PRISM: Detailed Workpackage descriptions

PRISM WP 1

Work package title General Coordination

Start date or starting event 1

Required input -

Lead contractor for this package 1

Required resources 54 person-months (42 requested from EU)

Contributing partners 1 2

Person-months (inst/EU) 6/42 6/0

Objectives:

Provide general guidance to the project, supervise project director
Organise, chair, and report on project meetings
Organise and chair Project Executive Committee
Organise mid-term user meeting
Provide external representation of the project
Report progress to the European Commission

Description of work
Under the responsibility of the coordinator and the co-coordinator, a project Director will monitor progress by the different groups involved, facilitate communication between the different groups (including the publication of a periodic "newsletter", organise the scheduled meetings of the project as well as small working sessions between groups when necessary. The Director will also regularly communicate with the European Commission in Brussels and help the coordinator and co-coordinator preparing recommendations for future and closer cooperation between climate modelling groups in Europe. Finally, an important task for the coordinators will be to establish links with similar initiatives in the United States and in Japan.

Deliverables
D11 Initial project brochure (month 3)

D12 Document for the first project meeting (month 6) compiling results provided by different work packages

D13 Report of the first project meeting

D14 Report on user consultation meeting (month 18)

D15 Document for the second project meeting (month 24) compiling results provided by different work packages

D16 Report of the second project meeting

D17 Recommendations for a European Climate Computing Facility (month 24)

D18 Measures for consolidation of the developed Earth System Modelling Infrastructure (month 36)

D19 Periodic progress reports to the EC (month 3, 6, . .)

D110 Final report + project brochure (month 36)

Milestones

Month 3: Deliver initial project brochure
Month 6 (m1): First project meeting. Objectives:

Overview of models, platforms, environments, tools
Detailed workplan for development and demonstration runs

Month 18 (m2): User consultation meeting. Objectives:

Review the status of the implementation of the Euroclivar (modelling) recommendations
Identification of interest in external participation in demonstration phase

Month 24 (m3): Second project meeting. Objectives:

Review progress with development and pilots
Approve detailed plan for demonstration phase
Initial recommendations for post-PRISM-up activities

Month 36 (m4): Final report, including recommendations for follow-up activities

PRISM WP 2a

Work package title System Specification

Start date or starting event 1

Required input

Coding rules, quality and portability from WP 2b
Data management requirements from WP 2c

Lead contractor for this package 9

Required resources 26 person-months (13 requested from EU)

Contributing partners 9 18 All other

Person-months (inst/EU) 12/12 1/1 < 1 per partner

Objectives

Set up the architecture of the PRISM system by a survey of existing component model candidates.
Perform a survey of diagnostic and post-processing needs.
Define coupler specifications for WP3a. In particular, define details of the coupling procedure. This includes coupling frequency and the averaging strategy.
Define specifications for a model interface library (to be developed in WP3a).
Specify suitable test procedures for the acceptance of candidate models or model components (in cooperation with workpackages 3a-h).
Specify requirements for other contributors willing to join PRISM at a later stage.

Description of work
This workpackage aims at setting up the scientific aspects of PRISM. Basic documents will be compiled allowing for the choice of candidate models and giving a thorough definition of coupler requirements and an overview of diagnostic needs.

WP2a focuses on the functional requirements of the system design. Coding rules and criteria for the choice of suitable platforms are covered in WP2b. Data management requirements are addressed in WP2c. This structure implies an active interchange of results between these three workpackages, which will be performed in the following tasks:

T1: Perform a survey of candidate models and give initial input to WP 2b and 2c (mo 1 and 2).

T2: Integrate requirements of I/O strategy (from WP 2c), coding and platform requirements (from WP 2b) and coupler and interface specifications into a basic document specifying the criteria for the system and the associated tests (mo 3-6).

T3: Specify criteria for other contributors to PRISM (mo 3-6).

T4: Review system specifications and compile PRISM specification document (containing detailed options for demonstration runs (mo 3-6).

T5: Review demonstration specification and compile PRISM demonstration specification report (mo 23-24).

T6: Review final system status (mo 33-36).

Input will be provided for two PRISM project meetings:

First project meeting (T1, T2, T3 and T4 completed, month 6) to discuss and agree on the specification document.
Second project meeting (T5 completed, month 24) to review the system status before the demonstration runs are started.

If necessary, additional meetings will be organized with the relevant partners.

Deliverables
D2a1 Survey of candidate models (T1 completed, mo 2)

D2a2 Integration of requirements from this workpackage and WP 2b and 2c into a basic document (T2, T3 and T4 completed, mo 6)

D2a3 Initial review of system status (T5 completed, mo 24)

D2a4 Final review of system status (T6 completed, mo 36)

Milestones

Month 6: Compilation of PRISM specification document
Month 24: Compilation of PRISM demonstration run document
Month 36: Compilation of PRISM demonstration run final report

PRISM WP 2b

Work package title Quality Assurance, Coding rules and Portability of Codes and Coupler

Start date or starting event 1

Required input Expertise in coding and optimisation of codes, related quality-assurance evaluation, portability evaluation and performance evaluations, correctness of results and defining ease of use criteria;
Input required from WP 2a, 2b, 2c, 3a - 3h, 5.

Lead contractor for this package 12

Required resources 56 person-month (43 requested from EU)

Contributing partners 12 8 18 19 20

Person months (inst./EU) 6/36 1/1 2/2 2/2 2/2

Objectives

Provide definition/overview of different computer architectures
Provide definition of Coding rules for codes and coupler for the different computer architectures
Provide portability criteria and ensure the compatibility of coding rules for models and coupler
Accelerate the adoption of portability criteria
Provide definition of benchmark criteria and rules
Perform, collect and evaluate benchmarks according to benchmark rules
Provide performing portable models and evaluate their ease of use criteria
Quality assessment for portability, scalability and ease of use of models and coupler

Description of work
The main goal of this workpackage is to assure that the code developments and assembly are meeting certain quality criteria.

First of all related coding and code optimisation rules for all code developments will be set up.

Secondly this workpackage will define the quality criteria for the coding, its optimisation and ease of use.

Of main interest are quality criteria with respect to the scalability, portability and performance of the codes of models that are being developed. Special focus will be on the coupling software and its scalability and portability.

Another criteria for quality will be the ease of use of the assembled system and its components.

As a basis for quality assurance the workpackage will clearly define and distinguish the different computer architectures, like mentioned above.

Assure Portability of all software development between these architectures is a core requirement for the project, since one of the differentiators of the PRISM project to its international competition shall be the evaluation of behaviour of different computer architectures for different tasks to perform within the project.

Therefore a detailed benchmarking is required. The workpackage will define benchmarking rules and performance criteria to be met. A definition for the quality of the results will be worked out as well.

The workpackage will perform benchmarks with codes and coupler and analyse scalability of codes and coupler by these benchmarks to assure high performance for the models on the different computer architectures. These benchmarks will also test the ease of use of the codes and coupler on different hardware platforms and the quality of results of the assembled software system.

Deliverables
D2b1 Technical review of scientific computer architectures (month 4)

D2b2 Definition of coding rules for codes and coupler (month 4)

D2b3 Portability criteria, scalability criteria and ease of use criteria (month 18)

D2b4 Portability, scalability and performance benchmarks, incl. result verification, result reproducibility etc. (month 36)

Milestones

Month 4: Define the coding rules for different computer architectures and review computer architectures
Month 6: Define portability criteria
Month 18: Contributes to user meeting
Month 24: Perform portability and performance benchmarks, verify ease of use
Month 36: Provide a final report (printed and electronic version).

PRISM WP 2c

Work package title Specification of the data management system

Start date or starting event 1

Required input Recommendations from system design

Lead contractor for this package 6

Required resources 26 person-months (13 requested from EU)

Contributing partners 6 4 8 3 18

Person-months (inst/EU) 2/2 6/6 1/1 3/3 1/1

Objectives
The main objective is to define the data management structure for the PRISM system. A system needs to be established which ensures an efficient production of diagnostic output files by the PRISM system and allows for the retrieval of files which are easy to process and analyse. On the production side a variety of hardware environments needs to be supported while on the diagnostic side the comfort of the user is the objective. The design of the system will be based on portable file formats and conventions for the meta-data, which fully describes the data to an external user. Furthermore the system to be designed should render the data easily accessible through the network.

Description of work
The design of the data management will be divided into two distinct but related phases of the model output handling. The first one will deal with the output of diagnostics from the model in such a way that the data can easily and efficiently be archived and handled on the various systems. The second part will concentrate on the processing, analysis and visualization of the data by the user on his workstation. For these two phases different constraints will govern the system design. WP4a and to some extent WP3 will deal with the implementation of the data management system.

T1 - Define the meta-data which needs to be written by the PRISM system together with the diagnostic output. This step will set the information needed from the model and the experimental design and kept in the output files. An analysis of existing model metadata representations will be carried out as part of this task.
T2 - Define the file formats, which need to be supported for model output so that the PRISM system can be used efficiently on the various climate research centres.
T3 - Write specifications for the output library to be used in the PRISM system. This means choosing the operations, which can be performed by the library and the allowed sampling strategies.
T4 - From the meta-data written by the model with the output library a subset needs to be selected for the processing, analysis and exchange of the data.
T5 - Because of the diversity of hardware installed in the climate research centres, different formats are needed for the model output. But for the processing, analysis and exchange of results a common data formats needs to be chosen. This data format must be capable of being read and written by widely used analysis packages, and must be suitable for easy exchange of data between institutions. This format needs also to be supported by the output library.
T6 - Existing diagnostic and processing tools will be surveyed to identify the operations, which are commonly required for data analysis. A template will be developed which describes this set of operations in terms compatible with the common meta-data format.
T7 - Data produced and needed by the various models will reside on different computers. To accelerate this exchange of data between the user and the model an efficient method needs to be implemented. Software allowing such a remote and partial access exists but an implementation is needed which makes best use of the PRISM data format and which takes into account the diversity of hardware.

Deliverables
D2c1: Definition of the meta-data for the model output (month 4)

D2c2: Definition of the file formats supported for output from the PRISM system (month 4)

D2c3: Specification of the output library (month 6)

D2c4: Definition of the meta-data for the data exchange (month 6)

D2c5: Definition of file format for the data exchange (month 6)

D2c6: Catalog of the most important diagnostic operations (month 8)

D2c7 A demonstration of remote data access system will be implemented which allows for the user remote access to model output and for the model to read in ancillary data sets (month 12)

Milestones

Month 4: completion of D2c1 and D2c2
Month 6: completion of D2c3, D2c4 and D2c5
Month 8: completion of D2c6
Month 12: completion of D2c7

PRISM WP 3a

Work package title Development of PRISM System Coupler

Start date or starting event 1

Required input

Coupler and model interface library specifications from WP2a
Coding rules and portability criteria from WP2b
Existing coupler from partner 8

Lead contractor for this package 8

Required resources 78 person-months (39 requested from EU)

Contributing partners 8 18 19 20 3

Person-months (inst/EU) 15/15 18/18 3/3 2/2 1/1

Objectives

Develop a parallel and scalable PRISM System coupler and model interface library according to specifications defined in WP2a, conforming to coding rules and portability criteria defined in WP2b, based on the existing coupler from partner 8.
Document the PRISM System coupler and the model interface library with a written documentation and a WEB presentation.

Description of work
The coupler will drive the whole PRISM system, ensuring the synchronisation of the different units and the exchange of the coupling fields directly between the units or via the coupler. When needed, the interpolation of the coupling fields under constraints typical in geophysical coupling will be performed by the coupler.

The starting point of the work will be the existing coupler developed by partner 8 consisting of a communication and an interpolation library. The functionalities of the coupler will be extended to meet specifications further defined in WP2a, including features regarding scalability and portability. In particular, the coupler itself will be parallelized and will allow a parallel exchange of information with parallel models. The following tasks will be performed:

T1- Develop a generic interface library, to be included in the models, allowing a standard communication with the coupler (or directly with the other units if specified). This first version of model interface will allow a parallel communication between the different processes of a parallel model and the monoprocess coupler. (mo 1 to 12)
T2- Improve the coupler communication library to meet specifications defined in WP2a and standard coding rules defined in WP2b, and to include the possibility of a parallel communication between a parallel model and the coupler. (mo 1 to 12)
T3- Rewrite and improve the coupler interpolation library to meet standard coding rules defined in WP2b and specifications defined in WP2a. (mo 1 to 12)
T4- Parallelize the coupler interpolation library and insure its scalability. (mo 13 to 21)
T5- Test the coupler on one target platform with a prototype configuration of the PRISM System reproducing a realistic exchange of information between highly simplified parallel models. At this point, the communication will take place directly between the different processes of the parallel models and the master process of the coupler; the coupler will be parallel regarding the interpolation operations performed on the coupling fields (mo 22 to mid-23)
T6- Improve the coupler communication library to allow a parallel communication between the different processes of the parallel models and the different processes of the parallel coupler. (mo 24 to 36)
T7- Document the coupler and model interface library with a written documentation and a Web-based presentation. (mo 34 to 36)

Deliverables
D3a1 Portable PRISM coupler and model interface library, allowing a parallel communication between the different processes of a parallel model and the monoprocess coupler, for preliminary interfacing of PRISM components (T1, T2, T3 completed; month 12).

D3a2 Portable, parallel and scalable PRISM coupler and model interface library, allowing a parallel communication between the different processes of a parallel model and the master process of the parallel coupler to be used in demonstration runs (T4 and T5 completed; month 24).

D3a3 Portable, parallel and scalable PRISM coupler and model interface library, allowing a parallel communication between the different processes of the parallel models and the different processes of the parallel coupler (T6 completed; month 36).

D3a4 The PRISM System coupler and model interface library documentation (T7 completed; month 36).

Milestones

Month 12 (T1, T2 and T3 completed) : Completion of D3a1
Month 24 (T4 and T5 completed): Completion of D3a2
Month 36 (T6 completed): Completion of D3a3
Month 36 (T7 completed): PRISM coupler and model interface library documentation (D3a4).

PRISM WP 3b

Work package title Interfacing of Atmosphere with PRISM System

Start date or starting event 1

Required input

System specifications from WP2a
Existing atmosphere models from the partners of WP3b
Model interface library from WP3a.

Lead contractor for this package 7

Required resources 66 person-month (18 requested from EU)

Contributing partners 7 6 5 1 4 13

Person month (inst/EU) 18/18 6/0 6/0 6/0 6/0 6/0

Objectives

To adapt existing atmospheric models used by European climate groups and to implement in these models standard interfaces so that they can interact with other climate components, directly or through a universal coupler.
To ensure that all software conforms to defined standards.

Description of work
Four of the partners have developed global Atmospheric General Circulation Models (AGCM) that are adapted for climate studies. All these AGCM have been coupled to ocean components and some of them to other components of the climate system (ocean, atmospheric chemistry, land/surface biosphere, etc.) but not in a distributed framework which allow interchange of the different models and components. Two of these global models offer the possibility of a variable resolution and can be used for regional climate simulations.

To meet the objectives of this work package, the following tasks will be performed:

T1 Common definition of the interface with participants to WP3c-d-e-f-g following the recommendations of WP2a: partitioning of the calculation between the different AGCMs and the other component models of the climate system, exchanged variables,.etc. (mo 1 to 6).

T2 Adaptation of the AGCMs for the calculation of the exchanged variables. This adaptation will include variable resolution versions of the AGCMs of partners 6 and 7 (mo 7 to 12).

T3 Interface the adapted AGCMs with an early version of the coupler using the interface library developed in WP3a (mo 13 to 18).

T4 Contribution to the analysis of the tests performed with the assembled models in the context of WP3h. (mo 19 to 24).

T5 Development of a complete documentation including scientific description and technical aspects of the AGCMs (mo 19 to 24).

Deliverables
D3b1 Component atmospheric models suitable for coupling to PRISM system (month 18)

D3b2 Documentation including scientific description and technical aspects of the models (month 24)

Milestones

Month 6 (T1 completed): Review existing codes and decide on best interface strategy.
Month 12 (T2 completed): Review model adaptation
Month 18 (T3 completed): Delivery of the new adapted atmospheric components to partners of WP3h for assembling in the PRISM system.
Month 24 (T4 and T5 completed): Delivery of the new adapted atmospheric components tested with the PRISM system for the demonstration runs.

PRISM WP 3c

Work package title Interfacing of Atmospheric Chemistry with PRISM System

Start date or starting event 1

Required input

System specifications from WP2a
Existing atmosphere chemistry models from the partners of WP3c
Model interface library from WP3a.

Lead contractor for this package 1

Required resources 59 person-months (21requested from EU)

Contributing partners 1 5 6 4 7 2

Person-months (inst/EU) 18/18 6/0 6/0 6/0 2/0 3/3

Objectives

To adapt existing atmospheric chemistry models used by European climate groups and to implement in these models standard interfaces so that they can interact with other climate components, directly or through a universal coupler.
To ensure that all software conforms to defined standards.

Description of work

T1: Chemistry climate models developed by different partners will be analysed and modified to be interfaced with the PRISM system. Different types of interface will be considered "(either direct inclusion of the chemical equations within the atmospheric model, exchange of information directly with the atmospheric model by the model interface library developed in WP3a, or indirect transfer of information through the PRISM coupler). Discussions between partners will take place to decide on the optimal complexity of the chemical scheme needed to address climate issues. The different numerical schemes used to solve the stiff chemical system will be compared on the basis of accuracy, mass conservation and computational efficiency. The solutions of the models provided by partners will be compared to a pure zero-dimensional chemical run using the Gear technique (used as a reference).

T2: Review interface with atmospheric models

T3: From these tests, the optimum chemical code will be chosen and made suitable for interface with PRISM

T4: The chemical model will be interfaced with the PRISM system.

Deliverables
D3c1: Provide report on model comparison and requirements for interface with PRISM (T1 and T2 completed; month 12)

D3c2: Component atmospheric chemistry models suitable for coupling to PRISM system including documentation (T3 completed; month 18)

D3c3: Atmospheric chemistry model interfaced with PRISM and tested (T4 completed; month 24)

Milestones

Month 6: (T1 completed) Review of existing codes and decision on best interface strategy
Month 12: (T2 completed) Review of model adaptation
Month 18: (T3 completed) Component model with early versions of the coupler interfaced
Month 24: (T4 completed) Component model interfaced with the PRISM system. Completion of tests

PRISM WP 3d

Work package title Interfacing of Land-surface schemes with PRISM

Start date or starting event 1

Required input

System specifications from WP2a
Existing land-surface schemes from the partners of WP3c
Model interface library from WP3a.

Lead contractor for this package 6

Required resources 60 person-month (22 requested from EU)

Contributing partners 6 4 1 5 7

Person-months (inst/EU) 18/18 6/0 6/0 6/0 2/4

Objectives

To adapt existing land-surface schemes used by European climate groups and to implement in these models standard interfaces so that they can interact with other climate components, directly or through a universal coupler.
To ensure that all software conforms to defined standards.

Description of work
The coupling between the land-surface and atmosphere is probably one of the most complex in the climate system because of the wide range to time scales involved. Up to now land-surface schemes were deeply interwoven in European atmospheric models. The proposed interface will change this situation by allowing for a clear distinction between surface and atmospheric processes and thus the integration of land-surface scheme into the PRISM system.

T1 - To evaluate the degree to which current land-surface schemes are interwoven in European atmospheric models. This will allow all partners to define the tasks needed to implement in their model the interface proposed by PILPS-4 (mo 1 to 6).

T2 - Implementation of the interface by the partners (mo 6-12).

T3 - To review the evolution of the implementation of the interface an inter-comparison of the land-surface schemes will be organised. All schemes will be run coupled to a simplified atmosphere. This will test the modularity of the schemes and their ability to adapt to new host models (mo 12-18).

T4 - Once a sufficient number of atmospheric models will have integrated the interface an evaluation of this part of the implementation will be performed. Pairwise the land-surface schemes will be exchanged to perform parallel sensitivity experiments. This will test the plug-compatibility of the atmospheric and surface component and the consistency of the new coupling (mo 12-24).

T5 - At this stage the interface will have gained sufficient maturity to finalise the documentation and integrate it in the PRISM system (mo 18-24).

Deliverables
D3d1 Land-surface schemes suitable for coupling to PRISM system including documentation (T3 completed, month 18)

D3d2 Documentation including scientific description and technical aspects of the models (T5 completed, month 24)

Milestones

Month 6: Review existing codes and evaluate the degree of re-coding needed by each group (T1 completed).
Month 12: Review the modularity of land-surface schemes (T2 completed and T3 underway).
Month 18: Interface land-surface schemes model with early versions of the coupler (T3 completed).
Month 24: Completion of tests (T4 and T5 completed).

PRISM WP 3e

Work package title Interfacing of Ocean with PRISM

Start date or starting event 1

Required input

System specifications from WP2a
Existing ocean models from the partners of WP3e
Model interface library from WP3a

Lead contractor for this package 5

Required resources 50 person-month (18 requested from EU)

Contributing partners 5 4 1 6

Person months (inst/EU) 18/18 6/0 6/0 2/0

Objectives

To adapt existing ocean models used by European climate groups and to implement in these models standard interfaces so that they can interact with other climate components, directly or through a universal coupler.
To ensure that all software conforms to defined standards.

Description of work
A number of existing ocean models will be evaluated in terms of the parameters that will need to be exchanged with the atmosphere and sea-ice bio-geochemistry or regional modules. In particular, a strategy for handling non-uniform data will be developed to enable the interfacing of dynamic sea-ice models with ocean models. Specifically the handling of variable sea ice margins and fractional sea ice coverage will need to be considered.

The ocean models will be modified to be interfaced with the PRISM system, paying particular attention to the conservation properties of the system with regard to energy, mass, salinity.

Test runs of the adapted ocean models will be performed with both one-way and two-way coupling to check the sensitivity of the results to coupling frequency, averaging procedures. In some runs, real cases (e.g. forcing the ocean models with atmospheric reanalyses) will be used so that the performance can be evaluated against observations. The performance of the ocean models in the fully coupled, final system will also be evaluated.

The adapted ocean models will be documented in terms of the parameters and methodology required for coupling, as well as the conservation properties.

Deliverables
D3e Three ocean models suitable for coupling to PRISM system including documentation (month 18)

Milestones

Month 6: Review existing codes and evaluate the degree of recoding needed by each group.
Month 12: Review the modularity of ocean models
Month 18: Ocean models with early versions of the coupler
Month 24: Complete tests

PRISM WP 3f

Work package title Interfacing of Sea-ice with PRISM System

Start date or starting event 1

Required input

System specifications from WP2a
Existing sea-ice models from the partners of WP3f
Model interface library from WP3a.

Lead contractor for this package 11

Required resources 66 person-month (28 requested from EU)

Contributing partners 11 17 4 6 1

Person-months (inst/EU) 18/18 6/6 6/0 2/4 6/0

Objectives

To adapt existing sea-ice models used by European climate groups and to implement in these models standard interfaces so that they can interact with other climate components, directly or through a universal coupler.
To ensure that all software conforms to defined standards.

Description of work
A dynamic and thermodynamic sea ice module is an integrated part of any climate modelling system. In this workpackage, existing sea ice models used by European climate groups will be reviewed. Based on the review, some of the most robust and best validated models will be adapted to the software requirements defined for the PRISM System. In the adaptation phase, emphasis will also be put on features of special importance for the atmosphere-sea ice-ocean interface, including proper treatment of partial ice and snow covered grid cells, proper treatment of the ocean-sea ice albedo, and strict conservation of heat, fresh water and salt. The adapted sea ice model component will be interfaced with the PRISM coupler, and tested. At the final stage of the project, the PRISM system runs will be supported, and results analysed based on available sea ice and snow observations, and comparable model integrations.

Deliverables
D3f Component models suitable for coupling to PRISM system including documentation (18 months)

Milestones

Month 6: Review existing codes and decide on best interface strategy
Month 12: Review model adaptation
Month 18: Interface component model with early versions of the coupler
Month 24: Contribute to system tests (WP 3i)

PRISM WP 3g

Work package title Interfacing of Ocean Bio-geochemistry with PRISM

Start date or starting event 1

Required input

System specifications from WP2a
Existing ocean bio-geochemistry models from the partners of WP3g
Model interface library from WP3a.

Lead contractor for this package 14

Required resources 57 person-month (30 requested from EU)

Contributing partners 14 6 1 4

Person-months (inst/EU) 18/18 3/3 3/6 3/3

Objectives

To adapt existing ocean bio-geochemistry models used by European climate groups and to implement in these models standard interfaces so that they can interact with other climate components, directly or through a universal coupler.
To ensure that all software conforms to defined standards.

Description of work
The coupling between bio-geochemistry and ocean models requires an efficient parameterisation of transport processes in the ocean which takes into account the particularities of chemical and biological sources and sinks. Up to now, bio-geochemical schemes were interwoven in ocean models with on the one hand no clear distinction between bio-geochemical schemes and ocean physics, and on the other hand, no clear use of the similarities between tracers, either active as used in ocean models (temperature and salinity) or passive as used in bio-geochemical schemes (carbon, nitrogen, etc.). The proposed interface will change this situation by clarifying the similarities and specificity of bio-geochemical schemes and integrating these in a way which will allow parallel but concerted developments in the modelling of ocean physics and bio-geochemistry.

T1 - To evaluate the degree to which current bio-geochemical schemes are interwoven in European ocean models, define a common interface to ocean models and to determine the modifications needed in ocean model to implement such an interface. (mo 1-6).

T2 - Implementation of the interface in the bio-geochemical schemes by the partners (mo 2-12).

T3 - Implementation of the interface for the transport of passive tracers in ocean models (mo 2-12)

T4 - Implementation of options to accelerate the integration time of passive tracers by decoupling the time steps used by the bio-geochemical schemes and by the ocean models, and/or decreasing the number of boxes over which the computation is to be performed (mo 7-18).

T5 - Review the evolution of the implementation of the interface with ocean models and test the modularity of the schemes and their ability to adapt to new host models (mo 12-23).

T6 - Finalise the documentation and integrate it in the PRISM coupler (mo 24).

Deliverables
D3g1 Bio-geochemical schemes suitable for coupling to PRISM system (T4 completed, month 18)

D3g2 Documentation including scientific description and technical aspects of the models (T6 completed, month 24)

Milestones

Month 6: Review of existing codes and evaluation of the degree of re-coding needed by each group (T1 completed).
Month 12: Interface of bio-geochemical schemes with ocean models (T2 and T3 completed).
Month 18: Interface of bio-geochemical schemes including options of acceleration (T4 completed).
Month 24: Completion of tests and documentation (T5 and T6 completed)

PRISM WP 3h

Work package title Interfacing regional models with PRISM System

Start date or starting event 1

Required input

System specifications from WP2a
Existing regional models from the partners of WP3h
Model interface library from WP3a.

Lead contractor for this package 10

Required resources 46 person-month (23 requested from EU)

Contributing partners 10 9 4 18

Person-months (inst/EU) 8/8 7/7 7/7 1/1

Objectives

To interface several existing European regional climate models with the PRISM system global models, by means of standardising and streamlining global model output/regional model input conventions
Explore the use of the PRISM coupler to more dynamically link regional and global model components
To ensure that all software conform to defined standards.
Participate in the PRISM demonstration in WP5 by extending the global simulations into dynamically downscaled regional simulations; provide these to potential users.

Description of work
Regional climate models are feasible tools to achieve high resolution regional scenarios requested by users performing impact analyses. These can in the end be used to set political negotiation targets and strategies as well as adaptive strategies on those societal aspects where the time horizon is long, such as investments in major infrastructure. 3 of the partners have regional climate models (RCMs) for dynamical downscaling of GCM-scenarios for the European region. Regional ocean/lake/hydrology components are also available. The first task is:

T1 - review present practices of GCM-RCM coupling and decide on the PRISM interface strategy. (mo 1-6)

To run regional climate models requires GCM data with high frequency output. The quality of RCM-simulations also strongly depends on the quality of the GCMs providing the large-scale data. So, there is a close coupling between the use of global and regional models. In PRISM, the flow of information from GCMs to RCMs is improved. To reduce the uncertainties in the GCM-RCM-scenario chain, it is important to facilitate the cross-use of many GCMs and many RCMs. For this, development of GCM-RCM coupling routines and co-ordination of GCM and RCM simulations, starting from the design of GCM output is needed. This is done in two tasks:

T2 - standardise, within the PRISM system, GCM output and the participating RCM input for offline use. (mo 6-18)

T3 - investigate the use of the PRISM system coupler to permit offline or online GCM-RCM coupling. (mo 12-24)

Finally, the regional models, with the improved interface to the global models will participate in the PRISM demonstration experiments in WP5.

Deliverables
D3h1 Component regional models suited for offline coupling in the PRISM system and documentation (T1, T2 completed; mo 18)

D3h2 Report on the use of the PRISM coupler to interface RCMs and GCMs (T3 completed; mo 24)

Milestones

Month 6: (T1 completed): Review of suitable RCMs and formulated RCM-GCM interface strategy
Month 18 (T2 completed): Off-line standardised GCM-RCM coupling.
Month 24 (T3 completed): Interface regional models with the PRISM system coupler.

PRISM WP 3i

Work package title Assembling of the PRISM System

Start date or starting event 6

Required input

Coupler and model interface library specifications from WP2a
Coding rules from WP2b
Coupler and models from WP3a to WP3g

Lead contractor for this package 3

Required resources 62 person-month (31 requested from EU)

Contributing partners 3 18 19 20 8

Person-months (inst/EU) 18/18 4/4 4/4 4/4 1/1

Objectives:

To combine components of the climate system as developed in WPs 3a-3h into the comprehensive PRISM system
To run this PRISM system for selected combinations of model components
To implement this PRISM system on different platforms

Description of work
The PRISM System will be composed of models simulating the diverse components of the Earth climate system interacting in a standard way with one or more other components.

In this particular work package the various climate system component models developed in the other work packages (WP3a-WP3g) will be combined to form complete climate system models. These complete climate system models will be run on a number of supercomputers. The vendors will help with their know how to install the codes or their particular machines.

However, meta-computing will not be attempted, since it is considered as too costly in human resources at present. This means that each coupled model assembled through the web interface will run entirely on one machine.

This package delivers the basis for the web based system developed in WP 4b.

The system will be tested in stages. The reproducibility of results on different system configurations for parallel runs will be verified. The definition of coding rules and the mechanism of software quality assurance defined in WP2b will be applied to all the software. The final product will be documented. The starting point will be the coupler developed in WP3a, one atmosphere model developed in WP3b and an ocean model developed in WP3e. The model components developed in the other WP3 c, d, e, f, g will be added gradually. The system will be tested on several platforms. Following tasks are envisaged:

T1 - Couple a system consisting of one atmosphere and one ocean model, run it on a single platform (mo 6 to 12)

T2 - Couple a system consisting of a different atmosphere model than in T1 and the same ocean model as in T1, run it on a single platform (mo 13 to 18)

T3 - Transfer the system developed in T1 and T2 to different platforms (mo 19 to 24)

T4 - Couple the system developed in T1 and T2 with the other components of the climate system, run it on a single platform (mo 25 to 30)

T5 - Transfer the systems developed in T4 to different platforms (mo 31 to 36)

The PRISM System will be composed of models simulating the diverse components of the Earth climate system interacting in a standard way with one or more other components.

In this particular work package the various climate system component models developed in the other work packages (WP3a-WP3g) will be combined to form complete climate system models. These complete climate system models will be run on a number of supercomputers. However, meta-computing will not be attempted, since it is considered as too costly in human resources at present. This means that each coupled model assembled through the web interface will run entirely on one machine.

This package delivers the basis for the web based system developed in WP 4b.

The system will be tested in stages. The reproducibility of results on different system configurations for parallel runs will be verified. The definition of coding rules and the mechanism of software quality assurance defined in WP2b will be applied to all the software. The final product will be documented.

Deliverables
D3i A system of different climate model components able to run in different combinations as climate system model on different computer platforms

Milestones

Month 18 (T1 and T2 completed): A coupled model consisting of two components running on one platform
Month 24 (T3 completed): A coupled model consisting of two components running on several platforms
Month 30 (T4 completed): A coupled model consisting on various components running on one platform
Month 36 (T5 completed): A coupled model consisting on various components running on several platforms

PRISM WP 4a

Work package title Development of Data Generation, Diagnostic and Visualization Tools

Start date or starting event 1

Required input Definition of the common PRISM metadata and data formats. Definition of the interface to the output library. Description of operations required for the processing library.

Lead contractor for this package 4

Required resources 180 person-month (90 requested from EU)

Contributing partners 4 6 3 2 19 18 20 21

Person-months (inst/EU) 18/18 12/12 12/12 21/21 9/9 2/2 10/10 6/6

Objectives
To provide a library that can generate output from PRISM models in the supported PRISM data formats. To provide a flexible library of tools to facilitate processing and analysis of data in the common PRISM data format and to promote sharing of data and analysis programs. To provide high-end and low-end graphical interfaces to display the data.

Description of work
An output library meeting the requirements of the defined interface will be written and tested to evaluate its ability to write data from the model into files with the chosen data formats and the common metadata convention. The library will then be optimised for the various architectures on which the PRISM system will run. This task includes machine independent vectorization and parallelisation.

Data produced and needed by the various models will reside on different computers. To accelerate this exchange of data between the user and the model an efficient method needs to be implemented. Software allowing such a remote and partial access exists but an implementation is needed which makes best use of the common PRISM data format and which takes into account the diversity of hardware. In order to do this, utilities will be needed for conversion between the common data format and any of the local data formats, which the climate centres decide will continue to be used in the future. Such utilities will be the responsibility of those institutions to provide.

The template for the processing library will be implemented as a software library. The language used will be chosen according to the requirements that it can pipe its output into graphics packages, it can manipulate data and metadata together (e.g. using derived data types or objects), it can be compiled and run on any machine and preferably that it is widely used and non-proprietary. If possible, existing processing software will be built upon in producing the library. The library will be able to read and write data in the common data format. An analysis program will be written with this library to demonstrate its usefulness for data from any PRISM model in the common format.

Existing graphic packages will be surveyed to identify the operations, which are required to use them for the analysis programs developed in PRISM. A low-end graphical display of PRISM data will be demonstrated and a high-end graphical display system will be developed together with the vendors.

Deliverables
D4a1 A first version of the output library which can be used in models, thus validating the user interface

D4a2 An performance benchmark for the output library representative of PRISM

D4a3 An optimised version of the output library which takes into account the performance enhancement possible on the various hardware used by the PRISM system

D4a4 A remote data access system will be implemented which allows for the user remote access to model output and for the model to read in ancillary data sets.

D4a5 Review of existing graphics packages and definition of common requirements for graphics for ordinary analysis purposes (low-end) and high-quality presentational purposes (high-end).

D4a6 A developed processing library that meets the architectural requirements, delivers the functionality defined in D4a5 and agrees with the data formats of D4a2.

D4a7 Identification or provision of graphics packages which meet the low-end and high-end graphics requirements.

D4a8 A demonstration programme including high-end and low-end graphical display using the defined data formats.

D4a9 A final report providing advice to members on development of existing systems to allow continued integration and proposing future development areas.

Milestones

Month 12: Completion of D4a1, D4a2, D4a5
Month 24: Completion of D4a3, D4a4
Month 30: Completion of D4a6 and D4a7
Month 36: Completion of D4a8 and D4a9

PRISM WP 4b

Work package title System Installation, Communication and Networking

Start date or starting event 1

Required input

Existing model presentation systems from partner 16
The component models as produced by WP3a to WP3g
The assembled PRISM system as set up by WP3i

Lead contractor for this package 3

Required resources 101 person-month (61 requested from EU)

Contributing partners 3 16 15 8

Person-months (inst/EU) 18/18 18/36 3/6 1/1

Objectives

to generate a web-based system which is capable:
to assemble desired model configurations,
to submit these models on different computer platforms,
to monitor the progress of model integration,
to control the database archiving of the model data,
to link the model experiment with the diagnostic package (WP4a).

Description of work
The PRISM System will be composed of models simulating the diverse components of the Earth climate system interacting in a standard way with one or more other components.

The PRISM system will link the different model components, take care that the appropriate input data are provided and submit the model to a selected computing site. It will supervise the experiment during its execution, will supervise the archiving of the generated data and will eventually start off a diagnostic programme.

The PRISM system will be based on an existing system used by partner 16 to submit and to run a single model on a single computing site. The system consists of a Java based web browser interface and a supervisor programme, which controls and monitors the models simulations. Partner 15 will help with the definition of the functionality of the system.

It is planned to extend the existing system to work with different models, to be able to assemble a model from different components, to work in different computing environments, and to supervise the archiving of the results.

The PRISM system will be tested in stages. The mechanism of software quality assurance defined in WP2b will be applied to all the software. The final product will be documented.

Following tasks will be performed:

T1 The portability and adaptability of the existing system from partner 16 to the needs of the climate modelling community will be assessed, and the integrability of similar systems of the other project partners into a common system will be analysed. (mo 1 to 3)

T2 The web-surface and functionality of the system will be defined and added to the PRISM Specification Document (mo 4 to 6)

T3 A prototype of a web-based system will be set up to run different models on a single site (partner 16) (mo 7 to 18)

T4 The web-based system has to implemented at different sites and tested with various models (mo 19 to 27)

T5 The system has to be linked with the Diagnosis system developed in WP4a (mo 28 to 36)

Deliverables
D4b A web-based system will be delivered, which will be capable of assembling and running climate system models, monitoring the progress of numerical experiments with this model, supervising the archiving the data generated by these experiments and starting off the diagnostic tools developed in WP4a.

Milestones

Month 6 (T1 and T2 completed):

Well defined functionalities of the web-based system as well as a plan how to adapt the existing system to these functionalities

Month 18 (T3 completed):

A prototype of the web-based system runs for different models on a single site (partner 16)

Month 27 (T4 completed):

The web-based system is implemented at different sites and tested with various model configurations

Month 36 (T5 completed):

The system is linked with the Diagnosis system developed in WP4a

PRISM WP 5

Work package title Demonstration experiments with PRISM

Start date or starting event 6

Required input Results from wp3i, wp4a, wp4b

Lead contractor for this package 13

Required resources 172 person-month (86 requested from EU)

Contributing partners 13 9 5 6 7 3 12 10 4 8 2 18

Person months (inst/EU) 30/30 6/6 6/6 3/3 3/3 3/3 3/3 6/6 3/3 1/1 18/18 4/4

Objectives

Demonstrate the feasibility and the efficiency of the PRISM system with a number of different sample integrations with various combination of the PRISM system on different platforms.
Analyse results to show the correctness of the approach.

Description of work
Partner 13 will co-ordinate the design of the demonstration experiments within PRISM and their analysis. The runs will be performed on different computer platforms in Europe. Functional and technical tests of each model component will be the responsibility of the originating groups (wp3a - wp3h). The demonstration runs will provide information on the overall performance of the system for different model configurations and for different computer architectures. Examples of possible runs include:

a. Atm/ocean in 3 configurations

b. One configuration with bio-geochemistry

c. One run atm/ocean + regional

d. A short integration with a very high resolution model a few seasons long

e. A multi-model ensemble integration (at T63, or thereabouts) for a couple of seasons, including the 1997/98 ENSO year

At the first project meeting different options for the demonstration runs will be considered and discussed. At the second project meeting the final selection will be made.

The demonstration runs will take place on the supercomputers already available to PRISM partners. They will not require massive amount of computing resources. Given the high modularity and flexibility of PRISM it will not be possible to demonstrate all possible combinations of modules, but a few will be selected for this task, with one coupler being at the centre of all combinations:

The following tasks are envisaged:

T1- Definition of the demonstration runs (mo 6-18)

T2- Preparation of demonstration experiments (mo 19-24)

T3- Execution of demonstration experiments (mo 25-32)

T4- Analysis of the results and final wrap-up (mo 33-36)

Deliverables
D51 Contribution to the PRISM System Specification Report (month 6).

D52 Contribution to the PRISM Demonstration Specification Report (month 24).

D53 Results from demonstration runs (month 36).

Milestones

Month 6 Contribution to PRISM Specification Document (more detailed options for demonstration runs)
Month 24 Final selection of PRISM Demonstration runs
Month 36 (T3 completed): Completion of demonstration runs and analysis of results
Month 36 (T4 completed): Final report

Back to Part B

To Part C

Last update: 27.09.2000 by Andreas Villwock

Work package title	System Specification
Start date or starting event	1
Required input	Coding rules, quality and portability from WP 2b Data management requirements from WP 2c
Lead contractor for this package	9
Required resources	26 person-months (13 requested from EU)
Contributing partners	9	18	All other
Person-months (inst/EU)	12/12	1/1	< 1 per partner
Objectives Set up the architecture of the PRISM system by a survey of existing component model candidates. Perform a survey of diagnostic and post-processing needs. Define coupler specifications for WP3a. In particular, define details of the coupling procedure. This includes coupling frequency and the averaging strategy. Define specifications for a model interface library (to be developed in WP3a). Specify suitable test procedures for the acceptance of candidate models or model components (in cooperation with workpackages 3a-h). Specify requirements for other contributors willing to join PRISM at a later stage.
Description of work This workpackage aims at setting up the scientific aspects of PRISM. Basic documents will be compiled allowing for the choice of candidate models and giving a thorough definition of coupler requirements and an overview of diagnostic needs. WP2a focuses on the functional requirements of the system design. Coding rules and criteria for the choice of suitable platforms are covered in WP2b. Data management requirements are addressed in WP2c. This structure implies an active interchange of results between these three workpackages, which will be performed in the following tasks: T1: Perform a survey of candidate models and give initial input to WP 2b and 2c (mo 1 and 2). T2: Integrate requirements of I/O strategy (from WP 2c), coding and platform requirements (from WP 2b) and coupler and interface specifications into a basic document specifying the criteria for the system and the associated tests (mo 3-6). T3: Specify criteria for other contributors to PRISM (mo 3-6). T4: Review system specifications and compile PRISM specification document (containing detailed options for demonstration runs (mo 3-6). T5: Review demonstration specification and compile PRISM demonstration specification report (mo 23-24). T6: Review final system status (mo 33-36). Input will be provided for two PRISM project meetings: First project meeting (T1, T2, T3 and T4 completed, month 6) to discuss and agree on the specification document. Second project meeting (T5 completed, month 24) to review the system status before the demonstration runs are started. If necessary, additional meetings will be organized with the relevant partners.
Deliverables D2a1 Survey of candidate models (T1 completed, mo 2) D2a2 Integration of requirements from this workpackage and WP 2b and 2c into a basic document (T2, T3 and T4 completed, mo 6) D2a3 Initial review of system status (T5 completed, mo 24) D2a4 Final review of system status (T6 completed, mo 36)
Milestones Month 6: Compilation of PRISM specification document Month 24: Compilation of PRISM demonstration run document Month 36: Compilation of PRISM demonstration run final report

Work package title	Quality Assurance, Coding rules and Portability of Codes and Coupler
Start date or starting event	1
Required input	Expertise in coding and optimisation of codes, related quality-assurance evaluation, portability evaluation and performance evaluations, correctness of results and defining ease of use criteria; Input required from WP 2a, 2b, 2c, 3a - 3h, 5.
Lead contractor for this package	12
Required resources	56 person-month (43 requested from EU)
Contributing partners	12	8	18	19	20
Person months (inst./EU)	6/36	1/1	2/2	2/2	2/2
Objectives Provide definition/overview of different computer architectures Provide definition of Coding rules for codes and coupler for the different computer architectures Provide portability criteria and ensure the compatibility of coding rules for models and coupler Accelerate the adoption of portability criteria Provide definition of benchmark criteria and rules Perform, collect and evaluate benchmarks according to benchmark rules Provide performing portable models and evaluate their ease of use criteria Quality assessment for portability, scalability and ease of use of models and coupler
Description of work The main goal of this workpackage is to assure that the code developments and assembly are meeting certain quality criteria. First of all related coding and code optimisation rules for all code developments will be set up. Secondly this workpackage will define the quality criteria for the coding, its optimisation and ease of use. Of main interest are quality criteria with respect to the scalability, portability and performance of the codes of models that are being developed. Special focus will be on the coupling software and its scalability and portability. Another criteria for quality will be the ease of use of the assembled system and its components. As a basis for quality assurance the workpackage will clearly define and distinguish the different computer architectures, like mentioned above. Assure Portability of all software development between these architectures is a core requirement for the project, since one of the differentiators of the PRISM project to its international competition shall be the evaluation of behaviour of different computer architectures for different tasks to perform within the project. Therefore a detailed benchmarking is required. The workpackage will define benchmarking rules and performance criteria to be met. A definition for the quality of the results will be worked out as well. The workpackage will perform benchmarks with codes and coupler and analyse scalability of codes and coupler by these benchmarks to assure high performance for the models on the different computer architectures. These benchmarks will also test the ease of use of the codes and coupler on different hardware platforms and the quality of results of the assembled software system.
Deliverables D2b1 Technical review of scientific computer architectures (month 4) D2b2 Definition of coding rules for codes and coupler (month 4) D2b3 Portability criteria, scalability criteria and ease of use criteria (month 18) D2b4 Portability, scalability and performance benchmarks, incl. result verification, result reproducibility etc. (month 36)
Milestones Month 4: Define the coding rules for different computer architectures and review computer architectures Month 6: Define portability criteria Month 18: Contributes to user meeting Month 24: Perform portability and performance benchmarks, verify ease of use Month 36: Provide a final report (printed and electronic version).

Work package title	Specification of the data management system
Start date or starting event	1
Required input	Recommendations from system design
Lead contractor for this package	6
Required resources	26 person-months (13 requested from EU)
Contributing partners	6	4	8	3	18
Person-months (inst/EU)	2/2	6/6	1/1	3/3	1/1
Objectives The main objective is to define the data management structure for the PRISM system. A system needs to be established which ensures an efficient production of diagnostic output files by the PRISM system and allows for the retrieval of files which are easy to process and analyse. On the production side a variety of hardware environments needs to be supported while on the diagnostic side the comfort of the user is the objective. The design of the system will be based on portable file formats and conventions for the meta-data, which fully describes the data to an external user. Furthermore the system to be designed should render the data easily accessible through the network.
Description of work The design of the data management will be divided into two distinct but related phases of the model output handling. The first one will deal with the output of diagnostics from the model in such a way that the data can easily and efficiently be archived and handled on the various systems. The second part will concentrate on the processing, analysis and visualization of the data by the user on his workstation. For these two phases different constraints will govern the system design. WP4a and to some extent WP3 will deal with the implementation of the data management system. T1 - Define the meta-data which needs to be written by the PRISM system together with the diagnostic output. This step will set the information needed from the model and the experimental design and kept in the output files. An analysis of existing model metadata representations will be carried out as part of this task. T2 - Define the file formats, which need to be supported for model output so that the PRISM system can be used efficiently on the various climate research centres. T3 - Write specifications for the output library to be used in the PRISM system. This means choosing the operations, which can be performed by the library and the allowed sampling strategies. T4 - From the meta-data written by the model with the output library a subset needs to be selected for the processing, analysis and exchange of the data. T5 - Because of the diversity of hardware installed in the climate research centres, different formats are needed for the model output. But for the processing, analysis and exchange of results a common data formats needs to be chosen. This data format must be capable of being read and written by widely used analysis packages, and must be suitable for easy exchange of data between institutions. This format needs also to be supported by the output library. T6 - Existing diagnostic and processing tools will be surveyed to identify the operations, which are commonly required for data analysis. A template will be developed which describes this set of operations in terms compatible with the common meta-data format. T7 - Data produced and needed by the various models will reside on different computers. To accelerate this exchange of data between the user and the model an efficient method needs to be implemented. Software allowing such a remote and partial access exists but an implementation is needed which makes best use of the PRISM data format and which takes into account the diversity of hardware.
Deliverables D2c1: Definition of the meta-data for the model output (month 4) D2c2: Definition of the file formats supported for output from the PRISM system (month 4) D2c3: Specification of the output library (month 6) D2c4: Definition of the meta-data for the data exchange (month 6) D2c5: Definition of file format for the data exchange (month 6) D2c6: Catalog of the most important diagnostic operations (month 8) D2c7 A demonstration of remote data access system will be implemented which allows for the user remote access to model output and for the model to read in ancillary data sets (month 12)
Milestones Month 4: completion of D2c1 and D2c2 Month 6: completion of D2c3, D2c4 and D2c5 Month 8: completion of D2c6 Month 12: completion of D2c7

Work package title	Development of PRISM System Coupler
Start date or starting event	1
Required input	Coupler and model interface library specifications from WP2a Coding rules and portability criteria from WP2b Existing coupler from partner 8
Lead contractor for this package	8
Required resources	78 person-months (39 requested from EU)
Contributing partners	8	18	19	20	3
Person-months (inst/EU)	15/15	18/18	3/3	2/2	1/1
Objectives Develop a parallel and scalable PRISM System coupler and model interface library according to specifications defined in WP2a, conforming to coding rules and portability criteria defined in WP2b, based on the existing coupler from partner 8. Document the PRISM System coupler and the model interface library with a written documentation and a WEB presentation.
Description of work The coupler will drive the whole PRISM system, ensuring the synchronisation of the different units and the exchange of the coupling fields directly between the units or via the coupler. When needed, the interpolation of the coupling fields under constraints typical in geophysical coupling will be performed by the coupler. The starting point of the work will be the existing coupler developed by partner 8 consisting of a communication and an interpolation library. The functionalities of the coupler will be extended to meet specifications further defined in WP2a, including features regarding scalability and portability. In particular, the coupler itself will be parallelized and will allow a parallel exchange of information with parallel models. The following tasks will be performed: T1- Develop a generic interface library, to be included in the models, allowing a standard communication with the coupler (or directly with the other units if specified). This first version of model interface will allow a parallel communication between the different processes of a parallel model and the monoprocess coupler. (mo 1 to 12) T2- Improve the coupler communication library to meet specifications defined in WP2a and standard coding rules defined in WP2b, and to include the possibility of a parallel communication between a parallel model and the coupler. (mo 1 to 12) T3- Rewrite and improve the coupler interpolation library to meet standard coding rules defined in WP2b and specifications defined in WP2a. (mo 1 to 12) T4- Parallelize the coupler interpolation library and insure its scalability. (mo 13 to 21) T5- Test the coupler on one target platform with a prototype configuration of the PRISM System reproducing a realistic exchange of information between highly simplified parallel models. At this point, the communication will take place directly between the different processes of the parallel models and the master process of the coupler; the coupler will be parallel regarding the interpolation operations performed on the coupling fields (mo 22 to mid-23) T6- Improve the coupler communication library to allow a parallel communication between the different processes of the parallel models and the different processes of the parallel coupler. (mo 24 to 36) T7- Document the coupler and model interface library with a written documentation and a Web-based presentation. (mo 34 to 36)
Deliverables D3a1 Portable PRISM coupler and model interface library, allowing a parallel communication between the different processes of a parallel model and the monoprocess coupler, for preliminary interfacing of PRISM components (T1, T2, T3 completed; month 12). D3a2 Portable, parallel and scalable PRISM coupler and model interface library, allowing a parallel communication between the different processes of a parallel model and the master process of the parallel coupler to be used in demonstration runs (T4 and T5 completed; month 24). D3a3 Portable, parallel and scalable PRISM coupler and model interface library, allowing a parallel communication between the different processes of the parallel models and the different processes of the parallel coupler (T6 completed; month 36). D3a4 The PRISM System coupler and model interface library documentation (T7 completed; month 36).
Milestones Month 12 (T1, T2 and T3 completed) : Completion of D3a1 Month 24 (T4 and T5 completed): Completion of D3a2 Month 36 (T6 completed): Completion of D3a3 Month 36 (T7 completed): PRISM coupler and model interface library documentation (D3a4).

Work package title	Interfacing of Atmosphere with PRISM System
Start date or starting event	1
Required input	System specifications from WP2a Existing atmosphere models from the partners of WP3b Model interface library from WP3a.
Lead contractor for this package	7
Required resources	66 person-month (18 requested from EU)
Contributing partners	7	6	5	1	4	13
Person month (inst/EU)	18/18	6/0	6/0	6/0	6/0	6/0
Objectives To adapt existing atmospheric models used by European climate groups and to implement in these models standard interfaces so that they can interact with other climate components, directly or through a universal coupler. To ensure that all software conforms to defined standards.
Description of work Four of the partners have developed global Atmospheric General Circulation Models (AGCM) that are adapted for climate studies. All these AGCM have been coupled to ocean components and some of them to other components of the climate system (ocean, atmospheric chemistry, land/surface biosphere, etc.) but not in a distributed framework which allow interchange of the different models and components. Two of these global models offer the possibility of a variable resolution and can be used for regional climate simulations. To meet the objectives of this work package, the following tasks will be performed: T1 Common definition of the interface with participants to WP3c-d-e-f-g following the recommendations of WP2a: partitioning of the calculation between the different AGCMs and the other component models of the climate system, exchanged variables,.etc. (mo 1 to 6). T2 Adaptation of the AGCMs for the calculation of the exchanged variables. This adaptation will include variable resolution versions of the AGCMs of partners 6 and 7 (mo 7 to 12). T3 Interface the adapted AGCMs with an early version of the coupler using the interface library developed in WP3a (mo 13 to 18). T4 Contribution to the analysis of the tests performed with the assembled models in the context of WP3h. (mo 19 to 24). T5 Development of a complete documentation including scientific description and technical aspects of the AGCMs (mo 19 to 24).
Deliverables D3b1 Component atmospheric models suitable for coupling to PRISM system (month 18) D3b2 Documentation including scientific description and technical aspects of the models (month 24)
Milestones Month 6 (T1 completed): Review existing codes and decide on best interface strategy. Month 12 (T2 completed): Review model adaptation Month 18 (T3 completed): Delivery of the new adapted atmospheric components to partners of WP3h for assembling in the PRISM system. Month 24 (T4 and T5 completed): Delivery of the new adapted atmospheric components tested with the PRISM system for the demonstration runs.

Work package title	Interfacing of Land-surface schemes with PRISM
Start date or starting event	1
Required input	System specifications from WP2a Existing land-surface schemes from the partners of WP3c Model interface library from WP3a.
Lead contractor for this package	6
Required resources	60 person-month (22 requested from EU)
Contributing partners	6	4	1	5	7
Person-months (inst/EU)	18/18	6/0	6/0	6/0	2/4
Objectives To adapt existing land-surface schemes used by European climate groups and to implement in these models standard interfaces so that they can interact with other climate components, directly or through a universal coupler. To ensure that all software conforms to defined standards.
Description of work The coupling between the land-surface and atmosphere is probably one of the most complex in the climate system because of the wide range to time scales involved. Up to now land-surface schemes were deeply interwoven in European atmospheric models. The proposed interface will change this situation by allowing for a clear distinction between surface and atmospheric processes and thus the integration of land-surface scheme into the PRISM system. T1 - To evaluate the degree to which current land-surface schemes are interwoven in European atmospheric models. This will allow all partners to define the tasks needed to implement in their model the interface proposed by PILPS-4 (mo 1 to 6). T2 - Implementation of the interface by the partners (mo 6-12). T3 - To review the evolution of the implementation of the interface an inter-comparison of the land-surface schemes will be organised. All schemes will be run coupled to a simplified atmosphere. This will test the modularity of the schemes and their ability to adapt to new host models (mo 12-18). T4 - Once a sufficient number of atmospheric models will have integrated the interface an evaluation of this part of the implementation will be performed. Pairwise the land-surface schemes will be exchanged to perform parallel sensitivity experiments. This will test the plug-compatibility of the atmospheric and surface component and the consistency of the new coupling (mo 12-24). T5 - At this stage the interface will have gained sufficient maturity to finalise the documentation and integrate it in the PRISM system (mo 18-24).
Deliverables D3d1 Land-surface schemes suitable for coupling to PRISM system including documentation (T3 completed, month 18) D3d2 Documentation including scientific description and technical aspects of the models (T5 completed, month 24)
Milestones Month 6: Review existing codes and evaluate the degree of re-coding needed by each group (T1 completed). Month 12: Review the modularity of land-surface schemes (T2 completed and T3 underway). Month 18: Interface land-surface schemes model with early versions of the coupler (T3 completed). Month 24: Completion of tests (T4 and T5 completed).

Work package title	Interfacing of Ocean with PRISM
Start date or starting event	1
Required input	System specifications from WP2a Existing ocean models from the partners of WP3e Model interface library from WP3a
Lead contractor for this package	5
Required resources	50 person-month (18 requested from EU)
Contributing partners	5	4	1	6
Person months (inst/EU)	18/18	6/0	6/0	2/0
Objectives To adapt existing ocean models used by European climate groups and to implement in these models standard interfaces so that they can interact with other climate components, directly or through a universal coupler. To ensure that all software conforms to defined standards.
Description of work A number of existing ocean models will be evaluated in terms of the parameters that will need to be exchanged with the atmosphere and sea-ice bio-geochemistry or regional modules. In particular, a strategy for handling non-uniform data will be developed to enable the interfacing of dynamic sea-ice models with ocean models. Specifically the handling of variable sea ice margins and fractional sea ice coverage will need to be considered. The ocean models will be modified to be interfaced with the PRISM system, paying particular attention to the conservation properties of the system with regard to energy, mass, salinity. Test runs of the adapted ocean models will be performed with both one-way and two-way coupling to check the sensitivity of the results to coupling frequency, averaging procedures. In some runs, real cases (e.g. forcing the ocean models with atmospheric reanalyses) will be used so that the performance can be evaluated against observations. The performance of the ocean models in the fully coupled, final system will also be evaluated. The adapted ocean models will be documented in terms of the parameters and methodology required for coupling, as well as the conservation properties.
Deliverables D3e Three ocean models suitable for coupling to PRISM system including documentation (month 18)
Milestones Month 6: Review existing codes and evaluate the degree of recoding needed by each group. Month 12: Review the modularity of ocean models Month 18: Ocean models with early versions of the coupler Month 24: Complete tests

Work package title	Interfacing of Sea-ice with PRISM System
Start date or starting event	1
Required input	System specifications from WP2a Existing sea-ice models from the partners of WP3f Model interface library from WP3a.
Lead contractor for this package	11
Required resources	66 person-month (28 requested from EU)
Contributing partners	11	17	4	6	1
Person-months (inst/EU)	18/18	6/6	6/0	2/4	6/0
Objectives To adapt existing sea-ice models used by European climate groups and to implement in these models standard interfaces so that they can interact with other climate components, directly or through a universal coupler. To ensure that all software conforms to defined standards.
Description of work A dynamic and thermodynamic sea ice module is an integrated part of any climate modelling system. In this workpackage, existing sea ice models used by European climate groups will be reviewed. Based on the review, some of the most robust and best validated models will be adapted to the software requirements defined for the PRISM System. In the adaptation phase, emphasis will also be put on features of special importance for the atmosphere-sea ice-ocean interface, including proper treatment of partial ice and snow covered grid cells, proper treatment of the ocean-sea ice albedo, and strict conservation of heat, fresh water and salt. The adapted sea ice model component will be interfaced with the PRISM coupler, and tested. At the final stage of the project, the PRISM system runs will be supported, and results analysed based on available sea ice and snow observations, and comparable model integrations.
Deliverables D3f Component models suitable for coupling to PRISM system including documentation (18 months)
Milestones Month 6: Review existing codes and decide on best interface strategy Month 12: Review model adaptation Month 18: Interface component model with early versions of the coupler Month 24: Contribute to system tests (WP 3i)

Work package title	Interfacing regional models with PRISM System
Start date or starting event	1
Required input	System specifications from WP2a Existing regional models from the partners of WP3h Model interface library from WP3a.
Lead contractor for this package	10
Required resources	46 person-month (23 requested from EU)
Contributing partners	10	9	4	18
Person-months (inst/EU)	8/8	7/7	7/7	1/1
Objectives To interface several existing European regional climate models with the PRISM system global models, by means of standardising and streamlining global model output/regional model input conventions Explore the use of the PRISM coupler to more dynamically link regional and global model components To ensure that all software conform to defined standards. Participate in the PRISM demonstration in WP5 by extending the global simulations into dynamically downscaled regional simulations; provide these to potential users.
Description of work Regional climate models are feasible tools to achieve high resolution regional scenarios requested by users performing impact analyses. These can in the end be used to set political negotiation targets and strategies as well as adaptive strategies on those societal aspects where the time horizon is long, such as investments in major infrastructure. 3 of the partners have regional climate models (RCMs) for dynamical downscaling of GCM-scenarios for the European region. Regional ocean/lake/hydrology components are also available. The first task is: T1 - review present practices of GCM-RCM coupling and decide on the PRISM interface strategy. (mo 1-6) To run regional climate models requires GCM data with high frequency output. The quality of RCM-simulations also strongly depends on the quality of the GCMs providing the large-scale data. So, there is a close coupling between the use of global and regional models. In PRISM, the flow of information from GCMs to RCMs is improved. To reduce the uncertainties in the GCM-RCM-scenario chain, it is important to facilitate the cross-use of many GCMs and many RCMs. For this, development of GCM-RCM coupling routines and co-ordination of GCM and RCM simulations, starting from the design of GCM output is needed. This is done in two tasks: T2 - standardise, within the PRISM system, GCM output and the participating RCM input for offline use. (mo 6-18) T3 - investigate the use of the PRISM system coupler to permit offline or online GCM-RCM coupling. (mo 12-24) Finally, the regional models, with the improved interface to the global models will participate in the PRISM demonstration experiments in WP5.
Deliverables D3h1 Component regional models suited for offline coupling in the PRISM system and documentation (T1, T2 completed; mo 18) D3h2 Report on the use of the PRISM coupler to interface RCMs and GCMs (T3 completed; mo 24)
Milestones Month 6: (T1 completed): Review of suitable RCMs and formulated RCM-GCM interface strategy Month 18 (T2 completed): Off-line standardised GCM-RCM coupling. Month 24 (T3 completed): Interface regional models with the PRISM system coupler.

Work package title	Assembling of the PRISM System
Start date or starting event	6
Required input	Coupler and model interface library specifications from WP2a Coding rules from WP2b Coupler and models from WP3a to WP3g
Lead contractor for this package	3
Required resources	62 person-month (31 requested from EU)
Contributing partners	3	18	19	20	8
Person-months (inst/EU)	18/18	4/4	4/4	4/4	1/1
Objectives: To combine components of the climate system as developed in WPs 3a-3h into the comprehensive PRISM system To run this PRISM system for selected combinations of model components To implement this PRISM system on different platforms
Description of work The PRISM System will be composed of models simulating the diverse components of the Earth climate system interacting in a standard way with one or more other components. In this particular work package the various climate system component models developed in the other work packages (WP3a-WP3g) will be combined to form complete climate system models. These complete climate system models will be run on a number of supercomputers. The vendors will help with their know how to install the codes or their particular machines. However, meta-computing will not be attempted, since it is considered as too costly in human resources at present. This means that each coupled model assembled through the web interface will run entirely on one machine. This package delivers the basis for the web based system developed in WP 4b. The system will be tested in stages. The reproducibility of results on different system configurations for parallel runs will be verified. The definition of coding rules and the mechanism of software quality assurance defined in WP2b will be applied to all the software. The final product will be documented. The starting point will be the coupler developed in WP3a, one atmosphere model developed in WP3b and an ocean model developed in WP3e. The model components developed in the other WP3 c, d, e, f, g will be added gradually. The system will be tested on several platforms. Following tasks are envisaged: T1 - Couple a system consisting of one atmosphere and one ocean model, run it on a single platform (mo 6 to 12) T2 - Couple a system consisting of a different atmosphere model than in T1 and the same ocean model as in T1, run it on a single platform (mo 13 to 18) T3 - Transfer the system developed in T1 and T2 to different platforms (mo 19 to 24) T4 - Couple the system developed in T1 and T2 with the other components of the climate system, run it on a single platform (mo 25 to 30) T5 - Transfer the systems developed in T4 to different platforms (mo 31 to 36) The PRISM System will be composed of models simulating the diverse components of the Earth climate system interacting in a standard way with one or more other components. In this particular work package the various climate system component models developed in the other work packages (WP3a-WP3g) will be combined to form complete climate system models. These complete climate system models will be run on a number of supercomputers. However, meta-computing will not be attempted, since it is considered as too costly in human resources at present. This means that each coupled model assembled through the web interface will run entirely on one machine. This package delivers the basis for the web based system developed in WP 4b. The system will be tested in stages. The reproducibility of results on different system configurations for parallel runs will be verified. The definition of coding rules and the mechanism of software quality assurance defined in WP2b will be applied to all the software. The final product will be documented.
Deliverables D3i A system of different climate model components able to run in different combinations as climate system model on different computer platforms
Milestones Month 18 (T1 and T2 completed): A coupled model consisting of two components running on one platform Month 24 (T3 completed): A coupled model consisting of two components running on several platforms Month 30 (T4 completed): A coupled model consisting on various components running on one platform Month 36 (T5 completed): A coupled model consisting on various components running on several platforms

Work package title	Development of Data Generation, Diagnostic and Visualization Tools
Start date or starting event	1
Required input	Definition of the common PRISM metadata and data formats. Definition of the interface to the output library. Description of operations required for the processing library.
Lead contractor for this package	4
Required resources	180 person-month (90 requested from EU)
Contributing partners	4	6	3	2	19	18	20	21
Person-months (inst/EU)	18/18	12/12	12/12	21/21	9/9	2/2	10/10	6/6
Objectives To provide a library that can generate output from PRISM models in the supported PRISM data formats. To provide a flexible library of tools to facilitate processing and analysis of data in the common PRISM data format and to promote sharing of data and analysis programs. To provide high-end and low-end graphical interfaces to display the data.
Description of work An output library meeting the requirements of the defined interface will be written and tested to evaluate its ability to write data from the model into files with the chosen data formats and the common metadata convention. The library will then be optimised for the various architectures on which the PRISM system will run. This task includes machine independent vectorization and parallelisation. Data produced and needed by the various models will reside on different computers. To accelerate this exchange of data between the user and the model an efficient method needs to be implemented. Software allowing such a remote and partial access exists but an implementation is needed which makes best use of the common PRISM data format and which takes into account the diversity of hardware. In order to do this, utilities will be needed for conversion between the common data format and any of the local data formats, which the climate centres decide will continue to be used in the future. Such utilities will be the responsibility of those institutions to provide. The template for the processing library will be implemented as a software library. The language used will be chosen according to the requirements that it can pipe its output into graphics packages, it can manipulate data and metadata together (e.g. using derived data types or objects), it can be compiled and run on any machine and preferably that it is widely used and non-proprietary. If possible, existing processing software will be built upon in producing the library. The library will be able to read and write data in the common data format. An analysis program will be written with this library to demonstrate its usefulness for data from any PRISM model in the common format. Existing graphic packages will be surveyed to identify the operations, which are required to use them for the analysis programs developed in PRISM. A low-end graphical display of PRISM data will be demonstrated and a high-end graphical display system will be developed together with the vendors.
Deliverables D4a1 A first version of the output library which can be used in models, thus validating the user interface D4a2 An performance benchmark for the output library representative of PRISM D4a3 An optimised version of the output library which takes into account the performance enhancement possible on the various hardware used by the PRISM system D4a4 A remote data access system will be implemented which allows for the user remote access to model output and for the model to read in ancillary data sets. D4a5 Review of existing graphics packages and definition of common requirements for graphics for ordinary analysis purposes (low-end) and high-quality presentational purposes (high-end). D4a6 A developed processing library that meets the architectural requirements, delivers the functionality defined in D4a5 and agrees with the data formats of D4a2. D4a7 Identification or provision of graphics packages which meet the low-end and high-end graphics requirements. D4a8 A demonstration programme including high-end and low-end graphical display using the defined data formats. D4a9 A final report providing advice to members on development of existing systems to allow continued integration and proposing future development areas.
Milestones Month 12: Completion of D4a1, D4a2, D4a5 Month 24: Completion of D4a3, D4a4 Month 30: Completion of D4a6 and D4a7 Month 36: Completion of D4a8 and D4a9

Work package title	System Installation, Communication and Networking
Start date or starting event	1
Required input	Existing model presentation systems from partner 16 The component models as produced by WP3a to WP3g The assembled PRISM system as set up by WP3i
Lead contractor for this package	3
Required resources	101 person-month (61 requested from EU)
Contributing partners	3	16	15	8
Person-months (inst/EU)	18/18	18/36	3/6	1/1
Objectives to generate a web-based system which is capable: to assemble desired model configurations, to submit these models on different computer platforms, to monitor the progress of model integration, to control the database archiving of the model data, to link the model experiment with the diagnostic package (WP4a).
Description of work The PRISM System will be composed of models simulating the diverse components of the Earth climate system interacting in a standard way with one or more other components. The PRISM system will link the different model components, take care that the appropriate input data are provided and submit the model to a selected computing site. It will supervise the experiment during its execution, will supervise the archiving of the generated data and will eventually start off a diagnostic programme. The PRISM system will be based on an existing system used by partner 16 to submit and to run a single model on a single computing site. The system consists of a Java based web browser interface and a supervisor programme, which controls and monitors the models simulations. Partner 15 will help with the definition of the functionality of the system. It is planned to extend the existing system to work with different models, to be able to assemble a model from different components, to work in different computing environments, and to supervise the archiving of the results. The PRISM system will be tested in stages. The mechanism of software quality assurance defined in WP2b will be applied to all the software. The final product will be documented. Following tasks will be performed: T1 The portability and adaptability of the existing system from partner 16 to the needs of the climate modelling community will be assessed, and the integrability of similar systems of the other project partners into a common system will be analysed. (mo 1 to 3) T2 The web-surface and functionality of the system will be defined and added to the PRISM Specification Document (mo 4 to 6) T3 A prototype of a web-based system will be set up to run different models on a single site (partner 16) (mo 7 to 18) T4 The web-based system has to implemented at different sites and tested with various models (mo 19 to 27) T5 The system has to be linked with the Diagnosis system developed in WP4a (mo 28 to 36)
Deliverables D4b A web-based system will be delivered, which will be capable of assembling and running climate system models, monitoring the progress of numerical experiments with this model, supervising the archiving the data generated by these experiments and starting off the diagnostic tools developed in WP4a.
Milestones Month 6 (T1 and T2 completed): Well defined functionalities of the web-based system as well as a plan how to adapt the existing system to these functionalities Month 18 (T3 completed): A prototype of the web-based system runs for different models on a single site (partner 16) Month 27 (T4 completed): The web-based system is implemented at different sites and tested with various model configurations Month 36 (T5 completed): The system is linked with the Diagnosis system developed in WP4a