Water Availability as a System: Lessons from Barbados That Still Hold Today

In the summer of 2013, I arrived in Barbados with a research mandate and a question that sounds deceptively simple: how much water is actually available here, and for whom?

What followed was a ten-week deep dive at the Centre for Resource Management and Environmental Studies (CERMES) at the University of the West Indies. The result was the Speightstown Water Availability Model — SWAM — a System Dynamics Model built to assess water availability across a 31.3 km² catchment in the Parish of St. Peter. Over a decade later, the core insight from that project still shapes how I think about water systems – it is less about the numbers and more about the relationships.

The Paradox of a Water-Rich Island

Barbados is not an obvious candidate for water stress. The island receives 127 to 152 centimetres of rainfall annually, and its distinctive limestone geology acts like a natural sponge, funnelling precipitation into a productive aquifer system that supplies approximately 98.6% of the country's potable water. Access is near-universal: 100% of the population has access to drinking water, and 99% have piped connections.

And yet, water managers in Barbados face compounding pressures that can't be captured by a single supply-demand ratio. The tourism sector alone brings 400,000 to 500,000 visitors per year, many arriving from high-consumption countries where water scarcity is abstract. The island's celebrated "Platinum Coast", which is a stretch of luxury hotels and private villas along the west, sits directly above the island's most important groundwater resources. And climate-driven variability is steadily eroding the predictability of aquifer recharge.

Barbados has responded with desalination: three plants now operate on the island, the largest producing 7.9 million gallons per day or roughly 20% of the potable supply. It's an impressive investment. Desalination is also a question about cost of energy — an important question for an island that imports the majority of its energy through diesel.

But desalination doesn't solve the governance question: who gets water, at what cost, with what reliability, and at what risk to the underlying system. These questions were at the core of the research into a Speightstown Water Availability Model (SWAM).

Why the Index Matters Less Than the Process

SWAM was built around the Water Poverty Index (WPI), a multi-component framework developed by Caroline Sullivan that evaluates water availability across five dimensions: Resources, Access, Capacity, Use, and Environment. Each sub-model captures a different facet of the water system, from physical aquifer dynamics to the affordability of piped water for low-income households.

The WPI produces a score. But here's what I learned building SWAM: the score is almost beside the point.

The real value of constructing an integrated model isn't the output number. It is the discipline of mapping the linkages. When you sit down to build a causal loop diagram for a place like Speightstown, you're forced to articulate how tourism growth affects seasonal demand, how infrastructure leakage affects effective supply, how land-use changes affect aquifer recharge, and how household income affects water affordability. These connections exist regardless of whether anyone models them. The model just makes them visible and navigable.

This is a criticism the WPI's own developers have made publicly: the development process yields more useful information than the final index value. I saw this firsthand. The sub-models for Capacity and Access, built from Barbados Statistical Services data, surfaced questions that no physical hydrology study would have raised — questions about educational attainment, land ownership, and health outcomes as proxies for a community's capacity to manage water stress. These are the variables that determine whether a household bounces back from a water outage.

System Dynamics as a Language for Water Managers

The modelling platform for SWAM was VENSIM, developed by Ventana Systems. System Dynamics Modelling (SDM) is well-suited to water availability problems for a specific reason: the relationships in a water system are rarely simple ratios. They involve time delays, feedback loops, and non-linear thresholds.

Consider aquifer recharge. Rainfall doesn't translate immediately or uniformly into groundwater storage. Vegetation intercepts a portion; buildings capture another fraction. What infiltrates must pass through the soil layer, where it competes with evapotranspiration demand. The recharge that eventually reaches the aquifer faces pressure from coastal diffusion (a slow but continuous loss to the sea). Build in the effects of salinization from overdraft, and the system becomes genuinely complex.

SDM handles this naturally. It also handles the integration of qualitative and quantitative data in a single model, which is a practical necessity in places like Barbados, where rich longitudinal data exists for some variables and anecdote-level knowledge exists for others. The Access and Capacity sub-models in SWAM relied heavily on Census data and qualitative indicators. The Resource sub-model was anchored in gauge station precipitation records from 1988 to 2004. A single coherent model held both.

What SDM cannot do well is predict precise future quantities. The model is most useful as a vulnerability map: it shows which parts of the system are most sensitive to change, and where management interventions are likely to have the greatest leverage. That's a different product than a supply forecast, and it serves a different purpose.

What Was Left Unfinished — and Why It Still Matters

Honest research acknowledges its limits. SWAM had several.

The connection between the Environment sub-model and the Resource sub-model was never fully quantified. The causal relationship exists. Water quality degradation affects effective supply. However, translating that into defensible equations required data that wasn't available within the scope of the project. The sub-models ran in parallel, and their outputs were compared qualitatively rather than linked computationally.

Rainwater harvesting was identified as a significant gap. The Speightstown area, with its mix of agricultural land, residential development, and commercial rooftops, has meaningful harvesting potential. But estimating it required a rooftop inventory (i.e. dimensions, slope, and material for each structure) that couldn't be completed with one researcher and a ten-week timeline. The Caribbean Environmental Health Institute publishes a practical methodology for exactly this calculation, developed with UNEP support. It probably remains an open opportunity for the area.

Perhaps most significantly, SWAM was built without meaningful community participation. Participatory System Dynamics, where stakeholders actively shape the model's structure and assumptions, was recognized in the literature as producing better models and more durable outcomes. It also builds trust. A water model built with a community is a tool that community can advocate for; a model delivered to a community is just a report.

From Barbados to the Boardroom

The work in Barbados was academic. What it produced, though, was a way of thinking about water that I've carried into every project since.

Water availability is not a tap pressure reading. It's an emergent property of a system that is shaped by physical hydrology, but also by infrastructure investment decisions, land-use patterns, pricing structures, demographic change, and the institutional capacity of the organizations responsible for management. Model one piece in isolation and you will consistently misunderstand the whole.

This is increasingly relevant beyond the public sector. Corporations operating in water-stressed regions face exactly the same systems challenge. The question "how much water do we have access to?" can only be answered honestly by mapping the surrounding system — competing demands, aquifer health, regulatory exposure, infrastructure reliability, climate trajectory. The organizations that understand this are starting to invest in structured water assessments rather than water meter readings.

The Speightstown Water Availability Model was built for a small catchment on a Caribbean island in the summer before smartphones became ubiquitous. The data was imperfect, the model incomplete, and the timeline compressed. But the framing was right: water availability is a relationship between physical systems and human systems, and it can only be understood by modelling both.

That framing has aged well.

Semyon Chaymann is the Founder and CEO of HydraLink Infrastructure Solutions, a digital water consultancy based in Thornbury, Ontario. HydraLink works with municipalities, developers, and corporations on hydraulic modelling, water capacity planning, and corporate water stewardship strategy.