A simple thermometer outside the window and a barometer in the living room used to be all a home needed. These days, advanced weather stations pop up in gardens and on balconies, measuring the local temperature, pressure, wind and rainfall. For the owners, such observations are fun to follow, for those hoping for better weather forecasts, a treasure.

Marie Pontoppidan, researcher at NORCE and the Bjerknes Centre tries to improve the computer models that are used in weather and climate predictions.

In a recent study Pontoppidan, with colleagues from NTNU, NINA and the University of Warsaw, used data from privately owned rain gauges to complete the picture given by official weather stations. With a more detailed overview of historical weather, they were able to test how well weather forecasting models represent the real world.

"Precipitation can be very local," she says.

A heavy shower in Bergen in 2023 fully confirmed her words.

Much wetter than officially measured

Norway has around one thousand official weather stations, approved and run by the Norwegian Meteorological Institute. The stations are located around the country, ensuring a good overview.

But the distance between neighboring stations can be large. One thousand stations may sound like a lot, but with a terrain made up of fjords, mountains and valleys, it is not enough to cover the locally varying weather. Weather stations set up in people's gardens may catch variations that would otherwise go unnoticed.

During a heavy rain shower in August 2023, a privately owned weather station in Fyllingsdalen, southwest of the city center of Bergen, received almost sixty millimeters of precipitation in less than an hour.

Neither of the Meteorological Institute's 14 rain gauges in the Bergen area registered as much as half of this.

"But we have radar data confirming the observation," says Marie Pontoppidan.

Even when large-scale weather conditions favor showers in an entire region, each cloud drops its cargo over a limited area. One kilometer away the difference can be significant.

The local signature of showers makes it hard to predict the outcome. That August day in 2023, warnings of heavy rainfall had been issued for the internal parts of Rogaland, but the rain turned out to be heavier farther north.

Filling gaps

The technological development during the recent decades has changed both the format and the accuracy of weather forecasts. The rain, temperatures and winds in Norway are displayed with more and more detail, as in an image with an increasingly higher resolution.

Weather prediction models place a grid over the map and calculate the weather in each grid cell. The size of the grid cells in the model currently used for ordinary weather forecasts, is 2.5 kilometers.

In the models Marie Pontoppidan and her colleagues use for research, there may be as little as one kilometer between each point where the weather is estimated.

The disadvantage of increasingly detailed models is that they have surpassed the observations. The picture given by official weather stations is too coarse to compare with the model. As a result, no one will know how well the fine-meshed model-generated image resembles the weather that really occurred.

That is why Marie Pontoppidan investigates the possibility of using weather stations from people's gardens to fill the gaps between official measuring sites.

"The weather stations are not free," she admits. "But they are already out there. So many people have already bought such stations. It would be too bad not to use the data, provided they are good enough."

Handed out weather stations

In collaboration with the municipality of Sunnfjord, people living in the flood-exposed valley Viksdalen were given weather stations to place at home. Stations were also handed out in the municipalities of Kinn and Osterøy.

"Many people were highly motivated," says Marie Pontoppidan.

She explains that the density of stations in a region must reach a certain level before the observations can be used.

The garden stations are simpler than the official ones, with poorer wind screening, permitting rain drops to blow past or bounce out of the gauge during heavy rains. To allow for data quality checking, the observations must be compared with those from stations nearby. If major deviations are found at a station, its data will not be included in the analysis.

So far, combining rain registered in people's gardens with official data is on the research stage. Temperatures measured at private weather stations are already included in the Meteorological Institute's weather prediction system.

From gardens to cabins

On that heavily rainy day in August 2023, Marie Pontoppidan and her colleagues received data from 62 private weather stations in Bergen. Would higher numbers allow for improvements?

"The more, the better. But I would rather see new stations in sparsely populated regions, or even better, at people's cabins in the mountains," says the rain researcher.
In the coming years Marie Pontoppidan will be involved in the newly established Centre for Mountains in Transitions. Though her work will include making her own observations of rainfall, she will appreciate data from private weather stations.

Reference

Pontoppidan, M., T. Opach, & J. K. Rød. 2025. “Demonstrating the Added Value of Crowdsourced Rainfall Data in Complex Terrain.” Meteorological Applications 32, no. 5: e70108. https://doi.org/10.1002/met.70108.