The app said 18°C, looked good, no complaints. You head outside and within five minutes you're cold and wondering what went wrong. Or the opposite: the forecast said grey and 14°C, you dressed heavily, and it turned out to be perfectly pleasant. Weather apps get it right often enough that we trust them, but they get it wrong in predictable ways.
Where the data comes from
Weather stations measure temperature in shade, at a standardised height above the ground, protected from direct radiation. That's the right method for producing consistent, comparable readings across thousands of locations. The problem is that those stations are sometimes kilometres away from where you're standing, and local conditions can diverge significantly from what the station recorded.
Urban heat islands are a well-documented example. City centres can run 2 to 4°C warmer than nearby airports or rural stations, which is where weather data is often collected. If your app draws on an out-of-town measurement, its city forecast may be systematically cold.
Local wind effects are another one. A 10 km/h average wind speed across an area can mean sheltered streets with barely a breeze and exposed junctions with a genuine gust. The app knows the average. It doesn't know about the wind tunnel on the corner where you're waiting for the bus.
What apps handle well and what they don't
| Factor | How well apps handle it |
|---|---|
| Regional air temperature | Good |
| Local temp differences (sun/shade, city/rural) | Less reliable |
| Wind direction shifts | Generally good |
| Urban wind tunnels | Less reliable |
| Daily precipitation totals | Generally good |
| Exact timing of rainfall | Less reliable |
| Cloud cover percentage | Generally good |
| Sun on a specific street | Less reliable |
The pattern is consistent: things that vary at a regional scale are captured well. Things that vary at a street or neighbourhood scale are approximated at best.
The time lag problem
Forecasts are predictions, not measurements. By the time you look at a forecast, conditions on the ground may have shifted. A front moving through faster than expected, a cloud that cleared an hour earlier than predicted, a temperature inversion that the model didn't capture. The gap between forecast and reality tends to be largest when the weather is changing quickly.
Can you wear shorts today?
Check it now for your location — free, no account needed.
How KorteBroekAan.nl fits into this
Using feels-like temperature rather than raw air temperature already makes the advice more realistic, because it incorporates wind, humidity, and cloud cover into the number rather than leaving them as separate icons you have to interpret yourself. That said, the site is also working from weather model data, which has the same station-distance and local-variation limitations as any other app.
The most useful thing is to combine the forecast with your own eyes. If it looks darker outside than the app suggests, or if it looks like the rain cloud is about to clear, trust what you're actually seeing. The app is useful for planning; your observation is more accurate for the moment you're actually standing there.
For a clearer picture of why the raw temperature number underpredicts how cold you can feel, this article covers why temperature alone isn't enough. The rest of the Weather Explained section goes through each factor in the calculation.
Further reading
Related articles in the Weather Explained section: