5. Climate scenarios for the Norwegian and Barents Seas

We here describe the work on and results from the climate downscaling simulations performed by IMR within the NorScen project. For this work to be understood, we first present the framework within which the development was done. This is related to the development of the Nemo-NAA10km ocean model, in climate downscaling mode. Nemo-NAA10km is an ocean modelling configuration based on the NEMO ocean engine, which development started in 2018 at IMR (Hordoir et al. 2022). Nemo- NAA10km is not a climate model and works in forced mode. Its purpose is to study ocean processes and climate impacts, especially surface processes such as mixed layer depth, which are important for primary production. Nemo-NAA10km has been used in several articles which goal is to estimate the impact of climate related changes on biogeochemical systems (Nilsen et al., 2023) and is being used to understand changes of transport towards the Barents Sea through the Barents Sea Opening (Jahanmard et al. 2024). Nemo-NAA10km is forced through its boundary conditions, either with the atmosphere at the surface of the ocean, or with the ocean at the open boundary conditions of the domain. The latest open boundary conditions are located at the latitude of southern Spain in the Atlantic Ocean, and in the Northern Pacific Ocean (Fig. 24). The coastal boundaries of Nemo-NAA10km are also forced as they provide freshwater in the form of river runoff to the ocean model. When used in hindcast mode, the atmospheric data to Nemo-NAA10km comes from the ERA5 reanalysis of Hersbach et al. (2020), while the open boundary conditions come from the GLORYS ocean reanalysis (http://marine.copernicus.eu). In hindcast simulations we assume the river runoff to be climatological, which means that it has seasonal variations, but no interannual variability. It is extracted from a database based on observations and statistics from Dai and Trenberth (2002).

The goal of our climate downscaling is to run Nemo-NAA10km based on forcing extracted from climate models, instead of re-analysis. When switching to climate downscaling mode, these three forcing components, atmospheric data, ocean boundary conditions and river runoff forcing, need to be adapted from the climate models. Two climate models are presently used to force Nemo-NAA10km in climate downscaling mode, NorESM2 and EC-Earth, to provide a wider range of possible results.

Runoff from rivers, either a few large ones or accumulated from many smaller, is important input to the climate downscaling simulations. Within NorScen we aimed at providing enhanced river runoff fields that are as accurate as possible and that follow the inter-annual variability of the climate model runoff to the downscaling simulations. This is quite challenging and work demanding. An earlier attempt, based on output from the NorESM2 climate model, provided unrealistic river runoff.

River runoff outputs of climate models are two-dimensional fields, which for every coastal point of the ocean component of a climate model provide a runoff value to the ocean. Each point corresponds to a drainage area outlet, but the representation of drainage areas varies between grid and models. Meanwhile, it is impossible to represent river runoff as two-dimensional fields, as they are on one hand discrete distributions, and on the other hand runoff conservation is essential which does not permit any spatial interpolation to enable comparison. A method is needed to perform this comparison.

To perform this task, the runoff of the climate model is re-routed to the Nemo-NAA10km, this means that for a given runoff point of the climate model, the runoff is re-routed towards the closest coastal point of the Nemo-NAA10km grid. River outflow locations can differ from one model to another, but integrating the river runoff along a coastline, should lead to the same accumulated value.  Therefore, our strategy is to integrate along a coastline that spawns the entire coast of the Nemo-NAA10km grid. 

This algorithm, once applied to the runoff fields of the NorESM2 and EC-Earth climate models, allowed for the creation of monthly runoff fields, for each year of the period 1950–2100. These fields follow the inter-annual variability of each climate model original runoff fields, while being bias corrected and consistent with the original runoff inputs of the Nemo-NAA10km hindcast simulations. A somewhat more detailed description of correction algorithms for river runoff and precipitation fields from climate models is given in Appendix 1.

The biogeochemical model used in the present study is the NORWegian ECOlogical Model system end-to-end, NORWECOM.E2E (Aksnes et al. 1995; Skogen et al. 1995; Skogen and Søiland 1998). NORWECOM.E2E is a coupled physical–chemical–biological model system originally used to study primary production, nutrient budgets and dispersion of particles such as fish larvae and pollution. The model system consists of an NPZD (Nutrient-Phytoplankton-Zooplankton-Detritus) model, several individual based models (e.g., Hjøllo et al. 2012), and modules for ocean acidification (Skogen et al. 2014) and contaminants.  In the present study, only the NPZD and the ocean acidification modules have been used. The model has two different types of phytoplankton (diatoms and flagellates) together with two types of zooplankton (micro-and mesozooplankton). The model has previously been validated by comparison with field data in the Nordic region and the Barents Sea (Hjøllo et al. 2012; Sandø et al. 2021; Skaret et al. 2014; Skogen et al. 2007, 2014, 2018). The model has been run in offline mode using 5 days mean values of the physical ocean fields (velocities, salinity, temperature, sea surface height, and sea ice) from the NEMO-NAA10km together with atmospheric fields from NorESM2. The horizontal grid used by the biogeochemical model is a sub-area of the original NEMO-NAA10km grid (Fig. 24) covering the Barents and Nordic Seas.

Figure 24. Domain and bathymetry (in m) of the Nemo-NAA10km configuration.

A new version of the NEMO-NAA10km model was developed for NorScen. This version, including updated downscaling with improved runoff distributions is hereafter referred to as NEMO-NAA10km-new. In the following we show time series and trends from the new NEMO-NAA10km-new runs for open ocean and coastal regions in the Norwegian Sea and Barents Sea. In addition, we have split the open Barents Sea in to an “Atlantic” and “Arctic” compartment as the environment and ecosystems are quite different (Fig. 25). 

Figure 25. The regions in the Norwegian Sea and Barents Sea that we apply our analyses too. Dark blue – Barents Sea Arctic; lighter blue - Barents Sea Atlantic; blue-green – Barents Sea coast; greyish yellow – Norwegian Sea Open; clear yellow – Norwegian Sea Coast.

Figs. 26–28 show updated downscaling results from NEMO-NAA10km-new for temperature, salinity, and mixed layer depth, respectively. The temperature development is quite different between the three scenarios. This is similar, but not identical, to what shown in Sandø et al. (2024), based upon the previous NEMO-NAA10km version. The SSP1-2.6 scenario has weak, negative temperature trends in most areas, except in the Norwegian Sea Open, while SSP2-4.5 and SSP5-8.5 have positive trends in all areas with strongest temperature increase in SP5-8.5. The smoothed time series, which represent variations on a decadal to multidecadal time scale, also indicate that the temperature has strong decadal to multidecadal variability (Fig. 26).

Figure 26. Projected annually averaged ocean temperature (°C) at 50–200 m for the period 2015–2100 in various ocean basins and corresponding coastal regions. From upper to lower: Barents Sea Arctic, Barents Sea Atlantic, Barents Sea Coast, Norwegian Sea Open, and Norwegian Sea Coast. Coloured time series indicate the scenarios SSP1-2.6 (blue), SSP2-4.5 (purple), and SSP5-8.5 (red), and corresponding straight lines their trends.

The salinity trends at 50–200 m are positive for all areas and scenarios and are therefore less mutually different (Fig. 27). Fig. 28 shows that the different temperature and salinity trends also impact the mixed layer depth differently with almost neutral to negative trends in the coastal areas, and mostly strong positive trends in the open ocean areas, at least for SSP5-8.5.

Figure 27. Projected annually averaged ocean salinity (unitless) at 50–200 m for the period 2015–2100 in various ocean basins and corresponding coastal regions. From upper to lower: Barents Sea Arctic, Barents Sea Atlantic, Barents Sea Coast, Norwegian Sea Open, and Norwegian Sea Coast. Colored time series indicate the scenarios SSP1-2.6 (blue), SSP2-4.5 (purple), and SSP5-8.5 (red), and corresponding straight lines their trends.

Figure 28. Projected annually averaged mixed layer depth (m) for the period 2015–2100 in various ocean basins and corresponding coastal regions. From upper to lower: Barents Sea Arctic, Barents Sea Atlantic, Barents Sea Coast, Norwegian Sea Open, and Norwegian Sea Coast. Coloured time series indicate the scenarios SSP1-2.6 (blue), SSP2-4.5 (purple), and SSP5-8.5 (red), and corresponding straight lines their trends.

Projections of net primary production (NPP) and gross secondary production (GSP) with the projected trend-lines are shown in Figs. 29 and 30. In general terms, the NPP in present climate is highest in coastal and ice-free regions with a maximum in the Norwegian Sea Coast box around 150 gC/m²/year and a minimum in the Barents Sea Arctic of approximately 35 gC/m²/year. The projected changes are small and similar between the different emission scenarios, except for the Barents Sea Arctic where there is an increase towards the mid of the century in all three scenarios, but thereafter SSP5-8.5 continue to increase so that the NPP at the end of the century is twice that in the beginning, while for the two other scenarios there is a decline so that SSP1-2.6 is at the same level in 2100 as in 2015, while SSP2-4.5 NPP is about 25% higher.

The level in GSP follows that for NPP with highest values in the coastal and ice-free regions. Except for the Norwegian Sea Coast, there is a general increase in GSP in all areas with the largest change for SSP5-8.5 in the Barents Sea Arctic, and the largest increase for SSP5-8.5 followed by SSP2-4.5 and SSP1-2.6 in general. In the Barents Sea Arctic the patterns for GSP and NPP are similar, with an increase towards the mid of the century before the three emission scenarios show different patterns.

Figure 29. Projected annually averaged NPP for the period 2015–2100 in various ocean basins and corresponding coastal regions. From upper to lower: Barents Sea Arctic, Barents Sea Atlantic, Barents Sea Coast, Norwegian Sea Open, and Norwegian Sea Coast. Coloured time series indicate the scenarios SSP1-2.6 (blue), SSP2-4.5 (purple), and SSP5-8.5 (red), and corresponding straight lines their trends.

Figure 30. Projected annually averaged GSP for the period 2015–2100 in various ocean basins and corresponding coastal regions. From upper to lower: Barents Sea Arctic, Barents Sea Atlantic, Barents Sea Coast, Norwegian Sea Open, and Norwegian Sea Coast. Coloured time series indicate the scenarios SSP1-2.6 (blue), SSP2-4.5 (purple), and SSP5-8.5 (red), and corresponding straight lines their trends.

Compared to the NEMO-NAA10km-old, NEMO-NAA10km-new is colder in the southern part of the Barents Sea, but less saline in the central part of the Barents Sea, providing shallower mixed layer in the central Barents Sea. How these changes will modify the conclusions in Sandø et al. (2024) must be analysed in detail, but the reduced temperature increase in NEMO-NAA10km-new should not aggravate the negative consequences for the stocks that will be negatively impacted by warming.  On the other hand, the positive impacts of less warming on stocks that benefits from increased temperatures might be more limited. In addition to direct temperature effects on plankton production and different fish stocks there are indirect effects of hydrography on the mixed layer depth which in turn impacts plankton production. The plankton production seems to be quite reduced in all areas compared to NEMO-NAA10km-old, but the trends are mostly the same.

The increased temperature will have a negative impact on ice-cover, and together with changes in salinity and wind-stress also an impact on the mixed layer depth. Ice cover and mixed-layer-depth was identified as the main drivers for changes in the Barents Sea primary production in, e.g., Mousing et al. (2023), and consistent patterns in these projections are seen in Figures 26, 27 and 28. The GSP in the Barents Sea partly mirrors the projected changes in NPP, but there are some exceptions. In the Barents Sea Atlantic region the projected increase in GSP is stronger than that in NPP. Mesozooplankton is the main feed for fish larvae, and prey concentration is thereby important for their growth and survival (Folkvord et al. 2009), hence important for recruitment success. The projected increase in GSP in all parts of the Barents Sea is therefore likely to have a potential positive impact for fish recruitment in the area. However, for the total impact on the commercial fish stocks in the area one also needs to take into account that the higher temperatures in a future climate also are expected to impact fish distribution to expand northwards and eastward, potentially also expanding larval distribution (Fossheim et al. 2015). Finally, also the rate of turnover in the phytoplankton community becomes faster with increased temperature. Consequently, the community structure will be more unstable, and, combined with alterations in phytoplankton diversity, this might lead to a loss of ecological resilience in the productivity and functioning of the marine environment (Henson et al. 2021).

The projected NPP and GSP in the Norwegian Sea boxes differs from the Barents Sea projections. In the Norwegian Sea Open there is a strong decrease in both NPP and GSP until 2030. This decrease is probably an artifact that is also seen in the mixed layer depth (Fig. 28) and is probably an initial adjustment due to an unbalance in the initial field. Even with the positive trend projected after 2030, the NPP does not quite return to the initial level in this area for all emission scenarios, while the GSP at the end of the century is larger for SSP2-4.5 and SSP5-8.5 than for present day. An interesting observation is that the lowest NPP is projected at the end of the century for the most extreme scenario (SSP5-8.5), while the opposite is true for GSP. In the Norwegian Sea Coast box, the results are more consistent with a negative trend towards 2100 with the lowest NPP and GSP for SSP5-8.5. The reduction in zooplankton in the coastal box suggests a negative impact for fish recruitment in that area. This applies to, e.g., Norwegian spring-spawning herring, which historically spawns on the coast from Møre and northwards with transport of eggs and larvae along the Norwegian coast and into the Barents Sea. The herring might face more unfavourable conditions in a future climate. A compensating adaptation might be to spawn further north, with lower distance into the Barents Sea with more favourable conditions for the fish larvae, and the last years this spawning is now concentrated from Lofoten and northwards (Salthaug and Stenevik 2024).

Since the aim of this report is to assess the effects of climate change at the end of this century, we have chosen to focus on downscaling several emission scenarios rather than reducing model errors by downscaling several global models. The selection of scenarios to be downscaled and analyzed for physical and biogeochemical variables in this report was made some time before SSP5-8.5 was widely known to be implausible, and before Burgess et al. (2023) was published. Our choice was made to span the range of possible outcomes from the most optimistic to the most pessimistic scenario. Another choice that was made was which global model should be downscaled, and we chose the Norwegian Earth System Model, NorESM2, because it was evaluated well regarding volume and heat transport to the Nordic region and the Barents Sea among 23 CMIP5 and CMIP6 models (Madonna and Sandø 2022). Despite this evaluation, Chylek et al. (2024), comparing different CMIP6 models for the SSP2-4.5 scenario, found that NorESM2-LM underestimates the future Arctic warming compared to the average of seven other models which were all found to be within 15% of the observed warming in the historical period from 1850–2015.  Preliminary results may indicate that temperatures in the SSP5-8.5 scenario from NorESM2 are closer to SSP2-4.5 temperatures in many models when focusing on the Barents Sea, and that this model thus does not necessarily exaggerate the future warming in the SSP5-8.5 scenario with subsequent effects on the marine ecosystem (Sandø et al. 2024). 

The temperature changes projected in our SSP3-7.0 scenario results (not shown) are consistent with the strength of emission scenarios when large basins are considered. This consistency is less pronounced in coastal regions, for which non-linear effects need to be considered. Specifically, the SSP3-7.0 generates what we believe is too high temperatures along the Norwegian coast. A detailed analysis of the underlying physical processes that can explain this unexpected result are beyond the scope of this report, but several hypotheses can be provided. Recent results of Jahanmard et al. (2024) showed for example that the barotropic transport at Barents Sea Opening, and therefore along the Norwegian coast, is most likely a non-linear process related with wind variability more than wind strength. Wind strength does not change significantly over the Nordic Seas at the end of the 21^st century, although the Barents Sea flow does increase, and Jahanmard et al. (2024) could show that a change hidden within the wind signal was indeed responsible. Such a change in wind variability could be for example specific to the SSP3-7.0 scenario.

We can also suppose that a change in stratification at the surface of the ocean, could drive a stronger heat from the surface downward, and explain why sub-surface temperature is more affected in this scenario. This change of stratification could be the consequence of a lower runoff flow along the coast, for example.