The initial version of the app was delivered in March 2023. In March 2025, we launched an updated version of the app. The new app has been designed to closely match the schools’ learning goals. Together with teaching instructors from Inspiria Science Center, we made the app more intuitive and matching high school requirements. At the same time, we extended and restructured the processing pipeline to produce more relevant outputs, such as time series, diversity and biomass estimates. Using our domain knowledge, we also contributed to exercises that the school classes can solve by using the app. We will continue improving and developing the app over the next months based on the students’ feedback. While the current version of the app is specifically focused on students, we also envision similar apps for general popular outreach and for research, where data and results from current and previous Oslofjord projects can be visualized as a help in project planning and management.
Methodology
Updated processing pipeline, results and visualization
The school cruise data include manually entered measurements, such as temperature, salinity and species counts from the bottom trawl, which are submitted to Dugnad for Havet. In addition, continuous measurements of meteorological (air temperature, pressure, etc.) and water column data (echosounder backscatter measurements) are conducted throughout the cruise. We have updated the pipeline for the echosounder data collected in the project, and the fusion of the echosounder results with the other two data sources for visualization in the app. The updated processing pipeline fulfills two main goals:
1. Make the pipeline and analysis more robust, consistent between platforms and less sensitive to single, potentially erroneous files, and
2. Improve the relevance of the output.
Based on a published python package that we use for converting the raw proprietary files, we developed our own python library for processing and analyzing the EK80 echosounder data coming from M/S Rognfjell and other platforms in the Frisk Oslofjord project. We apply dedicated post-processing, including spike noise removal through median filters and convolution and remove background noise caused by range- and frequency-dependent signal-to-noise ratios (see example in Figure 1). We furthermore detect the bottom depth to remove all signals at or below the sea floor (Figure 1d).
We apply threshold-based feature detection to identify fish schools in the water column and detect acoustic scattering layers. Finally, we use the multi-frequency backscatter signal, scattering layer information and features for a basic classification of the backscatter into plankton, fishes, aggregations (likely dense plankton patches), unknown/other and background signals (see example in Figure 2). So far, this classification is preliminary, based on domain knowledge and ad-hoc thresholding, but we are working on machine learning and AI -based solutions and aim to extend the classification to include jellyfish.
We then integrate the resulting classified backscatter to obtain estimates of acoustic biomass for, e.g., plankton and fish. The estimates are, together with various other measures and position information (latitude and longitude), saved in .csv result files with 1-minute resolution for use in the Innsikt Oslofjord app. The echosounder results are then combined with the data obtained from Dugnad for Havet.
A major change and improvement in the presentation of the data in the app is the added panel collecting all measurements obtained during school cruises. In addition to a map showing the location of all available stations, a time series enables the students to set their own measurements into perspective and to compare and discuss seasonal and regional gradients and differences. For further insights and an example, see below. We have added the biomass estimates from the echosounder data as informative output. In addition to the new panel, we have made all plots and figures more interactive to allow the students to zoom in and out and download the figures.