Skip to content

Methodology for producing spatial predictions of soil nutrients for SubSaharan Africa

Notifications You must be signed in to change notification settings

ISRICWorldSoil/AfricaSoilNutrients

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

7 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

SubSaharan Africa nutrient maps at 250 m resolution

Summary: This repository contains R code and some outputs of spatial predictions related with the production of soil macro and micro-nutrients maps for Sub-Saharan Africa at 250 m spatial resolution and for standard depth interval of 0–30 cm. Model training was run using soil samples from ca. 59,000 locations (a compilation of soil samples from the AfSIS, EthioSIS, One Acre Fund, VitalSigns and legacy soil data) and an extensive stack of remote sensing covariates, as well as landform, lithologic and land cover maps. For generating spatial predictions, an ensemble model from the machine learning algorithms random forest and gradient boosting, as implemented in R packages ranger and xgboost, was used. The results of cross-validation showed that apart from S, P and B, significant models can be produced for most targeted nutrients (R-square between 40–85 %). A limiting factor for mapping nutrients using the existing point data in Africa is high spatial clustering of sampling locations with many countries / land cover and land use groups completely unrepresented.

alt text alt text

Please cite as:

Download Maps

All maps are available for download under the Open Database License (ODbl) v1.0 and can be downloaded from www.isric.org without restrictions.

Disclaimer

These are results of spatial predictions based on using Machine Learning algorithms attached to the above-listed paper and hence some errors and artifacts are still possible. We aim at updating these maps regularly i.e. as the new training / point data arrives. See also ISRIC's general disclaimer. Predictions of potential yield (e.g. PotYield_crop1Maize_1km.tif) are based on the OFRA training data. These predictions are presented for experimental purposes only.

About

Methodology for producing spatial predictions of soil nutrients for SubSaharan Africa

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages