HTAP Report, Sub-Chap. 6 - Data/Info System
This wiki page is intended to be the collaborative workspace for Task Force members interested in the HTAP Information System development, implementation and use. The contents can be freely edited by anyone who is logged in. Comments and feedback regarding the wiki page or its content can also be sent by email to firstname.lastname@example.org.
The purpose of Chapter 6 is to discuss the need to integrate information from observations, models, and emissions inventories to better understand intercontinental transport. A draft outline of Chapter 6 is also on this wiki. This section focuses on the information system in support of the integration of observations, emissions and models for HTAP.
The previous chapters have assessed the current knowledge of HTAP through the review of existing literature and through a set of model intercomparison studies. These assessments highlight the complexities of atmospheric transport and our limited ability to consistently estimate the magnitude of hemispheric transport. A reconciliation of the observations with the current set of models is an even larger challenge.
Recent developments in air quality monitoring, modeling and information technologies offer outstanding opportunities to fulfill the information needs for the HTAP integration effort. The data from surface-based air pollution monitoring networks now routinely provide spatio-temporal and chemical patterns of ozone and PM. Satellite sensors with global coverage and kilometer-scale spatial resolution now provide real-time snapshots which depict the pattern of industrial haze, smoke, dust, as well as some gaseous species in stunning detail. Detailed phisico-chemical models are now capable of simulating the spatial-temporal pollutant pattern on regional and global scales. The ‘terabytes’ of data from observations and models can now be stored, processed and delivered in near-real time. The instantaneous ‘horizontal’ diffusion of information via the Internet now permits, in principle, the delivery of the right information to the right people at the right place and time. Standardized computer-computer communication languages and Service-Oriented Architectures (SOA) now facilitate the flexible access, quality assurance and processing of raw data into high-grade ‘actionable’ knowledge suitable for HTAP policy decisions. Last but not least, the World Wide Web has opened the way to generous sharing of data, models and tools leading to collaborative analysis in virtual workgroups. Nevertheless, air quality data analyses and data-model integration face significant hurdles. The section below presents an architectural framework, an implementation strategy, and a set of action-oriented recommendations for the proposed HTAP information system.
HTAP Information System (HTAP IS)
The HTAP Information System has to facilitate communication, provide a shared workspace for collaboration and offer tools and methods for data analysis. The HTAP IS needs to facilitate open communication between the collaborating analysts. The IT technologies may include wikis and other groupware, social software, science blogs, Skype, etc., along with the traditional communication channels. The HTAP TF also needs to have a work space where the tools and artifacts of the task force are housed. The most important component of the HTAP IS is the HTAP Integrated Data System (HTAP IDS) discussed below.
Thus, the HTAP IS will consists of a connectivity infrastructure (cyberinfrastructure), as well as a set of user-centric tools to empower the collaborating analysts. The HTAP IS will not compete with existing tools of its members. Rather it will embrace and leverage those resources through an open, collaborative federation philosophy. Its main contribution is to connect the TF participating analysts, their information resources and their tools.
A focal point of HTAP IDS is the integrated dataset to be used for model evaluation and pollutant characterization. The front end of the data system is designed to produce this high quality integrated dataset from the available observational, emission and modeling resources. The back-end of the data system is aimed at deriving knowledge from the data, i.e. a variety of the comparisons: model-model, observation-observation, model-observation, etc.
The primary goal of the data system is to allow the flow and integration of observational, emission and model data. The model evaluation requires that both the observations and if possible the emissions are fixed. For this reason it is desirable to prepare an integrated observational database to which the various model implementations can be compared to. On one hand, the integrated dataset should be as inclusive as possible. On the other hand, such a goal needs to be tempered by many limitations that preclude a broad, inclusive data integration
The proposed HTAP data system will be a hybrid combination of both distributed and fixed components as illustrated schematically in Figure ??.
HTAP Integrated Data System (HTAP IDS)
The HTAP IDS will facilitate seamless access to distributed data, allow easy connectivity of data processing components through standard interfaces, and provide a set of basic tools for data processing, integration, and comparisons.
Traditionally data processing, integration, and model comparisons have been performed using dedicated software tools that were handcrafted for specific applications. The Earth observations and modeling of hemispheric transport is currently pursued by individual projects and programs in the US and Europe. These constitute autonomous systems with well defined purpose and functionality. The key role of the Task Force is to assess the contributions of the individual systems and to integrate those into a system of systems.
Both the data providers as well as the HTAP analysts-users will be distributed. However, they will be connected through an integrated HTAP database which should be a stable, virtually fixed database. The section below describes the main component of this distributed data system.
The multiple steps that are required to prepare the integrated dataset are shown on the left. The sequence of operations can be viewed as a value chain that transforms the raw input data into a highly organized integrated dataset.
The operations that prepare the integrated dataset can be broken into distinct services that sequentially operate on the data stream. Each service is defined by its functionality and by firmly specified service interface. In principle, the standards-based interface allows linking of service chains using formal work flow software. The main benefit of such a Service Oriented Architecture (SOA) is that allows the building of agile application programs that can respond to changes in the data input conditions as well as the output requirements.
The service oriented software architecture is an implementation of the System of Systems approach, which is the design philosophy of GEOSS. Each service can be operated by autonomous providers and the "system" that implements the service is behind the service interface. Combining the independent services constitutes System of Systems. In other words, following the SOA approach, not only the data providers but also the processing services can be distributed and executed by different participants. This flexible approach to distributed computing allows the distribution of labor and the creation of different processing configurations.
The part of the integrated data system to the right of the integrated dataset (Figure ??) aids the analysist in performing high level analysis such as data data..., in particular, the comparison of models and observations. At this time, neither the specific model evaluation protocols nor the supporting information system is well defined. It is anticipated, however, that the observational and modeling members of the HTAP TF will develop such protocols soon. The service oriented architecture is well suited for the rapid implementation of model intercomparison techniques.
Interoperability Work within HTAP TF
The adoption of a set of interoperability standards is a necessary condition for building an agile data system for HTAP. During 2006/2007, members of HTAP TF have made considerable progress in evaluating and selecting suitable standards. They also participated in extending several international standards, such as standard names and a standard data query language.
The methods and tools for model inter comparisons was a subject of a productive workshop at JRC Ispra in March 2006. European and American members of the HTAP TF presented their respective approaches to model and data intercomparisons as part of their respective projects ENSEMBLES, Eurodelta, ACCENT, AEROCOM, GEMS, and DataFed. Several recommendations were made to improve future use of intercomparison data. The most important recommendation was to agree on a common data format, netCDF CF conventions and to devolop a list o standard names.
The naming of individual chemical parameters will follow the CF convention used by the Climate and Forecast (CF) communities. The existing names for atmospheric chemicals in the CF convention were inadequate to accommodate all the parameters used in the HTAP modeling. in order to remedy this shortcoming the list of standard names was extended by the HTAP community under leadership of C. Textor. She also became a member of the CF convention board that is the custodian of the standard names. The standard names for HTAP models were developed using a collaborative wiki workspace. It should be noted, however, that at this time the CF naming convention has only been developed for the model parameters and not for the various observational parameters.(See Textor, need a better paragraph).
For modeling data, the use of netCDF-CF as a standard format is recommended. The use of a standard physical data format and the CF naming conventions allows, in principle, the seamless connection between data provider and consumer services. The netCDF CF data format is most useful for the exchange of multi-dimensional gridded model data. It was also demonstrated that the netCDF format is well suited for the encoding and transfer of station monitoring data. Traditionally, satellite data were encoded using the HDF format.
The third aspect of data interoperability is a standard data query language through which user services request specific data from the provider services. It is proposed that for the HTAP data information system adapts the Web Coverage Service (WCS) as the standard data query language. The WCS data access protocol is defined by the international Open Geospatial Consortium (OGC), which is also the key organization responsible for interoperability standards in GEOSS. Since the WCS protocol was originally developed for the GIS community, it was necessary to adapt it to the needs of "Fluid Earth Sciences". Members of the HTAP group have been actively participating in the development and testing of the WCS interoperability standards.
The first stage of IDS is wrapping the existing data with standard interfaces for data access. This is the approach taken in the federated data system DataFed. Given a standard interface to all datasets, the Quality Assurance service can be performed by another provider that is seamlessly connected to the data access service. Similarly, the service that prepares a dataset for integration can be provided by another service in the data flow network. This flexibility offered through the chaining and orchestration of distributed, loosely coupled web services is the architectural framework for the building of agile data systems for the support of future demanding HTAP applications. These value adding steps have to be performed for each candidate dataset that is to be used in the integrated dataset.
Standards-based data access can be accomplished by ‘wrapping’ the heterogeneous distributed data into standardized web services, callable through well-defined Internet protocols (SOAP, REST). Homogenization can also be accomplished by physicaly transfering and uniformly formating data in a central data warehouse. The main benefit of virtualy or physucaly homogenious data access is that further processing can be performed through reusable processing components that are loosely coupled. The result of this ‘wrapping’ process is an array of homogeneous, virtual datasets that can be accessed through a standard query language and the returned data are packaged in standard format, directly usable by the consuming services.
(This is just a place holder. Need help from WMO. Here we need to explain the quality assurance steps that are needed to prepare the HTAP Integrated Dataset. For true quality assurance and for data homogenization the data flow channels for individual datasets need to be evaluated separately, as well as in the context of other data. QA should be ubiqutous process occuring throughout the data flow and processing. This means, that IS needs to provide channels for a QA feedback to the data providers. )
Homogenization and Integration
The HTAP Integrated Dataset will be used to compare models to observations. It will be created from the proper combination of a set of surface, upper air and satellite observations. Monitoring data for atmospheric constituents are now available from a variety of sources, not all of which are suitable for the integrated HTAP dataset. A set of criteria for the data selection is given below.
- The suitability criteria may include the measured parameters, their spatial extent and coverage density, as well as the time range and sampling frequency with major emphasis on data quality (defined as???).
- Initially, focus should be on PM and ozone, including their gaseous precursors. Subsequently, data collection should also include atmospheric mercury and persistent organic pollutants (POP).
- In order to cover hemispheric transport, the dataset should accept and utilize data from outside the geographical scope of EMEP.
- Special emphasis should be placed on the collection of suitable vertical profiles from aircraft measurements as well as from surface and space-borne lidars.
- The data gathering should begin with the existing databases in Europe (RETRO, TRADEOFF, QUANTIFY, CRATE, AEROCOM) and virtual federated systems in the US (DataFed, Giovanni, NEISGEI, and others).
- TF data integration should also contribute to and benefit from other ongoing data integration efforts, most notably, with ACCENT project in Europe, and similar efforts in the US.
However, before the inclusion into HID, each dataset will need to be scrutinized to make it suitable for model comparison. The scrutiny may include filtering, aggregation and possibly fusion operations.
A good example is the 400-station AIRNOW network reporting hourly ozone and PM2.5 concentrations over the US. The network includes both urban sites that are strongly influenced by local sources. These need to be removed from or flagged in the integrated dataset since they are not appropriate for comparison with coarse resolution global models. Developing the specific criteria and procedures for the HTAP integrated dataset will require the attention of a HTAP subgroup.
Given the limited scope and resources of HTAP, it will be necessary to select a suitable subset of the available observational data for the preparation of the 2009 assessment. The evaluation of suitable observational datasets for model validation and fusion will require close interaction between the modeling and observational communities. A wiki workspace for open collaborative evaluation of different datasets is desirable.
HTAP Information Network Implementation
The above described architecture needs to be implemented as soon as possible so that the HTAP integrated dataset can be created and the model data comparisons can commence. An important (incomplete) set of nodes for the HTAP information network already exist as shown in Figure ??. Each of these nodes is, in effect, is a portal to an array of datasets that they expose through their respective interfaces. Thus, connecting these existing "data portals" would provide an effective initial approach of incorporating a large fraction of the available observational and model data into the HTAP network. The US nodes DataFed, NEISGEI and Giovanni are already connected through standard (or pseudo-standard) data access services. In other words, data mediated through one of the nodes can be accessed and utilized in a companion node. Similar connectivity is being pursued to the European data portals Juelich, AeroCom and EMEP and others.
(Here we could say a few words about each of the main provider nodes) Federated Data System DataFed; NASA Data System Giovanni; Emission Data System NEISGEI; Juelich Data System; AeroCom; EMEP.
HTAP Datasets - need list of others
See 20 selected datasets in the federated data systems, DataFed; TOMS_AI_G - Satellite; SURF_MET - Surface; SEAW_US - Satellite; SCIAMACHYm - Satellite; RETRO ANTHRO - Emission; OnEarth JPL - Satellite ; OMId - Satellite; NAAPS GLOBAL - Model; MOPITT Day - Satellite; MODISd G - Satellite; MODIS Global Fire - Satellite; MISRm G - Satellite; GOMEm G - Satellite; GOCART G OL - Model; EDGAR - Emission; CALIPSO - Satellite; AIRNOW - Surface; VIEWS OL - Surface; AERONETd - Surface; AEROCOM LOA - Model
HTAP Relationship to GEO and GEOSS
There is an outstanding opportunity to develop a mutually beneficial and supportive relationship between the activities of the HTAP Task Force and that of the Group of Earth Observations (GEO). The national and organizational members of GEO have adapted a general architectural framework for turning Earth observations into societal benefits. The three main components of this architecture are models, and observations, which feed into decision support systems for a variety of societal decision making processes.
This general GEO framework is well suited as an architectural guide to the HTAP program. However, it is void of specific guidelines and details that are needed for application areas such as HTAP. The HTAP program provides an opportunity to aplly and extend the GEO framework. In case of HTAP, the major activities are shown in the architectural diagram of Figure ?? The HTAP modeling is conducted through global scale chemical transport models. The observations arise from satellite, ground-based and airborne observations of chemical constituents and their hemispheric transport. In case of HTAP, a third input data stream is needed for emissions which i neither observation, nor a model.
The HTAP decision support system consists primarily of humans. They need to be supported by an IT infrastructure and a set of tools that amplify ....
The first cluster is composed by the analysts who are the members of the HTAP task force. Their products are the 2007 and 2009 assessment reports (rectangle in Figure ??) submitted to the HTAP co-chairs and to EMEP. A second shorter report is prepared and submitted to the EMEP executive body, which is the decission making body of the LRTP convention. (Terry, Andre this description of the HTAP DSS needs your help).
Developing a higher resolution design chart for the HTAP DSS is an important task because it can guide the design and and implementation of the supporting information system. Furthermore, the more detailed DSS architectural map may also serve as a communications channel for the interacting system of systems components. The insights gained in developing the HTAP DSS may also help the DSS design in similar applications.
The implementation of the GEO framework utilizes the concept of Global Observing System of Systems (GEOSS). Traditionally, Earth observations were performed by well defined systems such as specific satellites and monitoring networks which were designed and operated using systems engineering principles. However, GEO recognized that the understanding of Earth system requires the utilization and integration of the individual, autonomous systems. The key difference between the Systems and System of System approaches are highlighted in Table 1.
While system science is a well developed engineering and scientific discipline, the understanding and development of System of Systems is in its infancy. The work of HTAP TF may provide an empirical testbed for the study of this new and promising connectivity architecture. Since, the HTAP TF activity encompasses virtually all aspects of GEOSS system of systems integration, it is an attractive "near-term opportunity" to demonstrate the GEOSS concept. An initial low-key demonstration could be accomplished as part of the HTAP TF 2009 assessment. Such a GEOSS demonstration is particularly timely since the data resources, data mediators and the connectivity infrastructure is nearly ready to be connected into a system of systems. Also, there are strong sociatal drivers to extend an update of LRTP convention to include the air pollution impacts of one continent to another.
An HTAP-GEOSS demonstration would also demonstrate System of Systems approach not through stovepipe but through a dynamic network approach.
The HTAP program can also be considered an early demonstration of the GEO concepts through its end to end approach. Both the atmospheric modeling, as well as the observations currently exist through the operation of existing modeling and observation systems. The Task Force has agreed organizing and evaluating those models, assembling and integrating the observational datasets and then reconciling the models with the observations will be the focus of the Phase II effort. The Task Force will also prepare and deliver a report to LRTP to aid its deliberations and its decision making processes. This sequence of activities constitutes an end to end approach that turns observations and models into actionable knowledge for societal decision making. One could say that this is an octagonal approach to more deliberate step by step development of GEOSS where in Phase I interoperability, in Phase II ???
HTAP Relationship to IGACO
The HTAP program can also have a mutually beneficial interaction with the Integrated Global Atmospheric Chemistry Observations (IGACO) program. The IGACO is part of the Integrated Global Observing Strategy (IGOS). IGACO proposes a specific framework for combining observational data and models. It also specifies the higher level information products that can be used for creating social benefit.
In the IGACO framework, the data products from satellite, surface, and aircraft observations are collected for inclusion into an integrated data archive. A key component in the IGACO data flow is the mandatory quality assurance and QA QC that precedes the archiving. This is particularly important for HTAP where multi-sensory data from many different providers are to be combined into an integrated database. In the IGACO framework, a key output from the data system is a high integrated spatial-temporal dataset which can be used in a multitude of applications that produce social benefit. The goal of creating such an integrated dataset is shared by the IGACO and the HTAP programs (?? Len, this section could benefit from your input)
HTAP Relationship to other Programs
There are various ways of source apportionment and those need to be compared receptor oriented methods and also reconcile the receptor and the forward oriented methods