asd

Description

This Study, began in April 2024, is led by David Ford from the Met Office. Additional contributors to this Study are Pablo Ortega and Joan Llort from BSC. The main CCI ECVs used in this Study are Sea Surface Temperature, Sea Surface Salinity, Sea Ice, Sea Level, and Ocean Colour.

The models EC-Earth4 and GloSea6/MEDUSA are used and comprises three main parts:


Background

Recent progress in Earth System Models (ESMs), in particular the incorporation of biogeochemistry in the ocean models, has enabled the use of ESMs for predicting changes in key biogeochemical variables that act as ecosystem drivers (e.g., pH, oxygen, net primary production, chlorophyll) at seasonal to decadal time scales (Park et al., 2019). Such ESM-based predictions have the potential to be used for predicting variations in fish populations and yields, and provide useful information to aquaculture, fishers and policy makers (Tommasi et al., 2017). Furthermore, they may play an important role in environmental monitoring and protection, for instance in relation to coral reefs and marine protected areas.

Seasonal predictions are commonly initialized from reanalyses that assimilate observations into dynamical forecasting systems (Acosta Navarro et al., 2022). Assimilation of CCI Sea Ice Concentration (WP3.8 in the previous phase of CMUG) demonstrated added value on summer prediction in the Northern Hemisphere (Acosta Navarro et al., 2022). Mean state wind stress correction leads to a modest but significant improvement in predictive skill in ecosystem drivers (Sea Surface Temperature, Chlorophyll, Primary Production). Correcting the full field leads to large predictive skill, demonstrating the dominant role of the wind in ocean biogeochemistry.

This study aims to address the following science questions:


Results and conclusions

Initial results and conclusions are presented here; full results will be published in the scientific literature once the Study is complete.

The following ocean-ice-biogeochemistry runs have been performed with EC-Earth4 at 1° resolution, with an evaluation period of 1998-2022:

Identifier Variables
assimilated 		Data sources
CONTROL 3D
temperature 		EN4
SST_CCI 3D
temperature
Sea
surface temperature 		EN4
SST-CCI
SST_ORAS5 3D
temperature
Sea
surface temperature 		EN4
ECMWF ORAS5 reanalysis
SST_CCI+BGC 3D
temperature
Sea
surface temperature
Phytoplankton
carbon biomass 		EN4
SST-CCI
Derived from OC-CCI
SST_ORAS5+BGC 3D
temperature
Sea
surface temperature
Phytoplankton
carbon biomass 		EN4
ECMWF ORAS5 reanalysis
Derived from OC-CCI

The following ocean-ice-biogeochemistry runs have been performed with GloSea6/MEDUSA at 1/4° resolution, with an evaluation period of 2016-2020:

Identifier Variables
assimilated 		Data sources
FREE - -
STANDARD 3D
temperature and salinity
Sea
surface temperature
Sea
ice concentration
Sea
level anomaly 		EN4
SST-CCI
SI-CCI/OSI SAF
SL-CCI/CMEMS
STANDARD+SSS 3D
temperature and salinity
Sea
surface temperature
Sea
ice concentration
Sea
level anomaly
Sea
surface salinity 		EN4
SST-CCI
SI-CCI/OSI SAF
SL-CCI/CMEMS
SSS-CCI
STANDARD+SSS+OC 3D
temperature and salinity
Sea
surface temperature
Sea
ice concentration
Sea
level anomaly
Sea
surface salinity
Chlorophyll 		EN4
SST-CCI
SI-CCI/OSI SAF
SL-CCI/CMEMS
SSS-CCI
OC-CCI
SSS Sea
surface salinity 		SSS-CCI

Figure 1 shows the difference in annual mean water column-integrated primary production between each EC-Earth4 experiment and CONTROL. All four runs show differences compared with CONTROL, especially in the equatorial Pacific. Both sea surface temperature and phytoplankton carbon biomass have an impact on primary production when assimilated, demonstrating both direct and indirect influences on the biogeochemistry. The impact is more pronounced when sea surface temperature from the ECMWF ORAS5 reanalysis is assimilated than when the SST-CCI product is assimilated.

Figure 1 shows the difference in annual mean water column-integrated primary production between each EC-Earth4 experiment and CONTROL. All four runs show differences compared with CONTROL, especially in the equatorial Pacific.
Figure 1 shows the difference in annual mean water column-integrated primary production between each EC-Earth4 experiment and CONTROL. All four runs show differences compared with CONTROL, especially in the equatorial Pacific.

Figure 2 shows the annual mean water column-integrated primary production from each of these EC-Earth4 experiments, and an observation-based product derived from OC-CCI data. Note that the satellite and model products are broadly equivalent, but not precisely comparable in terms of the quantities they represent. Figure 2 demonstrates that the changes introduced by the assimilation (Figure 1) remain relatively small in comparison with the overall model biases.

Figure 2: Annual mean water column-integrated primary production (mg C m-2 d-1) derived from OC-CCI satellite data (intpp_CCI) and four EC-Earth4 model simulations. Note that the satellite and model products represent broadly but not precisely comparable
Figure 2: Annual mean water column-integrated primary production (mg C m-2 d-1) derived from OC-CCI satellite data (intpp_CCI) and four EC-Earth4 model simulations. Note that the satellite and model products represent broadly but not precisely comparable

It appears that nudging towards surface temperature and phytoplankton carbon biomass is insufficient to significantly alter the complex physical-biogeochemical processes governing primary production. Potential ways to increase the impact of the assimilation in EC-Earth4 include parameter-based ensembles and machine learning techniques. Furthermore, these experiments assimilated monthly-resolution phytoplankton carbon biomass data, as higher temporal resolution products were not available. Preliminary results from another ESA project show greater impact from assimilating higher temporal resolution data, and generating daily or along-track products could be considered in the new ESA PHYTO-CCI project. Phytoplankton carbon biomass data are attractive for data assimilation purposes, as carbon is the base currency of many biogeochemical models.

Figure 3 shows monthly mean water column-integrated primary production from the OC-CCI-derived product and each of the GloSea6/MEDUSA experiments. As with EC-Earth4, the satellite and model products represent broadly but not precisely comparable quantities. The standard assimilation configuration (sea surface temperature, in situ temperature and salinity profiles, sea level anomaly, sea ice concentration) results in elevated primary production compared with the free run, especially around the equator.

Figure 3: Monthly mean water column-integrated primary production (mg C m-2 d-1) for May 2017 derived from OC-CCI satellite data and five GloSea6/MEDUSA model simulations.
Figure 3: Monthly mean water column-integrated primary production (mg C m-2 d-1) for May 2017 derived from OC-CCI satellite data and five GloSea6/MEDUSA model simulations.

This is related to a known issue with excessive vertical mixing triggered by updates to the 3D physical state in this region. Assimilating sea surface salinity data, either on its own or in addition to the other physics variables, has an impact on primary production mostly at mid-latitudes. Adding assimilation of daily-resolution OC-CCI chlorophyll data results in the best correspondence with the OC-CCI-derived primary production product.

Figure 4 shows monthly mean sea surface partial pressure of carbon dioxide (pCO2) for May 2020 from four GloSea6/MEDUSA model simulations. Given the close dependence of the carbonate system on salinity, a specific aim is to investigate the impact of sea surface salinity assimilation on model carbonate outputs. Figure 4 indicates some effects at mid-latitudes, especially when sea surface salinity is assimilated on its own, but the overall impacts are relatively small. This may be due at least in part to the assimilation of sea surface salinity data being restricted to the region between 40°N and 40°S, and it being given relatively high uncertainty estimates (Martin et al., 2022).

Figure 4: Monthly mean sea surface partial pressure of carbon dioxide (pCO2) (µatm) for May 2020 from four GloSea6/MEDUSA model simulations.
Figure 4: Monthly mean sea surface partial pressure of carbon dioxide (pCO2) (µatm) for May 2020 from four GloSea6/MEDUSA model simulations.

Initial results demonstrate distinct impacts from assimilating physics and biogeochemistry ECVs in both EC-Earth4 and GloSea6/MEDUSA, with a mixture of direct and indirect effects on the biogeochemical model state. Nonetheless, neither model is fully constrained. Whilst improvements are mostly likely to be achieved from updates to the model and assimilation methods, the requirement for high frequency observations with reduced uncertainties is highlighted.

References