EGU General Assembly 2020
© Author(s) 2020. This work is distributed under
the Creative Commons Attribution 4.0 License.

Future changes in atmospheric rivers and extreme precipitation in Norway

Kirien Whan1, Jana Sillmann2, Nathalie Schaller2, and Rein Haarsma1
Kirien Whan et al.
  • 1The Royal Netherlands Meteorological Institute, De Bilt, Netherlands
  • 2Center for International Climate Research (CICERO), Oslo, Norway

Atmospheric rivers (AR) are associated with flooding events in Norway, like the flood that impacted Flåm in 2014. We assess trends in Norwegian AR characteristics, and the influence of AR variability on extreme precipitation in Norway. After evaluating the global climate model, EC-Earth, compared to the ERA-Interim reanalysis, we show that ARs increase in both intensity and frequency by the end of the century. In two regions on the west coast, the majority of winter precipitation maxima are associated with AR events (> 80% of cases). A non-stationary extreme value analysis indicates that the magnitude of extreme precipitation events in these regions is associated with AR intensity. Indeed, the 1-in-20 year extreme event is 17% larger when the AR-intensity is high, compared to when it is low. Finally, we find that the region mean temperature during winter AR events increases in the future. In the future, when the climate is generally warmer, AR days will tend to make landfall when the temperature is above the freezing point. The partitioning of more precipitation as rain, rather than snow, can have severe impacts on flooding and water resource management in Norway.

How to cite: Whan, K., Sillmann, J., Schaller, N., and Haarsma, R.: Future changes in atmospheric rivers and extreme precipitation in Norway, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-18594,, 2020

Comments on the presentation

AC: Author Comment | CC: Community Comment | Report abuse

Presentation version 1 – uploaded on 01 May 2020
  • CC1: Comment on EGU2020-18594, Linda van Garderen, 04 May 2020

    Dear Kirien,

    I tried to ask a question or two during the chat but attendees were muted by the conveners and thus I will try this way. I actually have two questions:

    1) You use PRIMAVERA data for 4 slost of 5 years, this may not be enought to fully understand trends in the data. How do you think this may have influenced your results and are your planning on looking at the full timescales when PRIMAVERA provides it?

    2) I do not see a clear clarification on what is 'intense' for the AR and what is 'extreme' for precipitation. Could you elaborate on that?

    Thank you!

    Linda van Garderen

    • CC2: Reply to CC1, Kirien Whan, 04 May 2020

      Hi Linda,

      Thanks for the questions. 

      1. We use 30 years per period - 6 ensemble members of the 5 years. You can see more information on how the ensemble was generated here:

      I agree that more years would be better. I don't have plans to look at PRIMAVERA data but I think it's a super interesting question. 

      One limitation of the study is clearly that we use only one GCM. We needed a high-resolution GCM to be able to simulate the cyclone features well, and this limited the number of GCMs that were available. This is another good reason to extend the project and look in PRIMAVERA models

      2a. ‘Intense AR’: In the GEV analysis I take the 99th percentile of ARs as an intense AR. 

      2b. ‘Extreme precipitation’: I fit the GEV to the annual maxima of region-mean precipitation. I show how the 1-in-20 year extreme precipitation event increases in magnitude when the AR is intense (at the 99th percentile).

      Please continue the chat or email me at if you have other questions 



  • CC3: Comment on EGU2020-18594, Chris Weijenborg, 04 May 2020

    Hi Kirien,

    Thank you for uploading the presentation! I was wondering what kind of threshold you used in the fixed and variable definitions for the Atmospheric river detection? Is the variable threshold based on a certain percentile of the seasonal IVT?


    • CC4: Reply to CC3, Kirien Whan, 19 May 2020

      Hi Chris,

      We use the 95th percentile of IVT in each period to define the variable threshold. We don’t split by seasons, and so we use the same threshold for summer and winter. You can see the thresholds in Figure 1 of the paper (

      Let me know if you have any more questions (