The aurora borealis is like a distant theater: the show may be on, but whether the audience sees it depends on the curtain. Here, the curtain is clouds and moonlight. Earlier forecasts tried to predict both the occurrence of auroras and viewing conditions at once, mixing everything together. That's like predicting whether a play will be well received without separating the actors' performance from the visibility from the back row.
The authors of the study took a different approach. They built a two-step system. The first step is an algorithm trained on years of observations that uses 51 parameters of solar wind and Earth's magnetic field to compute the probability that an aurora is actually present. The second step is a simpler model that, given that an aurora exists, estimates whether clouds might hide it or the Moon might wash it out. Multiplying these two probabilities gives the overall chance of seeing the aurora with the naked eye.
Key clues for the first step turned out to be the interaction of the planetary geomagnetic activity index with the night side of Earth, local magnetic time, and the distance to the auroral oval — together they account for 39% of the predictive power. The system already works as a prototype and has been tested at three different sites in Scandinavia. This approach not only helps aurora hunters plan their outings but also gives researchers a tool to study the Sun–Earth connection.
The auroral oval is a ring around the magnetic pole where auroras occur most frequently. Its diameter is about 3000 kilometers, and it is slightly shifted toward the night side of Earth.