An Integrated Sensitivity and Surrogate Modelling Framework for Stakeholder-Aware Groundwater Management under Salinity Stress

Coastal aquifers are the hidden reservoirs that feed cities and farms along the world's shorelines, but they exist in a delicate balance. Beneath the sand and gravel, fresh water from rain and rivers floats atop a denser layer of saltwater that has seeped in from the ocean. This boundary is not fixed; it shifts constantly. When people pump too much fresh water from the ground, the pressure drops, and the heavy saltwater pushes inland, turning the fresh supply into a brine that cannot be used for drinking or irrigation. Managing this invisible war requires knowing exactly how much water can be taken from which wells without triggering a collapse of the entire system. For decades, scientists have tried to solve this by running complex computer simulations that mimic the underground flow, but these calculations are so heavy and slow that testing every possible pumping strategy becomes impossible.

A team of researchers has now developed a new way to navigate this problem, one that is faster, clearer, and designed to help the people who actually depend on the water make decisions. Focusing on the Kahak aquifer in central Iran, a region where a saline lake threatens nearby farmland, the team created a framework that replaces slow, black-box computer models with transparent, equation-based tools. Instead of treating the underground system as a mysterious machine that only outputs a result, their method builds a set of clear rules that explain exactly how pumping a specific well changes the water level and pushes the saltwater front. By combining these clear rules with a method for finding the best compromises between competing needs, the researchers were able to identify pumping strategies that satisfy farmers and protect the land, all while reducing the time needed to find a solution by nearly 98 percent.

The story begins in the southern Qom basin, where nearly 1,500 wells draw water for agriculture and domestic use. The area sits next to a salt lake, and the constant pumping has already pushed the salty water several kilometers inland, threatening to ruin the soil for thousands of hectares of farmland. In the past, trying to figure out how to stop this would require running a high-fidelity computer model thousands of times, each time adjusting the pumping rates for different wells to see what happens. This process is like trying to find a needle in a haystack by checking every single piece of hay one by one; it is accurate but takes far too long to be useful for real-time decision-making. The researchers realized that to manage the aquifer effectively, they needed a tool that could predict the outcome of a pumping plan instantly and explain why that outcome happened.

To build this tool, the team first used the slow, detailed computer model to generate a massive database of possible scenarios. They simulated how the aquifer would react to thousands of different pumping combinations, recording the changes in water levels and the movement of the saltwater. From this data, they trained a new type of computer program that does not just guess the answer but writes out the mathematical relationship between the input and the output. This program, a hybrid of two different analytical methods, learned to group the wells based on their location and the type of rock they sit in. It then derived specific equations for each group that describe exactly how much the water level drops and how far the saltwater moves when a well pumps at a certain rate. The result is a set of clear, readable formulas that act as a fast-forward version of the complex simulation, allowing the researchers to test thousands of strategies in the time it used to take to run just one.

With this fast and transparent model in hand, the researchers turned to the human element of the problem. In a region with five distinct agricultural zones, each group of farmers has its own needs and its own fears about the saltwater. Some zones are closer to the lake and more vulnerable, while others are further away. The goal was not just to find the mathematically perfect solution, but to find a solution that the stakeholders could agree on. The team used an algorithm to generate a wide range of possible pumping plans, each representing a different trade-off between how much water could be extracted and how much the saltwater would advance. They then applied three different methods of conflict resolution to these options: one that looked for the most balanced gain for everyone, one that ranked options based on how many people preferred them, and one that focused on the most robust plan in case of uncertainty.

The analysis revealed that there is no single perfect answer, but there are a few strategies that stand out as acceptable to all parties. The researchers found that 92 different pumping plans were mathematically efficient, but only three of them were simultaneously acceptable to all five stakeholder groups. Interestingly, while the different conflict-resolution methods picked different top choices, they all agreed on a second-best option, suggesting a strong consensus on a viable path forward. The most promising scenarios showed that by carefully redistributing where the water is taken from—pumping less from the wells closest to the salt lake and more from those further inland—it is possible to satisfy about 65 percent of the total water demand for all stakeholders while keeping the saltwater intrusion in check.

The study highlights a crucial insight: it is not just the total amount of water pumped that matters, but where it is pumped from. The researchers demonstrated that shifting the pumping load away from the sensitive edges of the aquifer can significantly slow the advance of the saltwater without drastically cutting off the water supply. This finding challenges the idea that the only solution is to simply pump less water overall. Instead, it suggests that smart management of the spatial distribution of wells can protect the resource while maintaining productivity. The new framework allows decision-makers to see these trade-offs clearly, understanding exactly how a change in one well affects the whole system, rather than relying on opaque computer outputs.

Ultimately, the work offers a practical path forward for managing groundwater in a world where fresh water is increasingly scarce and the threat of saltwater intrusion is growing. By replacing slow, unexplainable simulations with fast, transparent equations, the researchers have created a tool that can be used to negotiate and plan in real time. The framework successfully reduced the computational cost of finding these solutions by 98 percent, making it feasible to run the complex analyses needed for sustainable management. While the study was conducted in a specific region of Iran, the method is designed to be adaptable to other coastal aquifers facing similar pressures. The results suggest that with the right tools, it is possible to balance the competing needs of agriculture and environmental protection, ensuring that the water beneath our feet remains fresh for generations to come.