From 10 days to 3 weeks
Since the start in 2007, Yr has had a weather forecast 10 days ahead, but now we are extending the forecast period to 21 days. A weather forecast this far into the future is typically more uncertain than a weather forecast for the next few days. Therefore, we have divided the forecast into three rolling seven-day blocks. The uncertainty in the forecast means that the most important thing is to give an indication of the trend in temperature and precipitation further ahead. Check it out, then!
Where is the weather forecast?
The new 21-day forecast is displayed in a point forecast, and not in map format. This means that if you search for a place located within the blue square in the figure on the right (covering parts of the Nordic region), you will get a 21-day forecast for this place. The forecast is located under the tab "21-day forecast".
Figure text: coverage area marked in blue for the 21-day forecast
The data used in the forecast
Weather models and updates
The new 21-day forecast is originally based on three different weather forecast models, MEPS, EC medium range and EC extended range, which all update at different time intervals and have different geographical resolutions. A significant effort has been made to make the new forecast a seamless forecast, which also matches the original 10-day forecast at all times. The first 10 days of the 21-day forecast will thus correspond to the first 10 days of the normal weather forecast for Yr. After the first 10 days and up to 21 days, data sets for weekly values and daily values from ECMWF, the meteorological center with headquarters in Reading, are used. The first two days of the Yr forecast are updated every hour, from day three to day 14 the forecast is updated twice a day, and from day 15 to 21 it is updated once per day. At times when the weather forecast is updated with new forecasts, differences may appear for short periods in the new 21-day forecast and the other forecasts on Yr.
There is a lot of probability calculation and ensemble forecasts behind the 21-day data, and up to 100 possible weather scenarios are collected. Various values are then calculated, such as the expected high and low temperature (often corresponding to the maximum temperature of the day and the minimum temperature of the night), and the expected amount of precipitation. In addition, the extremes of possible high and low temperature and the probability of precipitation are calculated, which enables us to show the uncertainty in the forecast. Here we limit the data set to the 10 and 90 percentiles in the ensemble, so that the least likely values are cut out.
What does the weather forecast show?
The table shows the expected highest and lowest temperature for the three 7-day periods in the forecast. The frost parameter indicates how likely it is that the air temperature 2 meters above the ground will be below 0 degrees Celsius during a 7-day period. The chance of frost on the ground can often be somewhat higher than 2 meters up.
Figure text:Tables with expected precipitation, highest and lowest temperature, and chance of frost, divided into 7-day periods, three weeks into the future.
In the temperature graph, a long red or blue bar indicates large fluctuations in temperature throughout the day (i.e. that there is a large difference between the highest and lowest temperature), while a short red bar indicates little diurnal variation. A short shaded field shows little uncertainty in the warning, but a lot of shading indicates greater uncertainty.
Figure text: Graph with expected and possible temperature, given day by day three weeks into the future.
For precipitation, the expected precipitation for the three 7-day periods is shown in a table. The table shows both an expected value and an interval that shows the uncertainty in the forecast. In addition, there is a graph showing the probability of a little (0.5-10 mm) and a lot (more than 10 mm) of precipitation during a day. A high bar therefore means a high probability, and not necessarily that a lot of precipitation is expected.
Figure text: Graph with the chance of little (blue) and much (purple) precipitation, given day by day three weeks into the future.
Can you trust this weather forecast?
Generally speaking, one should have a higher expectation of accuracy for a short-term forecast than a long-term forecast, and the uncertainty typically becomes greater the further into the forecast period one arrives. These long-term forecasts will primarily tell something about the expected development and trend in the weather three weeks into the future. Therefore, both the level of detail and the time resolution in the weather forecast change depending on how far into the forecast period you get. On the next few days in the regular weather forecast, for example, you will be told at what time you can expect precipitation, while in week three instead, it is stated how likely it is that there will be little or a lot of precipitation during a given day. Correspondingly, the time resolution also changes the further into the forecast period you arrive. In the first few days you can see forecast every hour, after a few days you only get every six hours, and in the last half of the period you only get forecasts for one day at a time. In this way, the uncertainty in the forecast is taken into account. For industries that have to adapt operations to weather conditions, such as for food producers or ship traffic in challenging areas, trends in weather development are important information. Other industries that can benefit from weather forecasts for several weeks ahead are, for example, the power and insurance industries.
Although the uncertainty becomes greater beyond the forecast period, verification results nevertheless show that it is possible to say something about the weather three weeks ahead at the level of detail we have chosen to show. How strong a signal we get in week 3 will vary. If there is no strong signal this week, the days in week 3 will be quite similar and correspond to the weather in that location at that time of year. If, on the other hand, the signal is stronger, more variation will be shown between the days in week 3.
The new 21-day forecast has been under development for a while, and has been available in a test version on Beta-Yr. In the test phase, we collected feedback and experiences from you users, and adapted the forecast based on this. Although the 21-day forecast has now been launched, it is still possible to give feedback. Are there any parts of this forecast that were particularly helpful to you, and did you use the forecast to make any decisions? At the bottom of the page with the 21-day forecast, you will find a link with the opportunity to give your feedback.