A new and improved soil map for Africa

Recently I came across a news article about a new digital soil map of Africa which inspired me to do some online research and would love to share with you what I discovered for this week’s blog.

The iSDAsoil map gives soil variable data at a ‘farm-level’ 30m resolution for all potential agricultural land across Africa. This is the first time the soils of an entire continent have been studied at this level of detail. The 30m resolution means smallholder farmers across Africa now have access to specific, field-level information. iSDA claim the maps increase in resolution from other previously available datasets also increases the accuracy of the data as it more closely reflects the real variation in soils on the ground instead of averaging over large areas.

The maps were created using machine learning to predict soil properties for ~24 billion locations across Africa which were standardised using over 100,000 soil samples from 15 datasets across Africa. This was then combined with climatic, vegetation and terrain data from satellite imagery to create maps of over 20 soil variables (Hengl et al., 2020). 

The soil variables mapped include chemical properties such as carbon, nitrogen, pH; physical and landscape properties such as slope angles, soil content and texture; macro and micronutrients of the soils such as calcium, iron, magnesium and phosphorus; and agronomy information such as land cover and fertility capability. Values of each variable are available at two depth intervals (0–20cm and 20–50cm). Additionally, an uncertainty layer can be easily toggled on and off on the map therefore allowing the user to easily see the confidence of a value at a particular location.

Examples of soil variable maps from iSDAsoil. (a) Shows soil pH predictions at the 0–20 cm depth interval, (b) shows the corresponding uncertainty expressed as 1 s.d. prediction error. (c) Shows clay content at the 0–20 cm depth interval, (d) shows the corresponding uncertainty expressed as 1 s.d. prediction error, source: Hengl et al (2020) from iSDAsoil.

This open-access soil data could be incredibly useful to African farmers as achieving high soil quality will increase yields and water productivity of crops (Ali and Talukder, 2008). Farmers can now access field level soil information to choose the most suitable crops for their land and what fertilizers would be worth applying (if any) therefore maximising their productivity, profitability and limiting environmental damage. iSDA have future ambitions to continually improve and update the soil data available so I hope in future updates they will include variables such as soil water content which could be utilised by farmers managing droughts and/or implementing irrigation practices.

The variable which may prove most useful is the fertility capability classification layer which highlights the constraints of the soil which need to be dealt with in order to make the soil productive and fertile. This layer therefore summarises raw data of many other variables and formats them into one layer producing a map which can be easily used by farmers for agronomic decision making to improve their land. However, the database should be made more accessible to those that would benefit most from the data by, for example, publishing it in native languages. The iSDAsoil map is an exciting step towards a more data-driven approach to agriculture across Africa leading to decisions which could increase sustainable economic and environmental agricultural practices for individual smallholder farmers.