Modelling Extreme Water Levels of a Hydroelectric Dam under Natural Truncation

Imagine you are trying to predict the weather, but instead of worrying about whether it will rain tomorrow, you are trying to guess the wildest, most impossible storm that could ever hit your town. This is the job of a branch of math called Extreme Value Theory. While normal statistics focus on the "average" day—like the typical temperature or the usual amount of rain—Extreme Value Theory is obsessed with the outliers: the once-in-a-century floods, the record-breaking heatwaves, or the massive waves that crash over the highest cliffs. It asks a simple but terrifying question: "How high can the water really get?"

Why does this matter? Because when we build things like bridges, dams, or skyscrapers, we need to know the limits. If we guess wrong and think a flood will only reach the first floor, but it actually reaches the roof, the consequences are disastrous. Engineers use these mathematical tools to design safety nets, ensuring that when nature throws its worst punch, our structures can take it. But there's a catch: nature sometimes has a "ceiling." Just as a glass of water can't spill higher than the rim of the glass, a river trapped behind a dam can't rise higher than the top of the dam. This physical limit changes the math, and figuring out how to calculate the risk when a "ceiling" exists is the puzzle this study solves.

---

The Story of the Akosombo Dam and the Invisible Ceiling

Deep in Ghana, the Akosombo Dam stands as a giant guardian, holding back the Volta River to generate electricity for the country. It's a massive machine, but it has a very strict rule: the water inside cannot rise above a specific line, known as the maximum operating level of 278 feet. If the water tries to go higher, it spills over the top, which is dangerous.

For a long time, scientists have tried to use math to predict how high the water might get during a massive storm. They used standard tools that assume water levels could, in theory, go up to infinity if the storm was strong enough. It's like trying to predict how high a basketball could bounce if you kept hitting it harder and harder, ignoring the fact that the ceiling of the gym would stop it. The problem is, for a dam, the "ceiling" (the dam's crest) is very real and very close to the highest water levels we've ever seen.

This study, led by researchers from the University of Ghana, decided to fix this math problem. They asked: "What if we build a model that knows the water can't go higher than the dam's top?" They compared three different ways of doing the math:

1. The Old Way (GEV): A standard method that looks at the biggest water levels from each year.
2. The Popular Way (GPD): A method that looks at every time the water gets "too high" (above a certain line) but still assumes there's no hard ceiling.
3. The New Way (RT-POT): A special model that explicitly accounts for the "natural truncation"—the fact that the dam physically stops the water from rising further.

The Detective Work: Finding the Limit

The researchers looked at 49 years of daily water level data from the Akosombo Dam, totaling 17,533 measurements. They found that the water levels ranged from a low of 197.40 feet to a record high of 277.54 feet. That record high is incredibly close to the 278-foot limit, meaning the dam has been operating right at the edge of its capacity.

To solve the puzzle, they first had to pick a "threshold"—a water level high enough to be considered an "extreme" event. After analyzing the data with some fancy graphs (which act like a detective's magnifying glass), they decided that 272 feet was the perfect line to draw. Any water above this line was part of the "extreme tail" they needed to study.

Then came the big test: Is there a ceiling?
They ran a statistical test to see if the data showed signs of being "truncated" (cut off) by the dam's top. The result was a resounding yes. The data showed "rough truncation," meaning the dam's physical top is so close to the highest water levels that it is actively shaping the statistics. The standard models that ignored this ceiling were like trying to measure a room while pretending the roof doesn't exist; they were overestimating how high the water could go.

The Results: A Safer Picture

When the researchers applied their new Right-Truncated Peaks-Over-Threshold (RT-POT) model, the numbers changed in a reassuring way.

• The 100-Year Flood: Using the new model, they calculated that a "100-year return level" (a flood so big it happens, on average, once every century) would reach 277.13 feet.
• The Absolute Limit: The model estimated the absolute highest the water could possibly get under current conditions is 277.60 feet.

Here is the most important part: Both of these numbers are below the dam's maximum operating level of 278 feet.

This means that, according to the study, the dam has a safety margin. Even in a massive, once-in-a-century storm, the water is predicted to stop at 277.13 feet, leaving a gap of 0.87 feet before it would spill over. The estimated absolute limit of 277.60 feet is still below the 278-foot crest.

The researchers found that the chance of the water actually exceeding the 278-foot limit is incredibly tiny—so small it's almost zero. They also noted that the highest water level ever recorded in their 49-year dataset was 277.54 feet, which is a rare event with a probability of only 0.0008 (or 8 in 10,000).

Why This Matters

The study suggests that the old models, which didn't account for the dam's physical top, were predicting water levels that were too high (some previous studies suggested limits as high as 280.18 feet). By ignoring the "ceiling," those models were making the dam look more dangerous than it actually is.

The new RT-POT model is more precise because it respects the laws of physics. It tells engineers and dam managers that the current structure is likely strong enough to handle the worst storms nature has thrown at it in the past 50 years. The "ceiling" isn't just a theoretical idea; it's a real barrier that keeps the water in check.

However, the authors are careful to say this is based on the assumption that the climate and the dam's behavior haven't changed drastically over time. They suggest that future work should look at how changing weather patterns might affect these numbers. But for now, this study provides a clearer, more realistic map of the danger zone, showing that the Akosombo Dam has a comfortable buffer between its highest recorded water and the point of disaster.