This site provides links to view and obtain high resolution cropland and landcover maps developed by Clark University’s Agricultural Impacts Research Group for selected African countries using various machine learning approaches applied to Planet imagery.
There are two types of data currently available:
-
cropland: Annual (beginning in year 2018) crop field boundary maps of several African countries, developed using several different modeling approaches applied to Planet imagery (Estes et al, 2022a; Estes et al, 2022b; Wussah et al, 2023). Data are provided as vectorized boundaries, in both pmtile and geoparquet formats. These datasets are under active development, and more countries and annual maps are updated as they are created.
-
land cover: A 2018 multi-class land cover map for Tanzania developed using U-Net applied to Planet imagery and Sentinel-1 time series derivatives (Song et al, 2023). See here for more detail on the methods and larger project (led by Dr. Lei Song) for which this map was created.
These datasets can be downloaded from this bucket by AWS account holders. Data are stored under the following prefixes:
└── mappingafrica/
├── croplands/
│ ├── pmtiles
│ └── geoparquet
└── landcover
These can be viewed using the AWS command line interface (CLI):
aws s3 ls s3://mappingafrica/ PRE croplands/pmtiles/
PRE croplands/mbtiles/
PRE landcover/To download a dataset, please use the following an example command:
aws s3 cp \
s3://mappingafrica/landcover/tanzania_2018.tif \
~/Desktop/ download: s3://mappingafrica/landcover/tanzania_2018.tif to ../../..
/Desktop/tanzania_2018.tifThat will download a map of predicted land cover for Tanzania for the year 2019 to your desktop (you might need to replace ~/ with the full path to your home directory).
The land cover map and ancillary data can also be downloaded from the Open Science Foundation, and model code is here.
The datasets can be viewed through the web map hosted here (and accessible from here).
Maps can also be loaded and displayed using a Jupyter notebook (see the example here.
Note that field boundaries will often appear misaligned with the underlying imagery in the webmap. In some cases, this is due to model error (see Usage below), but often this reflects the fact that the model was applied to Planet imagery collected in a different year than the basemap imagery, when the fields themselves were different.
Use of these maps is governed by the terms of the Planet NICFI participant license agreement.
Users should also be aware of these maps’ limitations, which are released “as-is” and contain errors. These include:
- False positives;
- Under-segmentation, where adjacent fields are grouped together, particularly in dense agricultural landscapes;
- Inconsistent mapping of fields between years (fields missed, falsely predicted, or under- or over-segmented for the same location in different years).
These errors can vary by region. Map users should undertake their own accuracy assessments that are relevant to their area and application of interest. To better understand the nature of such errors, how they can be detected, and in some cases, corrected, please refer to Xiong et al (2026). The methods include details on how to use polygon shape metrics, which are provided with our geoparquets, to detect fields that are under-segmented and bias-correct their areas.
If you use the these maps in a publication, report, or other research product, please cite according to these guidelines:
Please cite the following paper:
Song, L., Estes, A.B. & Estes, L.D. (2023) A super-ensemble approach to map land cover types with high resolution over data-sparse African savanna landscapes. International Journal of Applied Earth Observation and Geoinformation, 116, 103152.
Please cite the Mapping Africa dataset:
Estes, L.D., Essuman, G., Xiong, S., Abedi, R., Chakraborty, T. (2026). Mapping Africa: Annual high-resolution cropland field boundary maps. Agricultural Impacts Research Group, Clark University. https://github.com/agroimpacts/mapping-africa. Accessed [date].
In addition to citing the dataset citation above, please cite the publication(s) below, which differ by country as the model and production methods varied.
For the current Zambia maps, please cite the following paper, which describes both the maps and the methods used to generate them:
Xiong, S., Li, W., Hadunka, P., Ross, G.D., Chakraborty, T., Khallaghi, S., Abedi, R., Daum, K., Xue, K., Yao, Y.-T., Chilenga, A., Rufin, P., Khan, A., Potapov, P., Baylis, K., Caylor, K. & Estes, L. (2026). High-resolution remote sensing reveals medium-scale farms at the frontiers of agricultural change in Africa. agriRxiv, 2026, 20260367814.
Khallaghi, S., Abedi, R., Abou Ali, H., Alemohammad, H., Dziedzorm Asipunu, M., Alatise, I., Ha, N., Luo, B., Mai, C., Song, L., Wussah, A.O., Xiong, S., Yao, Y.-T., Zhang, Q. & Estes, L.D. (2025) Generalization enhancement strategies to enable cross-year cropland mapping with convolutional neural networks trained using historical samples. Remote Sensing, 17, 474.
The methods used to generate these maps are drawn from the following two papers:
Xiong, S., Li, W., Hadunka, P., Ross, G.D., Chakraborty, T., Khallaghi, S., Abedi, R., Daum, K., Xue, K., Yao, Y.-T., Chilenga, A., Rufin, P., Khan, A., Potapov, P., Baylis, K., Caylor, K. & Estes, L. (2026). High-resolution remote sensing reveals medium-scale farms at the frontiers of agricultural change in Africa. agriRxiv, 2026, 20260367814.
Muhawenayo, G., Robinson, C., Khanal, S., Fang, Z., Corley, I., Wollam, A., Gao, T., Strnad, L., Avery, R., Estes, L., Tárano, A., Jacobs, N. & Kerner, H. (2026). PRUE: A Practical Recipe for Field Boundary Segmentation at Scale. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 6484–6495.
Estes, L.D., Wussah, A.O. & Asipinu, M.D. (2022a) Final report - Phase 1: Creating open agricultural maps and ground truth data to better deliver farm extension services
Estes, L.D., Ye, S., Song, L., Luo, B., Eastman, J.R., Meng, Z., Zhang, Q., McRitchie, D., Debats, S.R., Muhando, J., Amukoa, A.H., Kaloo, B.W., Makuru, J., Mbatia, B.K., Muasa, I.M., Mucha, J., Mugami, A.M., Mugami, J.M., Muinde, F.W., Mwawaza, F.M., Ochieng, J., Oduol, C.J., Oduor, P., Wanjiku, T., Wanyoike, J.G., Avery, R.B. & Caylor, K.K. (2022b) High resolution, annual maps of field boundaries for smallholder-dominated croplands at national scales. Frontiers in Artificial Intelligence, 4, 744863.
Khallaghi, S., Abedi, R., Abou Ali, H., Alemohammad, H., Dziedzorm Asipunu, M., Alatise, I., Ha, N., Luo, B., Mai, C., Song, L., Wussah, A.O., Xiong, S., Yao, Y.-T., Zhang, Q. & Estes, L.D. (2025) Generalization enhancement strategies to enable cross-year cropland mapping with convolutional neural networks trained using historical samples. Remote Sensing, 17, 474.
Song, L., Estes, A.B. & Estes, L.D. (2023) A super-ensemble approach to map land cover types with high resolution over data-sparse African savanna landscapes. International Journal of Applied Earth Observation and Geoinformation, 116, 103152.
Wussah, A.O., Asipinu, M.D. & Estes, L.D. (2022) Final report - Phase 2: creating next generation field boundary and crop type maps: Rigorous multi-scale groundtruth provides sustainable extension services for smallholders
Xiong, S., Li, W., Hadunka, P., Ross, G.D., Chakraborty, T., Khallaghi, S., Abedi, R., Daum, K., Xue, K., Yao, Y.-T., Chilenga, A., Rufin, P., Khan, A., Potapov, P., Baylis, K., Caylor, K. & Estes, L. (2026). High-resolution remote sensing reveals medium-scale farms at the frontiers of agricultural change in Africa. agriRxiv, 2026, 20260367814.