Agricultural Crop Area and Total Production
Access the data
Have more questions? Contact us
What is Atlas AI’s "Agriculture" layer?
Atlas AI’s Agriculture layer provides two types of statistics:
(1) crop area (hectares), i.e. land area under cultivation for specific crops, and
(2) total crop production (metric tonnes) as a measure of output from the cultivated area. The layer currently includes the crops maize, sorghum, and rice for the African continent (50 countries).
What is the spatial and temporal resolution of the Agriculture layer?
|Temporal Extent||2000 - 2020|
|Spatial Coverage||African continent (50 countries)
unavailable for Cape Verde, Comoros, Mauritius, and Sao Tome & Principe
|Spatial Resolution||Vectorized summary statistics at the national (admin 0), sub-national (admin 1), and local (admin 2 and below) levels of zonal aggregation|
We produce the Agriculture layer for every year from 2000 to 2020 inclusive. Datasets are available in the form of summary aggregations at zonal boundary levels such as national (admin 0), sub-national (admin 1), or local (admin 2 and below) in vector format (.shp) or tabular flat files (.csv).
What are the units of measurement for Agriculture?
Crop Area in hectares (ha) and Total Production in metric tonnes (mt)
Crop Area is a measure of agricultural intensity whereas Total Production is a measure of output.
How does Atlas AI estimate Crop Area and Total Production at the Admin 2 level?
Our methodology extracts trends in agricultural variables such as area, production, and yield for relevant crops. These trends enable us to make annual predictions for both Crop Area and Production for maize, sorghum, and rice.
- We aggregate SPAM data for area, production, and yield everywhere it is available in 2000, 2005, 2010, and 2017, at the Admin 2 level.
- We then fit a linear regression to these observations for area, yield, and production for each crop (maize, sorghum, and rice) and for each Admin boundary.
- Regression model parameters are corrected for outliers and then used to make predictions for both area and production for maize, sorghum, and rice annually.
- We validate the constructed statistics layers at the Admin 2 level and at the Admin 0 (country) level, based on the best published data currently available.
What are the main data sources we use?
We use the following publicly available data sources for model inputs, ground truthing, and model calibration. The primary input data for these statistics comes from International Food Policy Research Institute's (IFPRI) Spatial Production Allocation Model (SPAM). We also use country-level statistics from the United States Department of Agriculture (USDA), and local estimates from Ethiopia's Central Statistics Agency (CSA) for validation. These input data sources are described in detail below.
|data source||how we use|
|Spatial Production Allocation Model (SPAM)|
|The International Food Policy Research Institute's (IFPRI)’s HarvestChoice Spatial Production Allocation Model (SPAM) provides data on more than 40 crops, across five agricultural variables and six technologies.||We aggregate SPAM outputs from 2000, 2005, 2010, and 2017 for three crops (rice, maize, and sorghum), and three variables (physical area, yield, and production) for all years, to estimate area and production for maize, sorghum, and rice.|
|Production, Supply, and Distribution (PSD)|
|The United States Department of Agriculture (USDA) maintains the PSD Database as the official USDA data on production, supply, and distribution of agricultural commodities for the United States and key producing and consuming countries. This dataset provides statistics for barley, maize, millet, oats, rice, rye, sorghum, and wheat for several countries across the globe from 1960 to 2020.||We integrate country level agricultural statistics for maize, sorghum, and rice between 2000 to 2020 to validate our national-level (admin 0) statistics.|
|Ethiopian Central Statistics Agency (CSA) Crop Statistics|
|The Government of Ethiopia Central Statistics Agency provides all official surveys and censuses used to monitor economic and social growth.||We employ Admin 2 level statistics for area and production of maize, sorghum, and rice in Ethiopia between 2006 and 2016 to validate our Admin 2 level statistics.|
How can I learn more?
For more details on our methodologies for agriculture and published model metrics, we recommend the following citations as further reading:
- Jin, Z., Azzari, G., You, C., Di Tommaso, S., Aston, S., Burke, M., & Lobell, D. B. (2019). Smallholder maize area and yield mapping at national scales with Google Earth Engine. Remote Sensing of Environment: 228, 115-128 (2019). PAPER
- Jin, Z., Azzari, G., Burke, M., Aston, S., & Lobell, D. B. Mapping smallholder yield heterogeneity at multiple scales in Eastern Africa. Remote Sensing: 9(9), 931 (2017). PAPER
Please reach out to us at firstname.lastname@example.org if you have trouble accessing the full papers.
How can I explore the data?
Please reach out to us email@example.com if you have trouble creating a free evaluation account on our proprietary platfrom ApertureTM, or would like to request a free data sample for evaluation prior to purchase.
Access the data
Have more questions? Contact us