# Fine-scale mapping of irrigation suitability in South Africa using ensemble modelling

## TL;DR

This study uses machine learning to map irrigation suitability in South Africa, helping improve food and water security.

## Contribution

The novel approach combines multiple machine learning models to enhance irrigation suitability mapping accuracy.

## Key findings

- The ensemble model achieved a TSS of 0.66 and an AUC of 0.90, outperforming individual models.
- Land use, population density, and elevation were key factors in determining irrigation suitability.
- Northern Cape and Western Cape provinces showed the largest suitable areas for irrigation.

## Abstract

Food insecurity, exacerbated by a growing population and environmental change, poses a significant challenge in Southern Africa. Enhancing agricultural productivity through efficient irrigation practices is crucial for achieving food and water security and sustainable development goals. This study applied an ensemble modelling approach to identify and assess irrigation suitability areas across South Africa, combining the predictive power of Random Forest, Extreme Gradient Boosting (XGBoost), and Gradient Boosting Machine (GBM) algorithms. These machine learning models were applied using cropland presence/pseudo-absence data and a suite of predictor variables. The ensemble model, leveraging a weighted averaging approach based on individual model performance, outperformed the individual models, achieving a TSS of 0.66 and an AUC of 0.90. Land use, population density, and elevation were identified as key factors determining irrigation suitability. The ensemble model also revealed substantial spatial variation in irrigation potential across South Africa, with the Northern Cape and Western Cape provinces exhibiting the largest suitable areas. The results provide critical information for targeted irrigation development, enabling efficient resource allocation, and maximising agricultural productivity. This data-driven approach offers a robust framework for sustainable agrarian planning in the face of increasing food demands and climate change, contributing to enhanced food security and economic development in South Africa.