From ideas to impact: ‘Better forecasts need a new kind of data’

Weather forecasts are based on physical models and measurements from weather stations, satellites and other sources. However, the atmosphere close to the ground, where wind and solar power are produced, is difficult to measure accurately.

WeatherSens AI aims to fill this gap. Its drone flies vertical profiles through the lowest layer of the atmosphere, measuring wind, temperature, humidity and air pressure from near ground level up to a few hundred metres. A machine learning model uses these measurements to correct and refine conventional weather forecasts locally.

For wind power producers, more accurate local forecasts can have direct financial value. Wind farms report their expected electricity production to the transmission system operator in advance, and differences between forecast and actual output have a financial cost.

‘We started basically from zero,’ says Li. ‘First, we had to find out whether local weather data was a question of global interest, and whether it mattered enough for someone to be willing to pay to solve it.’ The team concluded that it did. The next step was to prove it.

What did the innovation grant make possible?

The team had an early prototype when it received an Aalto University student innovation grant. At this stage, many of the key assumptions behind the product idea were still untested. ‘It was so early stage, and such a big risk to put our own money in,’ says Laakso. ‘The grant let us test some risky stuff we wouldn't have been able to do otherwise.’

The grant provided resources to develop and test the concept further. The next version was smaller and designed to work with a drone. It showed that the sensing system could produce a usable signal rather than just noise.

‘We don't rerun it. We take the national forecast and correct it locally with real measurements. That is cheap to run, and the data comes from where you actually are. Better forecasts need a new kind of data.’ 

Building hardware is naturally slower than developing software, Nordström admits. ‘Slower, but harder to copy.’

What did you learn from customers?

The team initially approached wind power producers. There was interest, but energy companies wanted proven accuracy, and the system was not yet ready to demonstrate it. So the team tried another market: wakeboarding companies, for which local wind conditions also matter. Smaller companies were willing to try an early prototype, and the team secured its first customers. The need was the same, but in a seasonal business, the pilots did not become permanent.

Looking back, Li sees the experience as an important lesson. The need and interest were clear in the energy sector, but the team had not yet invested enough time in it. ‘We should have just kept on building and pouring more of our focus and validation into just the energy market. It's not like you can make a scrappy product and people want it. Precision is key to solving this problem.’

What happens next?

The team now works by constantly challenging its own assumptions. ‘We want to disprove our hypotheses as fast as possible,’ says Li. ‘We don't ask whether this sensor is in the right place. We ask: Can you skip this sensor? Can you buy this already made? Can this be done in another, more efficient way?’

The goal is a system that produces data accurate enough for customers to pay for and rely on. The team is now validating its accuracy in field tests, comparing measurements against reference measurements. ‘We're looking for partners who want to help solve problems in weather forecasting and data collection – in other sectors too,’ says Nordström.

Student innovation grants: backing those who try

The student innovation grant offers Aalto student teams fast, flexible support of €500–5,000 to build a prototype, test an idea or find out whether a product concept works in practice. The grant can make the difference between a promising idea staying on paper and getting the chance to develop.

WeatherSens AI used its grant to build and test early prototypes. The funding gave the founders room to take risks, learn from what did not work and keep developing their idea.

By donating to entrepreneurship and innovation, you can support students at a stage when even a small contribution can make a real difference. You also help create an environment where good ideas have the chance to grow into tomorrow's solutions.

Donate today – even a small gift can fuel great ideas