriculture Monitoring Organization for Earth Observation (GEOGLAM);
(8) Global Land Analysis and Discovery (GLAD);
(9) MapBiomas provides land use and land cover maps with a resolution of 30 meters for Argentina, Brazil, and Uruguay;
(10) Argentine multinational companies have provided detailed crop type maps with a resolution of 30 meters for Argentina from 2018 to 2022.
In order to reconstruct the dynamics of historical farmland and crop types, preprocessing is required for all datasets. Firstly, the high-resolution datasets (i.e. CGLS-LC100, GLC_FCS30D, GLAD, Argentina MNC, MapBiomas, and Uruguay LC) were aggregated to a resolution of 1 km to obtain partial farmland and crop types. Secondly, the bilinear method was used to resample HYDE3.2, SPAM2010, and GEOGLAM, achieving a spatial resolution of 1 km. Finally, the projection dataset was converted to WGS84 for further analysis, and all processes were carried out in Google Earth Engine.
Compared with existing data, reconstructed data has higher spatiotemporal resolution and can better capture the dynamics of crop type changes during historical periods.
This work is licensed under
CC BY 4.0 (Creative Commons Attribution 4.0 International License).
CSTR:11738.11.NCDC.ZENODO.DB6963.2025
10.5281/zenodo.14002960
