agroimpacts/mapping-africa

# High resolution, annual cropland and landcover maps for African countries ## Background 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. ## Datasets There are two types of data currently available: 1. **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. 2. **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. ## Accessing data ### From S3 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): ``` bash aws s3 ls s3://mappingafrica/ ``` ``` bash PRE croplands/pmtiles/ PRE croplands/mbtiles/ PRE landcover/ ``` To download a dataset, please use the following an example command: ``` bash aws s3 cp \ s3://mappingafrica/landcover/tanzania_2018.tif \ ~/Desktop/ ``` ``` bash download: s3://mappingafrica/landcover/tanzania_2018.tif to ../../.. /Desktop/tanzania_2018.tif ``` That 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). …