The Carbon Tracking Center (CTC)

Terug naar hoofdpagina

Science case

Integrated Monitoring

Although monitoring of greenhouse gas balances starts with high quality observations, interpretation of small signals and attribution to specific regions and processes depends on numerical models. This is because most observations inform about the total CO2, or CH4, or N2O present at a given time and location, but they do not carry a specific signature of for instance forest-carbon, crop-carbon, Dutch-methane, or Spanish-methane. The many contributions that together determine the total balance have to be modeled independently, and then integrated numerically to understand the observations.

In turn, the numerical models depend on the observational data to stay within credible bounds for their calculated balances. This is because without guidance from the observations, the numerical models will generate their own physically correct truth, but not necessarily the one we observe. The formal fusion of observational data and numerical models is referred to model-data fusion, or data assimilation.

Data-assimilation forms the heart of the daily weather forecast, in which millions of measurements are fused with a physics-based prediction of the state of the atmosphere (such as temperature, humidity, and pressure). The result is a detailed weather estimate that is consistent with both the observations, and with the process knowledge represented in the numerical model. Developing a similar data assimilation method for greenhouse gas exchange is a rapidly evolving scientific discipline in which the Dutch scientific community plays a pioneering role. It combines high-performance scientific computing, novel data-assimilation algorithms, and innovative modeling techniques.

Numerical Challenges

It is instructive to further compare the challenge that we face for greenhouse gases with numerical weather prediction. In numerical weather prediction millions of observations from balloons, surface stations, and satellites are integrated in a numerical model to provide a weather forecast. Every 6 or 12 hours, sophisticated data assimilation algorithms estimate a new atmospheric state that is most consistent with observations and physics, and from that point on a new (and better) weather forecast for the next 10 days is started. The analysis of the state thus allows improved forecasts, while decade long sequences of the analyzed states are used for scientific research.

In greenhouse gas research we also require a good analysis of the past and present state, and a reliable prediction for the future. However, several aspects make the problem more complex than data assimilation for the weather:

  1. We have a much more limited set of observations to work from, even if the proposed ICOS-EU monitoring infrastructure is fully developed.
  2. There are more modeling components needed to make a good physical description of the system with a numerical model. This includes for instance descriptions of atmospheric trace gas transport, climate variations, vegetation growth and decay, nutrient cycling in soils, and water table variations in wetlands.
  3. The modeling components have more complex interactions, and over a wider range of time scales. Some of these interactions are poorly characterized, and most of them cannot be constrained by observations.
  4. Large heterogeneity in land management and land-use history require very high model resolutions to resolve, and
  5. Human actions will have an enormous impact on the GHG balance for the next decade.

The complexity of different data streams, interaction between variables, and uncertainty in observation and model output that is integral to the multiple constraint approach, make data assimilation one of the fundamental tools to make progress in greenhouse gas cycle research. In essence, data assimilation is an integration tool. The integration refers to the combination of different streams of information from trace gas measurements, satellites, accounting efforts, agricultural statistics, and mechanistic models of the greenhouse gas cycle components involved in the context of a model system. The total information will only be greater than the sum of the parts if effective ways to combine the strengths of each part of the system can be found. Data assimilation techniques have shown the potential to fulfill this role.

The role of the CTC

Data assimilation at such a scale requires substantial computer resources, and a team of scientists, scientific programmers, and software developers to be successful. Quick and transparent access to the many streams of data must be facilitated, and operational efforts must co-exist with research and development of the system. The latter requires a well defined testing environment where new observations, new constraints, new scenarios, and new modeling components can quickly be assessed. Clearly, not every country in the EU can independently build and maintain this capacity for data assimilation. Within ICOS-NL, we aim to build this capacity through the ICOS-NL Carbon Tracking Center and offer it as a service to Europe. This center will:

  1. Build and host an ICOS-EU Enterprise Service Bus (ESB) that enables seamless exchange of observational data, modeling I/O, and Carbon Tracking products between partners, public, and policy makers.
  2. Make the ESB available to all ICOS-EU researcher that want to use its functionality, provide their data, access data, or contribute their greenhouse balance products to the Carbon Tracking Center
  3. Host a state-of-the-art ICOS-NL data assimilation system that can be used by all members
  4. Operationally produce multiyear greenhouse gas balances that are country and process specific and include uncertainties
  5. Deliver the source code, computing facilities, and expertise needed by others who want to contribute under an open source license
  6. Assist the development of the ICOS-NL data assimilation system by maintaining a research and testing environment for relevant numerical models.
Figures of an ESB design for ICOS-EU
Diagram showing the new ICOS-NL Carbon Tracking Center and the services connected to it (purple colors). The CTC will be a Central Facility within ICOS-EU (blue colors) and provide services for all ICOS-EU centers, as well as other EU projects related to greenhouse gas monitoring (yellow). External data providers and data users will be integrated through the Enterprise Service Bus, which provides seamless data exchange between all partners.

Operational Services

We plan to base the operational component of the center on one of the leading efforts in the carbon cycle community: CarbonTracker. CarbonTracker estimates global CO2 exchange on a 1x1 degree latitude/longitude grid (equivalent to ~ 60x100 km over the Netherlands) each week. In ICOS-EU, an even finer mesh is needed to accurately estimate CO2 and other greenhouse gas. This requires the ‘nesting’ of a mesoscale atmospheric model -the Weather Research Forecast (WRF) model- into the global CarbonTracker simulation. Mesoscale models are now capable of representing most of Europe on a 4x4 km grid, allowing much more detail in how the surface area is represented, and in resolving GHG transport patterns. Ample experience with meso-scale transport modeling is available within the ICOS-NL consortium.

At the land-surface we need a biogeochemistry model in which photosynthesis, respiration and carbon storage interacts with climate, nutrient availability, and human management of agricultural and forest ecosystems. The model should calculate the exchange of carbon dioxide and its 13C isotope, and possibly also the release of CH4 and N2O. The energy, water and greenhouse gas cycles must interact on all time scales to properly capture the effects of droughts, which is one of the major climatic disturbances of the European GHG balance. Since few models exist meet all these requirements, we will start from an easily available candidate and add required capabilities in research mode. Atmosphere-land interactions is an active topic of research within the ICOS-NL consortium and several candidate models (SIBCASA, LPJ, NOAH, SWAP-C) are available.

Operational Carbon Tracking products will include daily estimates of regional GHG exchange, estimates of fossil fuel emissions for the major industrial areas of Europe, reanalysis of major European carbon cycle events such as droughts, and also specialized products for policy makers such as GHG trends, country totals, and a break-down of GHG balances by sector. Uncertainty estimates based on multiple emission and management scenarios will be part of this. All results will be disseminated following the ICOS-EU open access policy, and be fed into the Enterprise Service Bus together with applications to extract, summarize, and visualize the data in different ways for different stakeholders.

Research & Development testbed

In addition to operational products, the ICOS-NL center will enable research across ICOS member countries by providing components from the operational system to research efforts. A strictly modular design of the data assimilation system will allow interested parties to take for instance the operational model infrastructure, but substitute their own meso-scale transport model that might have specific physics to deal with elevated terrain. Or similarly, some groups might want to use the infrastructure to develop and test new methods to assimilate oxygen measurements into the system. The ICOS-NL Carbon Tracking Center will provide such efforts with the computer code, system documentation, training, and supercomputing resources to complete their research. Our European partners can thus use the ICOS-NL infrastructure to perform scientific experiments, similar to service offered by for instance CERN and ECMWF.

Talent case

Innovation case

A major innovation that will speedup the progress and allow researchers to focus on the fundamental issues of data/model fusion in carbon cycle and greenhouse gas science is that the CPF will serve also as a data facility for transport models and assimilation systems. It will provide a unique set of unified interfaces to all the data needed for data assimilation systems, such as global and high resolution meteorological information for the off-line models, land use data sets, and of course observations. It also serves through another set of unique interfaces as a receiver of the modeled data results, such as generated prior and posterior flux fields and modelled values of concentrations at receptor locations.
The interfaces will be open source, extensible and multi-platform.
The interfaces can be used within the computing facilities of the CP, but also external online access will be provided.
Analysis tools to compare model results with observations and among other model results and to derive ensemble results where will be constructed using these interfaces, driven by user demands.

Partnership case

  1. IBM

Developing a robust, efficient and expandable software architecture for the Carbon Tracking Center requires specific expertise not readily available in the science community. To develop a state-of-the art infrastructure ICOS-NL will form a partnership with IBM-Research. The collaboration focuses on the development of a so-called ICOS Enterprise Service Bus (ESB) tailored to the needs of the Carbon Tracking Center.

Figures of an ESB design for ICOS-EU
An Enterprise Service Bus will be developed for ICOS-EU in partnership with IBM

The ICOS-ESB basically provides a set of communication protocols that facilitates the exchange and sharing of information and resources between users and providers of observed data. As such the facility enables links with data enhancement algorithms, model pre- and postprocessors, data repositories, and supercomputing facilities. The ESB itself is open source and transparent. Any entity wishing to hook up has to develop an interface between the ESB and their data/application. An example would be an application that extracts the latest CO2 measurements from the ESB and presents to a user the CO2 growth rate in a graphical format. As far as needed the ESB can also facilitate access control and payment services for proprietary products.

The infrastructure of the ICOS-ESB closely resembles the ‘Delta-ESB’, an initiative around measurement data and models that are used in the water management and water research sectors. IBM is currently involved in a feasibility study, in which ten applications will be connected to the Delta-ESB. In this study a case will be defined around the ICOS-ESB, allowing for a quick start. Results of this pilot project will be presented October 4th, 2011. In subsequent stages a dedicated project will further develop the ICOS-ESB in consultation with all national and international partners.

  1. EarthNetworks

Business case

Total Costs

non-NWO contributions

Requested NWO financing

Technical case

Possible focus for the Netherlands

Critical mass

Embedding

Proven willingness to cooperate

Reflection of social trends

Other relevant information

Unless otherwise stated, the content of this page is licensed under Creative Commons Attribution-ShareAlike 3.0 License