Transition of the 8-bin dust component into NMM-dust was performed by extending the NMM model through the development of an additional dust concentration module. Since the dust component has a similar (mass conserving) structure as, for example, the equation for specific humidity in NMM-dust, dust and specific humidity were handled similarly. They differ in the deposition (source) and emission (sources) as parts specific for dust. These two components were done in a separate routine, following the way of parameterization in the 8-bin code. The two models (8-bin and NMM-dust) are initially tested by comparing to the PHAiRS project's 4-bin code and matched the results in principle. The 8-bin and NMM-dust models are utilized for model interoperability studies and yield good results. Initial tests within the project have been performed with the DREAM-eta 4 bin system (Nickovic et al. 2001). In this system, DREAM model dust particle distribution is described with 4 sizes, ranging in radius from 0.7 - 40 µm. Since the hydrostatic NCEP/Eta cannot be applied with resolution higher than ~ 5km in horizontal, in the next step we replaced the NCEP/Eta with the Nonhydrostatic NCEP/NMM (Janjic et al., 2001) that may be used in a high resolution mode. At the same time NCEP/NMM has a possibility to be run in a parallel execution mode, which is not the case with NCEP/Eta. Theoretical thoughts was made to transfer the DREAM8 component (Nickovic, 2004; Nickovic 2005; Peres et al., 2006) into the NCEP/NMM, DREAM8 being the dust model version that resolves dust particle distribution with 8 particle bins ranging from submicron radius to 10 µm. Smaller particles are used in the project having in mind that such particle size range is more appropriate for possible future use of the modeling system for public health. Incorporating the knowledge gained, we developed the NMM-dust model in the project.
Through the project, interoperability is built into several activities: a) model interoperability is investigated, b) a data service is provided with web map service and web coverage service, c) data ingest is implemented with web coverage service. These aspects made it possible for broader adoption of the research results into other systems. This project made significant contributions to the interoperability capabilities of the foundation PHAiRS project through the development of enhanced OGC services. Specifically, the development and deployment of additional time-enabled WMS for DREAM-eta model outputs has facilitated the deployment of these Earth science products into public health decision support systems (SYRIS) while also demonstrating the value of developing standards-based delivery systems (i.e. KML) into other, more general visualization systems.
While the current project has identified limitations in the current generation of data delivery server software, it has also highlighted areas for further development in these applications, particularly in the areas of server and client support for the OGC standards. These lessons learned are outlined in the next section.
The interoperability concept utilized in this project aims to permit different modelling systems with different domains and resolutions to communicate in a general manner. This is achieved through the use of following approaches:
• a common pre-processor for both DREAM4-eta and NMM-dust; a corresponding decision was therefore made to use NMM with eta-like pre-processing.
• a common post-processor for both DREAM4-Eta and NMM-dust to provide
• common input formats for initial/boundary conditions for the higher resolution nested model members in the interoperability modelling chain
• common input for either GRADS or GRASS GIS post-processing and visualization
This required all initial/boundary conditions inputs prepared in the regular lat-lon grids at (10 in our setup) standard pressure levels for conventional and dust concentration data. As a common interface for these input parameters, the GRIB1 format was chosen to be the primary data format between models at all interface levels. Conventional GRIB encoder/decoder software programs were modified to be able to process dust parameters. The modification relates to the extension of the GRIB metadata information (i.e. extension of parameter tables that originally do not include aerosol data).
While achieving significant results in itself, this project has also identified several areas for future research that are worthy of continuing investment. These areas include technical the application aspects.
3.3.1 Technical
Migrate the model into a WRF-NMM or AGCM: The development choose the NMM model according to the progress and project team's experience. A future step of migrating the model onto a WRF-NMM or AGCM, the current and/or future operational model environment, would eliminate the gap for widely adoption of the model. It would also allow continuous scanning of North America, for example, for conditions leading to dust storms, and then focusing, or downscaling, to provide timely, "at your zip code," warnings and forecasts of dust concentrations for public safety and health.
Continue model testing and evaluation such as to document skill score and other measures of performance. The models tested here have been validated through comparison with one specific dust storm scenario. Utilizing historical data and EPA in-situ sensor observations to validate the accuracy of the model would be needed.
Improve model applications through data assimilation: More data, such as soil wetness and seasonal vegetative cover will be needed to improve the scientific value and credibility of the model.
Integrate into operation: integrate the model within an operation environment to support a GEOSS societal benefit area will further enhance the utility of the modelling system.
Continue to develop server applications that host OGC WCS services to support time-enabled services. The specific combination of applications used in this project includes MapServer and PostgreSQL/PostGIS, but other server platforms (i.e. ArcSDE/ArcGIS Server) could also benefit from their expansion of support for time-enabled WCS services.
Develop standards-based service interfaces for common meteorological forecasting models that allow for the integration of initialization parameters obtained from remote data servers.
Develop queue-based model execution environments in which subsequent model runs may be defined and executed based upon the results of previous model runs.
Develop high-performance computing capabilities that are capable of executing on-demand model runs that are customized based upon the outputs of other model runs.
3.3.2 Application
Public interest and unsolicited enquiries from a user base broader than the one targeted by the NASA-sponsored PHAiRS project, have been eye opening. PHAiRS has demonstrated that NASA's current observation system is essential if the World Meteorological Organization's International Sand and Dust Storm Warning and Assessment System and any new applications implied by these enquiries are to succeed. The UA and UNM current, NASA-funded, projects expanding upon DREAM and inventing new applications of NASA products include:
• DECISIONS, "Adding NASA Earth Science Results to EPHTN via the NM/EPHT System," a three-year project linking the universities of New Mexico and Arizona, will apply new A-Train and dust forecast products as well as already proven MODIS observations to the New Mexico Environmental Public Health Tracking System, New Mexico's contribution to CDC's Environmental Public Health Tracking Network (EPHTN);
• RPC, "Rapid Prototyping Capability For Environmental Factors Affecting Asthma and Allergies," aka the "Pollen Project," in collaboration with Marshall Space Flight Center and the Universities of New Mexico and Arizona, explores DREAM modifications and additional NASA "vegetation" products to simulate and predict pollen emission and downwind dispersal, a prototype for phenological properties and events.
One of us (Sprigg) leads the international drafting team for plans to implement the WMO Sand and Dust Storm Warning Advisory and Assessment System. The world community would welcome a modeling system empowered by the work accomplished here. This project is a step-function increase in opportunities and applications for our regional atmospheric models.
New clients and applications we may now anticipate include:
• Wildfire control-burn scheduling and management;
• Forest fire-fighting strategic and tactical support for fire-fighter safety;
• Assessment of superfund site remediation;
• Assessment of health risks due to toxic waste dumps, relaxation of cross-border truck emission standards with Mexico
• Assessment of proposed expansion of open-pit mining in Arizona as copper prices continue to rise;
• Assessment of proposed expansion of open-pit uranium mining on Navaho reservations in the four-corners region of New Mexico , Colorado , Arizona and Utah as nuclear power becomes a more popular energy-producing option;
• Forecasts and hindcasts of visibility-reducing airborne dust conditions for USAF flight support;
• Forecasts and warnings of visibility-reducing dust storms for highway safety and emergency response personnel; and
• Warnings and assessments related to natural, accidental, or intentional release of aerosols and particulates for purposes of homeland security and emergency response.
The team anticipates global use of the new models in the WMO Dust Storm Warning System and, concomitantly, providing a US contribution to the GEO. If carried forward, the models can be an adjunct to Natural Hazards (a service of NASA's Earth Observatory) , and contribute to UN-SPIDER , enhancing NASA programs, priorities and product applications.
But, we should take our current success a step further to truly capitalize on it. We will demonstrate that the models work together in a few realistic, but not operational, situations. However, the system should be run over a period of time under varying environmental conditions and larger geographical area to see if our interoperability tests are consistent and to estimate error and bias of the modeling system and the satellite data feeding it. We need to quantify joint observation/model reliability and performance as a measure of credibility for our current clients in public health services and to convince potential clients of broader applications for this satellite-based observation and modeling system. Furthermore:
(a) The product of the UA and UNM collaboration with Marshall Space Flight Center (the Phenology Regional Atmospheric Model) and the USGS's National Phenology Network, and the WRF-NMM should be added to this project' high-performance interoperable system. The models are similar. Considerable flexibility will be gained with little effort in comparison.
3.1 Model 
3.2 Interoperability