Increasing Water Use in Global Copper Production Threatens Freshwater Availability

Global demand for the critical raw material copper is projected to increase by approximately 75% by 2050, driven by the growing uptake of advanced technologies and rising material demand for housing and energy infrastructure, particularly in low- and middle-income. While copper recycling is expected to increase, with recycled materials expected to supply about half of total demand by 2050, this rising demand will also require significant expansions in copper mining.

One crucial challenge for the expansion of copper mining is its water use, which is critical at different stages in metal mining and processing. Differences in the processes in place at each site are, therefore, important determinants of water demand and water intensities, which can vary considerably among extraction sites.

Depending on the geographical location of the mine, availability and sources of water as well as the impacts related to their use can vary considerably. As metal mining is often taking place in vulnerable hydrological settings, water use can have considerable impacts on local water scarcity and water quality, leading to or aggravating competition for local water resources which often result in open conflicts. It is hence essential that assessments of mining impacts on water resources consider the local environmental and socio-economic contexts. However, comprehensive assessments of water use in copper mining are limited due to the lack of consistent datasets tracking the actual demand at individual mines worldwide.

This work presents a comprehensive spatially explicit analysis of water use in copper mining, taking into consideration multiple determinants of water input at mine site level. We employ machine learning (ML) to estimate mine-specific water use, which we then integrate with satellite-derived trends in freshwater availability and local water scarcity indicators at each mining site to assess the pressure on water resources. The ML models are trained on available water use data compiled from multiple sources with a predictors space including a wide set of variables: production quantity, primary or by-product classification, mine type, geological setting, process route, ore grade, and potential evapotranspiration. Since data on water is not often reported, the ML models enable to comprehensively estimate global water use across all known copper-producing mines.

The analysis reveals spatial and temporal variations in water use across global copper production in relation to local characteristics such as process types in place, geological setting, and water availability. Results show that global water intensity is two-fold higher than previously known. Between 2015 and 2019 copper mines withdrew 13.6 trillion litres of water, with water use increasing at a rate 50% higher than copper production. In 2019, more than half of global copper output came from sites with decreasing freshwater availability and rising water demand, with notable contribution from Latin America, the largest copper producer and water user.

Our analysis is relevant to public and corporate policy, revealing concerning spatial patterns on water use that can threaten future mine production, cause local conflicts and ultimately put global sustainability strategies at risk.