Icarus Allen, Tom Anderson, Corinne le Quéré Integrated Marine Biogeochemical Modelling Network to Support UK Earth System Research (the 'i-MarNet') i-MarNet aims to provide a framework and pathway for the development of a new biogeochemical model/model hierarchy for use by the UK science community Icarus Allen, Tom Anderson, Corinne le Quéré

Rationale for i-MarNet Ocean Biogeochemistry is an important part of the Earth System regulating the global cycle of C, N, P and providing a source of climatically active gases.

Rationale for i-MarNet The UK is a world leader in marine biogeochemical / ecosystem modelling with a diversity of models and range of expertise. If we are to maintain UK leadership in this field it is essential that the UK ESM community works together in a coordinated way that more effectively exploits the strengths of the Met Office and the NERC community.

Rationale for i-MarNet MEDUSA PlankTOM10 NanoPhyto Si Diatoms Fe N MesoZoo MicroZoo slow detritus Fast particle flux HadOCC nutrient phyto- plankton zoo- detritus Diat-HadOCC ERSEM Increasing structural complexity

Rationale for i-MarNet NERC ESM Strategy: to develop a new, unified, biological modelling approach that builds on the strength of existing models to provide a traceable hierarchy of models of different complexity using a modular approach with consistent process descriptions through a coordinated initiative by the wider UK community

Rationale for i-MarNet A more coordinated strategy for UK model development will help to improve the models, help identify key uncertainties, and ensure compatibility with parallel efforts (e.g. in shelf seas modelling).

Rationale for i-MarNet A more coordinated strategy for UK model development will help to improve the models, help identify key uncertainties, and ensure compatibility with parallel efforts (e.g. in shelf seas modelling). Recognise importance of appropriate approaches for different problems Need to understand the relationship between models of different complexity Use of common frameworks to facilitate pull through from research to core strategic modelling tools

Complexity Phytoplankton Nutrient Zooplankton Detritus

Complexity Si N P Fe multiple zooplankton multiple PFTs multiple slow Phytoplankton Nutrient Zooplankton Detritus multiple nutrients Si N P Fe fast sink slow spectrum of size classes These models can become muddles. Useful heuristic tools.

Increasing structural complexity MEDUSA PlankTOM10 NanoPhyto Si Diatoms Fe N MesoZoo MicroZoo slow detritus Fast particle flux HadOCC nutrient phyto- plankton zoo- detritus Diat-HadOCC ERSEM Increasing structural complexity

v v v v Complexity high light high N:P low Si Poorly understood ecology Umbellosphaera tenuis Florisphaera profunda Gephyrocapsa ericsonii v Aggregation v All in the interactions v Data, validation

Predicted diatom distributions Complexity Predicted diatom distributions March-May Sept.-Nov. NEMO OCCAM Sinha et al. (2010)

Complexity

3 Phase project Phase 1: i-MarNet will establish a community of stakeholders, model developers and model users to help determine the biogeochemical model hierarchy needed by the UK science community. Phase 2: i-MarNet will undertake a comparison of existing UK global biogeochemical models to provide a systematic body of evidence by spring 2013 to inform UKMO model selection for the next IPCC assessment. Phase 3: i-MarNet will define a roadmap for the development of the next generation ocean biogeochemical model in terms of specific areas to be addressed and possible approaches, and requirements for analytical tools to quantify model skill and demonstrate traceability

Phase 1: Establishing the Network Role: Inform the inter-comparison experiment, devise the roadmap Diverse membership Biologists Modellers Bio geochemists Computer scientists Mathematicians Climate scientists Three key workshops Biological assessments Skill assessment Roadmap Model evaluation projects 3 X 15K sub contracts Open call Potential Network Members Core member iMarNet Project partner Written expression of interest Verbal expression of interest

Phase 2: Comparison of existing models HadOCC, diat-HadOCC, MEDUSA, ERSEM, PlankTOM Common Computing Platform Monsoon Common Physics 1 degree Global NEMO Common Forcing CORE (1960-2007) UKMO RCP8.5 (1860-2100) Common Initial Conditions* World Ocean Atlas *where possible

Expert assessment of biological fidelity Independent assessment Phase 2: Assessment Validation Data Computational cost Expert assessment of biological fidelity 3. Model Skill Assessment of C cycle Univariate Metrics (R2, RMSE, Bias etc..) Relationships (e.g. pCO2 vs T) Spatial scales (wavelet analysis) Pdf’s Independent assessment Network evaluation projects Independent report (P Cox) 5. Synthesis Reanalysis Hindcast 1860 1960 2000 2100

Phase 3: Road map for the next generation Goal: to establish the vision for the next generation Ocean biogeochemistry model. Drawing line in the sand but building on on existing efforts we will establish the vision for a 5-10 yr. model development program. Address range of issues: Exploit state of the art knowledge Traceable hierarchy Novel modelling approaches Parameter uncertainty Software frameworks