Geospatial stable isotopes signatures of groundwater in United Arab Emirates using machine learning

Abstract

Study region: The research is conducted in United Arab Emirates (UAE) where limited water resources are negatively impacted by both natural and human-induced factors. Study focus: The focus of this study is to establish a cost-effective and accessible isotopic database to identify the sources of recharging country-wide aquifers in UAE. The hydrogen (δ2H) and oxygen (δ18O) data of five aquifer systems is integrated into a publicly accessible web mapping application. In addition, the Machine Learning (ML) approach is employed to develop a novel isotopic boundary clustering tool for the various aquifers. New hydrological insights: The results indicate a cost - time-effective web application could assist in any future research. The results also revealed that (1) the eastern gravel plain, ophiolite, and northern carbonate aquifers are isotopically comparable, indicating a recharge by modern precipitation, (2) Mixing and upward leakage along the deep-seated faults with modern precipitation is reflected by the isotopic signature of Jabel Hafeet carbonate aquifer and (3) The isotopic values of the coastal aquifers suggested an impact of the sea water intrusion. The ML clustering categorized the isotopic data into four main boundary decision zones (DZ) that are different from the aquifer boundaries indicating various recharge sources. This study provides a new country wide geospatial stable isotope distribution of groundwater which is vital for the sustainable management of water resources in arid regions.

Keywords

• Groundwater
• Stable isotopes
• Spatial-clustering, machine learning, UAE