Calendar

Workshop on Self-Organizing Maps

Time

08. April 2026, 10:30-11:45

Location

BCCR lecture room 4020, Jahnebakken 5

Dear all,

Mark Seefeldt from the University of Colorado - Boulder who is in Bergen for the spring as a visiting Fulbright Scholar kindly invites you to a workshop or short tutorial on Self-Organising Maps. A brief description of which is below, and the key details are:

When: Wednesday, April 8, 12:30-1:45 PM

Where: BCCR - Undervisningsrom 4020

Please send an email to Mark Seefeldt (mark.seefeldt@colorado.edu) if you would like to download the slides prior to the workshop. Or if you have any questions.

Kind regards,

Priscilla

Workshop Description

The method of Self-organizing maps (SOMs) is an artificial neural network data analysis technique that objectively stratifies large volumes of data into a smaller number of recurring patterns on a physically meaningful basis. SOMs have been used for the past 20+ years in the atmospheric and climate sciences identifying patterns in sea-level pressure, winds, relative humidity, and vertical profiles of potential temperature, to name a few applications. This workshop will provide an introduction to the application of SOMs in atmospheric and climate sciences by providing an overview of SOMs, a description of what is done during SOM training, and reviewing two common methods to create SOMs and corresponding analyses. Throughout the presentation there will be a variety examples provided showing the different ways in which SOMs have been applied to atmospheric and climate sciences.

More events

See all

1st BCCR AMOC forum

Dear All, We are pleased to announce the 1st BCCR AMOC forum, that will take place on December 10th 2026 in the Bjerknes meeting room 4020. We plan to provide some light snack or meal for lunch/coffee break during this event, within limits of budget and number of participants. Hence, we kindly ask participants to register by Friday December 4th using the link below, so we can plan for this. Link to registration form and tentative program: https://docs.google.com/forms/d/e/1FAIpQLSdlIqpgCVQNbE1D66BiwY2TLB-HTbn2WvBf8g1zEkhXZhMM1Q/viewform?usp=dialog

BCCR Seminar: Mind the Slope: Why Greenland’s Katabatic Storms Challenge Global Models.

Abstract Katabatic storms in southeastern Greenland are fierce, density-driven downslope wind events that influence both local weather and downstream climate. We assess their representation across three generations of ECMWF global reanalyses (ERA5, ERA-Interim, and ERA40) and a hierarchy of Community Earth System Model version 2 (CESM2) simulations at different horizontal resolutions, using in situ observations and the 2.5-km Copernicus Arctic Regional Reanalysis (CARRA) as benchmarks. The global reanalyses systematically underestimate wind speeds by approximately 30% in ERA5 and 50% in ERA-Interim and ERA40, while failing to capture key storm structures. Similar biases occur in CESM2 simulations at standard 1–2° resolutions. In contrast, variable-resolution CESM2 configurations with local grid refinement to 0.25° and 0.125° better resolve Greenland's topography and substantially improve the representation of storm intensity and structure. Sensitivity experiments further demonstrate that resolving the steep ice-sheet margins is critical for capturing katabatic acceleration. The reanalysis data also reveal that these storms generate fast-moving, high-amplitude ocean wave packets that propagate as far downstream as Scandinavia and the British Isles. Accurately representing these storms in global models and reanalyses is therefore essential for simulating North Atlantic climate variability beyond Greenland's immediate coastal region. Speaker information Marcus Lofverstrom is an associate professor in the Department of Geosciences at the University of Arizona. His research group conducts novel research in the fields of atmospheric dynamics, (paleo-)climatology, and glaciology to understand the evolution of the climate system over multi-millennial time scales. Numerical modeling is an important component of his research program, and the group is utilizing models over a broad range of complexities: from conceptual and linear steady-state models to fully coupled state-of-the-art Earth-system models.