Prototype System for Multidisciplinary Shared Cyberinfrastructure:...

ryhetas

dengls

nethurndngl-

onng

rnil:

mnsil:

er,les

mg,il:

m-intil:

i-n-3–

stasedic4-

Prototype System for Multidisciplinary SharedCyberinfrastructure: Chesapeake Bay

Environmental Observatory

CBEO Project Team

Abstract: A prototype system for a Chesapeake Bay Environmental Observatory �CBEO� is under development by a multidisciplinateam of researchers from the domains of environmental engineering, marine science, hydrology, ecology, and computer science. Tvision is to provide new means of coupling and synthesizing field sampled and model generated data in a way that will open up new dasources to researchers and managers interested in understanding and resolving some currently unanswered questions and problemconcerning hypoxia in the Chesapeake Bay. It will do so by developing advanced cyberinfrastructure to provide uniform nationwiaccess to new tools and a wide variety of data of disparate type, scale, and resolution, in both spatial and temporal domains, includimodel-derived data from past runs of major computational models for Chesapeake Bay hydrodynamics and water quality. Some key goaof the prototype project are to resolve the existing data source heterogeneities such that all relevant data are accessible through ointerface, to archive and facilitate the analysis of model input and output files, and to provide new shared tools for data analysis, all withe goal of transforming the way scientific research and science-based management is conducted on the Chesapeake Bay. There are foproject teams operating separately but in close and continuous communication. The teams’ objectives are to make simultaneous aparallel advances in �1� environmental observatory network design and nationwide network access to Bay data �CBEO:N�; �2� furtherithe educational missions and outreach at the host institutions �CBEO:E�; �3� providing a test-bed application that will allow the deveopment and testing of new cyberinfrastructure and data analysis tools �CBEO:T�; and �4� using all of the above-mentioned advancesfocused science questions to demonstrate the transformative nature of the CBEO for addressing research questions and improvi

1William P. Ball, Professor and CBEO Project Co-Director, Dept. ofGeography and Environmental Engineering, Johns Hopkins Univ., 3400North Charles Street, Baltimore, MD 21218. E-mail: [email protected]

2Damian C. Brady, Postdoctoral Associate and CBEO Test-Bed TeamMember, Dept. of Civil & Environmental Engineering, Univ. of Dela-ware, 301 DuPont Hall, Newark, DE 19716. E-mail: [email protected]

3Maureen T. Brooks, Faculty Research Assistant and CBEO ScienceTeam Member, Univ. of Maryland Center for Environmental Science -Horn Point Laboratory, 2020 Horns Point Rd., Cambridge, MD 21613.E-mail: [email protected]

4Randal Burns, Assistant Professor and CBEO Test-Bed Team Leader,Dept. of Computer Science, Johns Hopkins Univ., 3400 North CharlesSt., Baltimore, MD 21218. E-mail: [email protected]

5Benjamin E. Cuker, Professor and CBEO Education and OutreachTeam Leader, Dept. of Marine and Environnemental Science, HamptonUniv., Hampton, VA 23668. E-mail: [email protected]

6Dominic M. Di Toro, Professor and CBEO Project Director, Dept. ofCivil and Environmental Engineering, Univ. of Delaware, 345 DuPontHall, Newark, DE 19716. E-mail: [email protected]

7Thomas F. Gross, Program Specialist, UNESCO, 1, rue Miollis,75732 Paris, France; formerly Administrative Director for CBEO Projectand Research Scientist with Chesapeake Research Consortium, Chesa-peake Community Modeling Program, Edgewater, MD 21037. E-mail:[email protected]

8W. Michael Kemp, Professor and CBEO Science Team Leader, Univ.of Maryland Center for Environmental Science-Horn Point Laboratory,2020 Horns Point Rd., Cambridge, MD 21613. E-mail: [email protected]

9Laura Murray, Professor and CBEO Education and Outreach Team

Leader, Univ. of Maryland Center for Environmental Science—HoPoint Laboratory, 2020 Horns Point Rd., Cambridge, MD 21613. [email protected]

10Rebecca R. Murphy, Graduate Student and CBEO Test-Bed TeaMember, Dept. of Geography and Environmental Engineering, JohHopkins Univ., 3400 North Charles St., Baltimore, MD 21218. [email protected]

11Eric Perlman, Graduate Student and CBEO Test-Bed Team MembDept. of Computer Science, Johns Hopkins Univ., 3400 North CharSt., Baltimore, MD 21218. E-mail: [email protected]

12Michael Piasecki, Associate Professor and CBEO Network TeaLeader, Dept. of Civil, Architectural, and Environmental EngineerinDrexel Univ., 3141 Chestnut St., Philadelphia, PA 19104. [email protected]

13Jeremy M. Testa, Graduate Student and CBEO Science Team Meber, Univ. of Maryland Center for Environmental Science-Horn PoLaboratory, 2020 Horns Point Rd., Cambridge, MD 21613. [email protected]

14Ilya Zaslavsky, SDSC Spatial Information Systems Laboratory Drector and CBEO Network Team Leader, San Diego Supercomputer Ceter, Univ. of California San Diego, 9500 Gilman Dr., La Jolla, CA 92090505. E-mail: [email protected]

Note. Discussion open until March 1, 2009. Separate discussions mube submitted for individual papers. The manuscript for this paper wsubmitted for review and possible publication on June 14, 2007; approvon December 5, 2007. This paper is part of the Journal of HydrologEngineering, Vol. 13, No. 10, October 1, 2008. ©ASCE, ISSN 1080699/2008/10-960–970/$25.00.

960 / JOURNAL OF HYDROLOGIC ENGINEERING © ASCE / OCTOBER 2008

management approaches for large coastal systems that are heavily affected by humans. Finally, this CBEO cyberinfrastructure develop-rkalle

e;

ment is geared toward ensuring that the envisioned system is integrated into larger nationwide environmental observatory netwoinitiatives �e.g., Water and Environmental Research System, Ocean Research Interactive Observatory Networks, National EcologicObservatory Network, and Long Term Ecological Research� thus helping to lead the way toward the development of a continental-scaenvironmental observatory network.

DOI:

CE Database subject headings: Chesapeake Bay; Environmental engineering; Hydrodynamics; Research; Internet; DrainagEcology.

Motivation and Overview

n-hehehehed

a-ofe-chhehehe�.orretentn-

iscta-d

n-n-ita

edal

observatory networks, as schematically illustrated in Fig. 1. As aornd

-or

rytss,i�.iltF

er,edheis

ndsoI-d-als,to

altochedhea-e-illli-miaa-deheAa-c-ndr-s-y-toedo-ed

a

ge

The use of large-scale environmental observatory systems to coduct transformative research has received much attention in tenvironmental science and engineering community recently. TNational Science Foundation �NSF� is currently supporting tdevelopment of environmental observatory prototypes with taim of building a substantive knowledge base for conceiving anbuilding these systems. Some of NSF’s environmental observtory initiatives include the Consortium for the AdvancementHydrologic Science �CUAHSI 2007�, the Collaborative Largscale Engineering Analysis Network for Environmental Resear�CLEANER 2007�, both of which are now combined to form tWater Environmental Research Systems �WATERS� network, tNational Ecological Observatory Network �NEON 2007�, and tOcean Research Interactive Observatory Networks �OOI 2008Within this context, the NSF solicited proposals in late 2005 fprojects to develop and deploy a “prototype cyberinfrastructufor environmental observatories” �CEO:P� and then “demonstra�its� viability in order to inform the planning of, and developmeof, an environmental cyberinfrastructure for large-scale, enviromental observing systems” �NSF 2005�.

The Chesapeake Bay Environmental Observatory �CBEO�currently being designed by the multidisciplinary CBEO projeteam as a prototype to demonstrate the potential of cyberinfrstructure for transforming environmental research, education, anmanagement. Because the Chesapeake Bay is a large environmetal system that is host to numerous ecological, environmental egineering, hydrologic, coastal, marine, and social complexities,is ideally suited to serve as a prototype test bed designed forlarge and very diverse user community. In fact, it is envisionthat the system will later be integrated to multiple environment

Fig. 1. Conceptual integration of CBEO into the other three larscale environmental observatories

JOURN

result, this project pursues an end-to-end approach to two majcomponents of cyberinfrastructure, i.e., a test-bed observatory aa new type of node for an existing national network. These components will be useful for both regional applications and fshared cyberinfrastructure with other researchers nationwide.

The CBEO is under development by a multidisciplinaproject team that is led by PIs comprising marine/ocean scientis�Cuker, Gross, Kemp, Murray�, computer scientists �BurnZaslavsky�, and environmental engineers �Ball, DiToro, PiaseckThe NSF required that the observatory be designed and buusing existing cyberinfrastructure components from other NSprograms. Toward this end, the San Diego Supercomputer Centwhich has deployed software components of the NSF-sponsorGeoSciences Network �Baru 2004�, is working as part of tCBEO team to implement and develop a CBEO portal on thnetwork. The CBEO project not only uses software stacks aother approaches developed for the GEON web interface, but alintegrates metadata standards developed through CUAHSHydrologic Information Systems/WATERS and applies data feeration and querying approaches developed for the NationVirtual Observatory �Szalay et al. 2002; Malik et al. 2003�. Thuthe prototype components developed in this work are assuredbe flexible and amenable to extension and upgrade.

Finally, a key objective was that the project “works with reenvironmental data, includes a mechanism for general usersaccess the prototype, and demonstrates the utility of its approaby attracting users outside of those researchers directly involvin the project.” In this context, the CBEO team has targeted tseasonal depletion of dissolved oxygen, or hypoxia, in the Chespeake Bay as a fundamental issue around which to develop spcific environmental engineering and science questions that whelp to structure, test, and demonstrate the prototype’s capabities. The use of seasonal hypoxia as a key environmental probleis well suited because of the long-standing history of hypoxresearch in the Bay and the general importance of this issue ntionwide. In addition, the likelihood is high that it will provianswers to key questions about coastal hypoxia that will lie at tintersections of existing observational and modeling data sets.key premise of the work is that cyberinfrastructure and informtion technology are now at a point where new tools can be effetively developed and deployed to better link, analyze, avisualize multiple observational and modeling data sets of dispaate size, scale, and resolution �in both spatial and temporal apects� that will transform our ability to address large, policoriented science questions. The ability of cyberinfrastructureperform such functions, however, has yet to be properly explorand demonstrated. It is our hope that the CBEO and other prottypical projects and test beds will be able to provide the needresearch and development for such a demonstration.

The CBEO prototype development effort was motivated by

AL OF HYDROLOGIC ENGINEERING © ASCE / OCTOBER 2008 / 961

d-inl-

keo-gn

e-toce

edon

hypothesis that the development of new and better tools for fining, viewing, and analyzing multiple data sets and data streams,conjunction and comparison with each other and with modederived data, should lead to new knowledge about ChesapeaBay hypoxia and toward better understanding of management slutions. This manuscript provides a summary of the project beinundertaken to test this hypothesis, with a focus on the desigchallenges and project approach.

Fig. 2. Fixed stations for tidal water quality monitoring under theat multiple depths once or twice monthly �less frequently in winterfrom the U.S. EPA Chesapeake Bay Program Office, Annapolis, M

962 / JOURNAL OF HYDROLOGIC ENGINEERING © ASCE / OCTOBER

Science and Cyberinfrastructure Challengeson the Chesapeake Bay

Chesapeake Bay as a Study Site

The Chesapeake Bay �CB, Fig. 2� is an ideal location for rsearchers to demonstrate how cyberinfrastructure can be usedhelp answer complex and unresolved science questions, enhan

peake Bay Tidal Monitoring Program. Monitoring has been conduct1985 and at some stations for much longer. �Reprinted with permissi07.�

Chesa� sinced., 20

2008

the ability of educators to teach environmental science, and in-istha-al.,

ofc-ofal.c-n

mch.,

hep-asal.n

u-2

toll

hehe

ta,ved

hev/rehes-lea--

rs,n-dd

alo-al..,

eskeal.r-ergy

ofntn

oricb�orre

itychoral

ter

form management of environmental resources. The Chesapeakea well-studied and intensely monitored coastal ecosystem wiactive research and resource management activities. The degradtion of water quality, seagrass communities, and benthic animpopulations from human activities is well documented �e.gKemp et al. 2005�. The region has a long-standing historycollaboration between researchers and managers, with the objetive of basing management on sound scientific understandingsystem processes and responses to disturbance �e.g., Malone et1993; Boesch et al. 2001�. Federal mandates and multijurisditional initiatives have motivated an aggressive Bay restoratioprogram �Boesch et al. 2001�. A systematic monitoring prograhas been in place since 1984, and numerous integrated researprograms have assessed various aspects of system dynamics �e.gKemp et al. 2004, 2005; Roman et al. 2005�. In addition, tChesapeake Bay management program has supported develoment of a coupled hydrodynamic and water quality model that hbeen calibrated �Cerco and Cole 1993; Cerco 1995; Johnson et1993�, upgraded �e.g., Cerco and Noel 2004�, and repeatedly ruto produce multiyear �1986–2000� simulations of physical circlation and ecological dynamics, at spatial scales of 10 to 1 kmhorizontal and 1 to 5 m vertical and temporal scales of minuteshours �Cerco 1995�. These and other models are available to aresearchers and managers, access to which is facilitated by tChesapeake Community Modeling Program managed through tChesapeake Research Consortium �CRC et al. 2008�.

In addition, there is a wealth of collected and sampled daa wide variety of incoming data streams, and a very extensinumerically generated data. Some of the existing data andata streams have already been collected and organized by tU.S. EPA Chesapeake Bay Program Office �http://www.epa.goregion03/chesapeake/; http://www.chesapeakebay.net/�, whethey have been made available for public access through tChesapeake Information Management System �CIMS�. This sytem provides a rich source of information that is already availabfor enhanced use through the development of new cyberinfrstructure tools. The vision and challenge for the CBEO is to demonstrate that cyberinfrastructure can provide hypoxia researchemanagers, and educators with transformative new tools for sythesizing, harvesting, analyzing, and interpreting spatially antemporally distributed information of unprecedented diversity andensity.

Chesapeake Bay Science: Challenges, Questions,and Resources

Hypoxia as a Prototypical Challenge for Scientific ResearchThe seasonal depletion of dissolved oxygen �O2� from coastwaters, i.e., hypoxia, is a widespread problem of growing prportions that is tied to human influences �Rabalais et2002; Howarth et al. 2000� and is of worldwide concern �e.gRosenberg 1990; Diaz 2001�. Accumulating evidence indicatthat the intensity and extent of summer hypoxia in ChesapeaBay have been increasing since the early 1950s �Hagy et2004�. Although interannual variations in extent and duration corelate with fluctuations in freshwater discharge �Fig. 3�a� upppanel�, long-term increases follow broad trends of increasinnutrient loading from the watershed �Fig. 3�a� lower panel, Haget al. 2004�. Thus, hypoxia in coastal waters is a consequenceinteractions between watershed land-use �and associated nutrieloading�, hydrology, climate, and oceanographic processes. A

JOURN

Fig. 3. �a� Relationships between bay hypoxia and river flownutrient loading. �a� Time-integrated volume of hypoxic and anox��0.2 mg O2 l−1� water versus winter-spring river flow. �Mid-summer volume of anoxic water versus nitrate loading fearlier �1950–1979� and more recent years �1980–2001� �figuadapted from Hagy et al. 2004�. �b� Simulations of a water qualmodel for Chesapeake Bay �Cerco 1995� showing �a� close matfor bottom dissolved oxygen concentrations at seasonal scales fone location �CB5.2�, but �b� relatively poor match for interannuvariations in the integrated volume of summer hypoxic wa�C. Cerco, communication�.


environmental observatory could transform our process under-nston-ortar-

t�.

ia.at

3�,he–t-ser-utntu-he

ps.,lsarl-isee

y-erd

cen-esa-al.edi-

heutedte.

userntk-veg

orly,g

a-nttod

of

computer models �Table 1�. Overall, the time and space scales ofti-

kendls

ndis

heerh-e-ty,esntif-er,u-n.nds-asly

a-e-a

ndketothrer-

ta,ic

o-if-talb-ts,hentngnde-d.nse-ssa-ed

e-ndeyal,

standing, because some fundamental and unanswered questioabout hypoxia remain �Nixon 1995, Cloern 2001�. Our abilityunderstand and address these scientific questions would be ehanced in transformative ways with vastly improved methods ffinding, integrating, interpolating, and visualizing multiple dasets �including model output� that are complementary but dispaate in their structure, precision and scale �resolution and exten

Chesapeake Bay Science QuestionsThe team identified three related questions pertaining to hypoxThe first derives from Fig. 3�a� �lower panel�, which indicates thfor a given rate of N loading �total N is highly correlated to NOa larger hypoxic volume has generally been observed during tlast 20 years in contrast to the previous three decades �19501979�. The considerable scatter in the relationships is largely atributable to fluctuations in river flow. If proven significant, thedifferences would imply the existence of highly nonlinear undelying mechanisms that are currently not well understood bwhich have major implications for prediction and manageme�Kemp et al. 2005�. Conclusively validating these trends and elcidating their causes would transform our understanding of tprocesses causing hypoxia.

The second question pertains to the observed relationshibetween year-to-year variations in hypoxia and river flow �e.gFig. 3�a�, upper panel�. Close examination of these trends reveaconsiderable scatter which is poorly understood, and it is unclewhy the existing coupled hydrodynamic-water quality model, athough calibrated to reproduce seasonal patterns �Fig. 3�b��,incapable of simulating these interannual trends �Fig. 3�b��. Wsuspect that there are problems both with estimating the timcourse of hypoxic volume and with modeled processes, both hdrodynamic and biological. Exploring these issues requires bettintegration of existing observational data sets with each other anwith the model-derived output.

The third question involves the regional and seasonal balanbetween source and sinks of organic matter that regulate O2 cosumption and hypoxia in bottom waters. Although prior analyssuggested that organic matter production by algae in shallow wters may drive respiration in deep channel waters �Kemp et1997�, the general structure of the water quality model being usfor management �Cerco and Noel 2004� reflects the more tradtional view that phytoplankton production in surface waters of tdeeper regions of the Bay is the major source of organic inpfueling hypoxia �e.g., Malone et al. 1988�. A recently establishshallow water monitoring program should provide the requisidata �MD DNR 2005� that will be incorporated into the CBEO

Data Resources for the Chesapeake BayVaried observatories and monitoring programs generate numeroChesapeake Bay databases relevant to hypoxia including watquality conditions, physical structure and circulation, sedimecharacteristics, biogeochemical processes, abundances of planton, benthos, and fish populations. Extensive sets of data habeen and are being collected at a wide range of scales usindiverse means including grab samples, moored and towed senssystems, and satellite and aircraft remote sensing. Additionaldata are available from other disparate sources: acoustic samplinof water depth, sediment character and fish abundance; quantittive and qualitative chemical and physical data from sedimecores that have been used to recreate ecological history andmeasure the rates of biogeochemical processes �e.g., Cooper anBrush 1991�; and output data for and from a wide variety


the available data range from minutes to decades and from cenmeters to hundreds of kilometers.

An important goal of environmental observatories is to masuch disparate data sources accessible in a user-friendly way aon a web-based workbench that also provides the necessary toofor properly resolving and interpolating their different spatial atemporal scales and visualizing the results. The realization of thgoal would transform our ability to understand and interpret tdata. For example, Fig. 4 shows fine-scale distributions of watquality obtained from towed undulating sensor systems througout the Bay. Data from this monitoring program include measurments of physical, chemical, and biotic aspects of water qualibut the complementary large databases with key ecological rat�e.g., primary productivity, community respiration and nutrierecycling, and organic particle sinking� were obtained under dferent research programs at different scales and times. Moreovoutput of numerical model computations provides physical circlation, water quality, and ecological properties at a fine resolutioTo address the scientific questions posed, we must interpolate aintegrate the various databases so that computations of flux, tranformation, and coherence can be made and patterns as wellmechanistic relationships can be visualized and quantitativeanalyzed.

The CBEO team is focusing on the use of specific hypoxirelevant data sets from Table 1 to provide new insights into spcific science questions. Through parallel science activities onlocal server containing these multiple data sets �our test bed� athrough cyberinfrastructure activities that are designed to mathese data amenable to analysis over the network, we intenddemonstrate the scientific value of combining disparate data winew tools, and simultaneously develop the cyberinfrastructuneeded for better sharing such data and tools to other users. Oveall, this project aims to use prototypical science questions, daand innovative new approaches to obtain insight into the intrinsvalue of cyberinfrastructure for research.

Challenges and Objectives for SharedCyberinfrastructure on the Chesapeake

Resolving problems of structural, syntactic, and semantic hetergeneity across data sets maintained at different locations for dferent purposes is a major challenge facing all environmenobservatories. In addition, there are challenges relating to prolems of semantic and syntactic heterogeneities among data selack of relevant metadata descriptions for some data, and tmajor technical challenge of integrating data collected at differespatial and temporal resolutions, under different samplischemes, with different frequencies and extents of coverage, awith different levels of precision and uncertainty. Formal reprsentations and metadata for such disparate data sets are requireThis is a prerequisite for developing scalable and robust meafor querying, exploring, and integrating observation data, and prparing them for further analyses. The CBEO project will addrethese challenges by developing data registration, query, integrtion, and analysis tools for shared use within the CBEO test bas well as across multiple networks through the portal �Fig. 1�

Inconsistent Semantic and Syntactic RepresentationsA major challenge in the federation of disparate data sets is dveloping and implementing appropriate metadata standards adata registration systems that fully reflect data properties. The kto data interoperability at a basic level �search, viewing, retriev

2008

esero-teis

aydn,i-

u-altaheedicu-tyofmdse-s-o

veb-edrstaOtaherefi-lya-enantolyp-

Table 1. Space and Time Scales of Chesapeake Bay Databases Relevant to Hypoxia Questions

t

ut

ts

75

t

�,

,

a;

r�mrilan

and analysis� is resolution of syntactic and semantic differencthrough a mediation layer �Ludäscher et al. 2007�. This layreconciles the different metadata standards and implements vcabularies that allow connections between otherwise disparadata descriptions, e.g. “Gauge Height”�“Stage.” Although thexample is trivial, similar inconsistencies exist in the wmetadata annotations are encoded �plain text versus XML� anpublished, and the manner in which data are stored �locatioformat�. This is a serious obstacle to the interoperability of env

DatabaseVariablesincluded

Space

Grain

Grab sample monitoring T, S, Chl, N, P,C, Si, O2, kd

z, 1 mx, 10 km

Fixed sensors

• Shallow �DNR�• Deep �CBOS�

T, S, Chl, O2, kd

T, S, u, v, w0 km10 km

Underway sensors

• Shallow �Dataflow�• Deep �NSF TIES�

T, S, Chl, O2, kd

T, S, Chl, O2,zooplankton

1 mz, 1 m

x, 10 km

Remote sensing

• SeaWiFS �satellite�• ODAS �aircraft�

ChlChl

1–10 km5 m�50 m

Bathymetry Depth, volume z, 1 m,x, 1 km

River inputs Flow 10 km

Nutrient load N, P, Si 10 km

Climate T, rainfall, wind,tides

10 km

WQ model T, S, Chl, N, PC, Si, O2, kd

z, 1 m,x, 1 km

Hydro. model T, S,u, v, w, Kx,y,z z, 1 m,x, 1 km

Note: T�temperature; S�salinity; Chl�phytoplankton chlorophyll-a; Okd�diffuse light attenuation coefficient; u, v, w�velocity componentsaCartesian coordinates: “x” follows land–sea gradient; “y”�horizontal

Fig. 4. Dissolved oxygen �upper panel� and chlorophyll-a �lowederived from transects with a vertically undulating sensor syste�Scanfish� towed along the central axis of Chesapeake Bay �Ap2000�. Data from the NSF LMER:TIES project �Table 1; see Romet al. 2005�.

JOURN

ronmental observatory networks, because the ecological commnity networks are either already using Long Term EcologicResearch, or planning to adopt �NEON� the Ecological MetadaLanguage �EML 2008� for metadata descriptions, whereas tocean community network �OOI 2008� is slated to use IntegratOcean Observing System �IOOS�-developed �Federal GeographData Committee �FGDC�-based� Data Management and Commnications �OCEAN.us 2005� and the WATERS communi�through the Consortium of Universities for the AdvancementHydrologic Sciences, Inc., hydrologic information syste�CUAHSI-HIS�� is planning to use the International StandarOrganization �ISO�-based standard 19115 �ISO 2007�. First expriences with the Chesapeake Bay Information Management Sytem �CIMS� and its metadata annotations have revealed twimportant findings. First, in cases where data providers haadopted a metadata standard to annotate their data, this estalishes a relatively straightforward means to develop a formalizmetadata framework that can be used for crosswalks. The writehave now had experience in mapping the provided metada�CIMS is mostly FGDC based� to a structure based on the IS19115 and have found little difficulty in mapping the metadatags—see also the FGDC-based North American Profile for tISO 19115 �FGDC 2007�. The second finding, which is moimportant in the writers’ view, is that there has been some difculty in finding consistent vocabularies that can be propermapped and also difficulty in reconciling the fact that some metdata tags in one system have no equivalent in the other. This oftmeans that some metadata may be lost when mapping fromexpressive system to one that requires less entries, or havingcreate metadata annotations when trying to adopt from a sparseannotated system. First experiences have shown that the ma

a Time scales

Web site addressor data sourceExtent Grain

Extent�years�

0–30 m300 km

2–4 weaks 20 www.chesapeakebay.ne

1–10 km300 km

15 min15 min

2–55

www.hpl.umces.cbos.edwww.chesapeakebay.ne

www.cbos.org

1–10 km10 m

100 km

4 weeks3 months

1–35–6

www.eyesonthebay.newww.chesapeake.org/tie

Roman et al. 2005

200 km300 km

1–10 day1–4 weeks

1010

Harding et al. 1994Harding et al. 2005

300 km na na Cronin and Pritchard 19

300 km 1 day 50 www.usgs.gov

300 km 1 day 20 www.chesapeakebay.ne

300 km 1–24 h 50 www.noaa.gov

300 km 1 h 15 C. Cerco �US Army CoEprivate communication

300 km 5 min 15 B. Johnson �US CoE�private communication

solved oxygen; N�nitrogen; P�phosphorus; C�organic carbon; Si�silicy-, z-directions; Kx,y,z�turbulent mixing in x, y, z.

erpendicular to x; and “z”�vertical axis.

scales

2�disin x-,

axis p


pings work out relatively easily for most of the CIMS data, al-tahensrereds

ez

p-he.,

rur-tar-e-ise-edssd

a-e-ses,ssn-d

eshemalatveinheetedr,0/s-nterhea

rs

atbets.ale,

entachn,

or averaging. Thus, either the value of nitrate concentration or theri-

oto-ofeledlelde.as

aloralofi-

ltse-ndre

ry.aru-tsa-at

ndale-c-ticN,so

k-erd,s--

isn-ng

fi-kea

edg.,heoredite

r-esht

though the comparatively greater richness of the CIMS metadastructure did provide some challenges. In this regard, tCUAHSI-HIS-based system only requires about 36 annotatioper data value �or set�, which are very basic in their structu�CUAHSI 2007�. Higher degrees of interoperability will thereforequire that metadata tags be mapped, semantics reconciled, anboth syntactics and semantics implemented in a way that allowautomatic mediation without human intervention �e.g., Bermudand Piasecki 2006�.

Integration of Disparate Information SystemsIntegration of data collected within heterogeneous but overlaping cross-disciplinary observation networks requires that tCBEO rest on a sound but flexible technological foundation, e.gthat established by the GeoSciences Network, GEON �Ba2004�. Required components for data integration include: �1� fomat, projection, and unit conversion; �2� knowledge-based daregistration and query rewriting, and �3� spatiotemporal intepolation techniques. The most important challenges are to dvelop a framework that �1� can be modified and extended; �2�sufficiently flexible, transparent, and documented to allow intgration into future new systems; �3� is based on well-establishstandards with a high degree of provenance; �4� has an acceinterface that can be programmed against; and �5� is robust anscalable for the supported data sources, data types, and interpoltion models. Observatory developers must find conceptual reprsentations that describe both the storage structure �i.e., databaand file system� and the descriptive elements �e.g., metadata tagsemantic conventions�. It is also important to permit basic acceto data through web services that are operating system indepedent, can be integrated into multiple user applications anworkflows, and can use a uniform simple logic. Web servicthat provide data exposure to the “outside” world are one of tgreat promises of the Water Environmental Research Syste�WATERS� and other observatory networks and substantiprogress is being made. For example, SOAP web services thpermit access to USGS National Water Information System habeen made public through nationally supported servers for usecustom desktop or web-based end-user applications, such as tData Access System for Hydrology �CUAHSI 2007; Maidmental., 2006�. The CBEO is building on technologies developunder both the GEON and CUAHSI-HIS programs. In particulathe newly developed CBEO portal �http://geon16.sdsc.edu:808gridsphere� is based on GEON technologies, and supports regitration, discovery, and integration of data resources of differetypes, contributed by CBEO members as well as by the widearth sciences community. Access to observation data for tChesapeake Bay will be provided via a Workgroup HIS server,node in the emerging system of hydrologic observation servebeing developed within the CUAHSI-HIS project.

Merging of Data for User ApplicationsNew means will be needed to integrate data that are collecteddisparate spatial and temporal scales, and new approaches willneeded to merge observational data and model-derived resulMany Bay-related data sets fall into two classes: four-dimensiondata �4D=3 spatial+1 time�. If variable type �e.g., temperaturvelocity, concentrations� is also considered as a dimension, ththe data are five dimensional �5D�. In this context, a metadadescription is also an additional variable that is useful for sutasks as “weighting” other types of data for display, interpolatio


metadata about nitrate data quality can be viewed as the “vaable” that comprises the fifth dimension.

Moreover, for any given variable, many types of data do noccur at single locations in space and time, but are rather assciation with polygons in 4D space–time and represent averagessome sort, e.g., a concentration derived from a photograph pixor a measured �model-derived� concentration within a sampl�grid element� volume over the compositing time of the samp�or temporal discretization of the model�. Another example woube water velocity across some geometrically defined interfacThe rest of the data universe can be even more complex, suchdescriptive data for sediment cores and living resources.

The CBEO project is currently focusing on 5D observationwater quality data and the development of better techniques fcomparing among and between different kinds of observationand model-generated data. We are already facing challengeshow best to store, retrieve, and make effective use of the addtional polygonal information relevant to model-generated resuand pixilated data sets. Storage and use of other �even more dscriptive� types of data �such as that from biological sampling asediment cores� will represent even greater challenges, but abeyond the scope of our current prototype.

Configurability of a Virtual ObservatoryThe CBEO will be a multipurpose and multifaceted observatoHowever, each user’s individual view should meet their particulneeds. Such a custom CBEO view will likely represent a particlar spatiotemporal window over a subset of observatory data seand available processing workflows. Several existing cyberinfrstructure projects allow for such personal user areas maintainedgrid nodes, such as �myGEON 2006� in the GEON project, amyLEAD in the LEADPortal �2008� project. In the personareas, authorized users may reference data sets and other rsources found in resource catalogs and perform supported funtions using common software stacks that include semanannotation, data transformation, and online mapping. In GEOan ability to develop user-defined processing workflows is albeing added.

In the proposed system, we will extend the personal worbench concept to enable users to define an integrated view ovselected CBEO resources, specify how these resources are linkeand instantiate the symbolic representation as a standardcompliant XML document or a well-known format with accompanying metadata, which in turn can be used as input to analysprograms. The latter step would rely on GEON or Scientific Evironment for Environmental Knowledge services for resolvisemantic mismatches.

Challenges for InterpolationIntegrating and analyzing relationships between data sets is difcult because of their spatial and temporal heterogeneity. Unlimost database applications, data cannot be joined by equality orsimple predicate. Rather, data have complex relationships definby the underlying science and properties of data generation, e.sampling discipline and measurement error. Generally, putting tdata on a common basis requires interpolation algorithms. Fexample, a scientific question that examines point data collecthourly at buoys with pixilated data collected daily by satellrequires interpolation in space–time.

There are two important and challenging aspects to data intepolation. First, the distributed nature of the CBEO complicatthe evaluation of interpolation functions. Data can be broug

2008

together at a single site and interpolated locally. In a distributedgeel.he5�zepd

exhecthesslyat

n-rets.mren

Innd

ce

we

enr-gy

ayn-a-esh

x-netel-

ord

er.inl-

ined

aoratl�,al

ete-ntt.

Preserve the data polygonal structures. Visualize the results

fyri-orx-ve

x-e–s:ofg.,et

tictoic

alo-ri-oftoce

r-e/edc-ofhes.ndo-mher-ryn;ryll-ed

mOst

ngtassleu-heesndx-n-ngi-

ct.

environment, this strategy requires transfer of unnecessarily laramounts of data, e.g., all readings used in the interpolation kernFor performance reasons, interpolation should be evaluated at tdatabases resulting in smaller data transfers �Malik et al. 200thus improving performance. To allow scientists to customiretrieval and interpolation functions, we propose to develointerpolated spatial joins using SQL language extensions anuser-defined functions. User-defined functions allow complprograms written in Java, C�� and C# to be executed at tdatabase �Chamberlain 1998�. They are the fundamental construthat enables database federations to “bring the computation to tdata” �Szalay et al. 2002�. The challenge here is to provide clalibraries of interpolation functions that can be invoked directand provide customized interpolation functions to be executedthe database.

Second, a common feature of all observational data is its icompleteness in 4D, in contrast to model derived data that acontinuous in time and exist in all polyhedral model segmenThe problem of commonly used interpolation schemes, frosimple weighting schemes, e.g., inverse least squares, to mocomplex methods, e.g., kriging, is the need for an interpolatiometric to be used in relating observed to interpolated points.regard to kriging, for example, ordinary kriging based on locatiomay not be sufficient, as the system is heavily influenced by lanlocations, density gradients, and other conditions that influenthe movement of water and connectivity among data points.

The presence of model data, specifically hydrodynamic flofields, as part of CBEO suggests that an interpolation schemcan be devised that uses this information. This method has beused with flow fields that are estimated from the data to be intepolated �Ueng and Wang 2005; Yang and Parvin, 2003; Zhanand Akambhamettu 2003�. Model-derived flow fields from manyears of hydrodynamic model simulations of the Chesapeake B�e.g., Cerco 1995� are available to the CBEO. These are indepedent of the data and can therefore be used with sparse observtional data. In theory, barriers to transport such as pycnoclinand land boundaries could be properly taken into account througthe effective use of accurate hydrodynamic information, thus epanding our ability to view and interpret sparse observations. Oimportant challenge for the CBEO project team is to demonstrathat our cyber-based observatory can in fact facilitate the deveopment and application of tools of this type.

Challenges from a User’s PerspectiveFrom a user’s point of view, the CBEO should be the vehicle fperforming research using disparate collections of data sets anmodel outputs that could not previously have been used togethIn the context of hypoxia, a user might be most interestedapplying merged 5D data of the type discussed earlier. The folowing are examples in the form of query scenarios.1. Queries for Existence and Extent of Data. For a variable

a particular spatial and temporal domain, with a defingranularity, what is the data coverage? CBEO would return5D data cube representing presence/absence of data that fvisualization purposes could be subsampled: one variablea time �4D�, grouped into time intervals �three dimensionaand spatially sliced and collapsed into two-dimensionslices.

2. Queries for a Data Merge. Select from an extant 5D subsand produce the merged data set. Select an averaging elment in 4D space–time. Populate all the elements with poiestimates or multiple entries if overlapping data are presen

JOURN

using tools described in 1.3. Queries for Data Interpolation. For filled elements, speci

compositing algorithms in time and space �e.g., volumetcally weighted averages that incorporate point values�. Funfilled elements, choose an interpolation algorithm, for eample, through universal kriging with a variety of alternaticovariates.

4. Queries for Analysis. Possibilities are numerous. For eample: �1� Space–time averages: the volume of the spactime isosurface for O2�2.0 mg /L; �2� Areal mass fluxedaily average flux of nitrate across a plane in the bay. FluxO2 across the pycnocline, defined as a surface bisecting, e.a 1.0 g /L m vertical change in salinity; �3� Sources: the nmonthly production of algal biomass in the photosynthezone accounting for loss by turbulent mixing and settlingthe sediment or: the sediment flux of ammonia during oxand anoxic periods.

5. Queries for Model–Data Comparison. A typical annumodel run can produce 100 Gbytes of output for the hydrdynamics and the 25 computed biological and chemical vaables. During the same year there are perhaps 1 Gbyteobservational data. The CBEO can be queried in ordercompute and visualize statistics at various time and spascales.

6. Queries for Data Assimilation. Produce a “source diffeence” 5D cubes in which the fifth dimension is the sourcsink that must have been present to produce the observconcentration, corrected for transport effects using the veloity and mixing data. Arguably, this is the most useful formdata assimilation as the output is a direct quantification of tprocesses responsible for the changes in the state variable

These six query scenarios are the basis for web services auser interface development targeting the following three categries of users: �1� project PIs and their research teams, for whoconfigurability and flexibility of the CBEO environment and tability to quickly integrate data sets are paramount; �2� undegraduate and graduate students, with the emphasis on observatodata exploration, packaged queries, and curriculum integratioand �3� researchers and system integrators outside the observatodomain, for whom CBEO shall provide a gateway to wedocumented Chesapeake Bay observations and model-derivdata.

Project Approach: CBEO Development

To better manage and assign the various tasks the project teahas adopted a framework called NETS. NETS divides the CBEscope into for four subtopics, i.e., �N�etwork, �E�ducation, �T�ebed, and �S�cience. The CBEO:T and CBEO:N teams are workiwith selected databases, data streams, and model-derived dafrom the extensive body of existing data to provide an accemechanism for outside users. The CBEO:E team is responsibfor implementation and dissemination of material that is condcive for K–12 education as well as higher education. TCBEO:S team formulates the science questions and providfeedback to the test-bed and network teams for improving afurther developing cyberinfrastructure components. Previous eperience has shown that the development of a successful enviromental observatory requires a close and early cooperation amodomain scientists and computer scientists. The central role of scence in this project is a key aspect of the CBEO prototype proje


It is critically important that any developments, abstractions, oru-s-reinsts,

a-b-g,n-alis,l-dtsk,alo-hetantisNr-d

ree-

gedreinnseg

otdi-r-toa-n-ta-

a-n

ke-

hea-d

skgr-

CBEO:E—Multicultural Student Development

beonpsu-i-too-ndgh—r-erceatu-anchofal-

gha

h-il-e,s,erhehe

awrsheara-e-

s,e-b-ederd,o-anis

tonstaher-ale-to

ed

applications envisioned by the computer scientists must be evalated by domain scientists as contributory to the solution of presing science questions. This requirement is in addition to the moreadily recognized constraint that the needs/wants of the domascientists be technically feasible and properly balanced againthe time and effort that computer scientists must expend. Thuparallel activities and active dialog are important.

CBEO Network and Cyberinfrastructure Components„CBEO:N…

The CBEO end-to-end cyberinfrastructure will deliver observtion data services that include: �1� registering and annotating oservation data resources of different types; �2� searchinaccessing, and exploring the available metadata and data; �3� cofiguring and transforming the data into a common spatiotemporframework; �4� using the transformed data as input to analysmodeling, and visualization systems; and �5� providing personaized user interfaces to access the services. Such an end-to-ensystem would leverage several cyberinfrastructure componendeveloped in other projects including the Geosciences Networor GEON �Baru 2004; Zaslavsky 2006a,b�, the National VirtuObservatory �Szalay et al. 2002�, and the CUAHSI-HIS compnents �CUAHSI, 2007�. Particularly the deployment of tCUAHSI-HIS time series data structure, or Observational DaModel, a database design specifically geared toward storing poitime series data, together with a number of management tools,of great importance to the development of the CBEO. The GEOcyberinfrastructure follows the principles of services-oriented achitecture and represents a system of point-of-presence data ancompute nodes, each with a respective stack of layered softwacomponents �GEON pack�, for which a CBEO node will be dployed and extended.

Reusing core grid and data management services, includinsecurity and authentication management already implementwithin the GridSphere portal, will let CBEO:N focus on its cocontribution of providing support for additional types of dataCBEO, services for integrating disparate data sets and the meafor interfacing these contributions with tools for analysis. Somspecific contributions will include: �1� modeling and representinthe various types of data being integrated into the CBEO �e.g., nonly point-based time series, but also overflight satellite anmodel-generated data within volume elements as well as velocties and fluxes across interfacial areas�; �2� developing web sevices that extend GEON registration and search functionalitydata types associated with the many different types of observtional and model-generated data that exist for the Bay; �3� defiing standard interfaces for incorporating these additional datypes in the software stack; �4� developing methods for transforming and packaging a user-configured fragment of virtual observtory as an input for analysis, modeling, and visualizatiosoftware; �5� using powerful visualization software products, lithe Integrated Data Viewer, IDV �UNIDATA 2008� or the FERRET system �PMEL 2006�, that have been integrated with tGEON environment �SDSC 2006�; and �6� defining communiction interfaces between the CBEO services and applications, anthe GridSphere portal framework used in GEON. The latter tais particularly important as a step toward making the emergincyberinfrastructure components reuseable across different obsevation networks.


and K–12 Outreach

The educational and outreach component of the CBEO willdeveloped through activities that interface with existing educatiprograms. The goal of this effort is to create new partnershiamong scientists, science educators, teachers, and minority stdents to better convey information to a broad and diverse audence. More specifically, the CBEO prototype is being usedfacilitate and assist three different education and outreach prgrams. First, the CBEO is working to facilitate data access aanalysis for a Teacher Research Fellowship Program run throuthe Univ. of Maryland Center for Environmental SciencesMaryland Sea Grant Environmental Science Education Partneship. This program involves teachers with scientists from partninstitutions to enhance the teachers’ understanding of scienconcepts and to develop classroom applications built on thwork. Second, the CBEO is helping the development of new edcational tools for use in the NSF sponsored Center for OceScience Excellence, Mid-Atlantic. This center integrates researand education by using hypoxia in estuaries to illustrate issuesecosystem health. Finally, the CBEO is working directly withlong-standing NOAA sponsored program known as MAST �Muticultural Students At Sea Together�, which is operated throuHampton Univ. �Cuker 2003�. The MAST program involvesdiverse crew of undergraduate and graduate students in a montlong cruise of the Chesapeake aboard a 16.2 meter �53 foot� saing vessel as a means to combine the study of marine sciencpolicy, the heritage of African Americans and Native Americanand seamanship. Because MAST focuses on measuring watquality parameters such as oxygen and chlorophyll throughout tBay, the link with the CBEO is again a natural one. Through tCBEO, MAST students will be able to examine their data inbroader context and with new tools, and also contributing neresults to the overall project. By thus involving other educatoand students in the development of the CBEO prototype, tproject team hopes to gain additional insights into the particulsystem features that are most important for making the observtory useful not only for scientific research and resource managment, but also for public education and outreach.

CBEO Test Bed Development „CBEO:T…

The CBEO:T �test bed� is a platform on which to test codesynthesize data sets, develop interpolation routines, and also dploy and evaluate web services that operate within the larger oservatory networks �CUAHSI and GEON�. Locally the test bconsists of an Intel-based server that runs on the Microsoft Serv2003 operating system and has several software stacks installei.e., MS SQL Server 2005, ArcSDE Server, IIS, but also nonprprietary databases like postgreSQL. The test bed has more th2.25 Tbytes of capacity and over 1 Tbyte of data loaded at thtime.

The CBEO:T includes a database system that can be usedtest the integration of model-derived data with observatioacross disparate databases. The system allows observational dato be queried against model data by interpolating the values of tpoint sources onto the model grid so they can be compared, corelated, and otherwise analyzed. This makes the test bed an idecomputational platform for hypoxia research that, when intgrated with the GEON framework, makes it globally accessiblethe scientific community.

The data sets of the CBEO:T are constructed from select

2008

existing databases and data streams. These include archived out-s;

ke’s�;c

ve.,

es

nsnt

ern-le

e-nsed

esze

edteal

N.heri-inern

ed

il.y-eshey-

gisa-k

y-aln-2

ene-e-

ofale-

Summary

p-mx-atrenttetats

chlytsa

erorleled

-thurpecha-

onkechrs

ell

thI,

les

ken,.ay

of

on.se,

tal

n-re-

ic.”

put from Bay-scale water quality and hydrodynamic model runmonitoring data from the Chesapeake Bay Program’s ChesapeaInformation Management System �USEPA 2007�; MD DNR“Eyes on the Bay” program shallow water data �MD DNR 2005and aircraft remote sensing data �Harding et al. 1994�. Specifitasks for this team include:1. Integration of selected data sources and model data to resol

structural and semantic heterogeneity within the data, e.gdata format, units, spatial coordinate system, and ontologi�where appropriate�.

2. Development of SQL language constructs for spatial joithat resolve spatial and temporal heterogeneity among poidata, continuous data, and spatially averaged data.

• Adaptation of the middleware SQL parser and optimizdeveloped for the NVO to translate data set joins to stadard SQL operations and schedule these across multipdatabases.

• Building of a library of user-defined functions that implment interpolation functions for temporal and spatial joiexecuted at the databases, including the velocity-directinterpolation �III-C5�.

3. Provision of web-services interfaces to the test-bed databasso that identified users may invoke, manage, and visualiscientific questions using the CBEO:T across the Internet.

Chesapeake Bay Science and Management „CBEO:S…

The CBEO:S team oversees the science connection to the planncyberinfrastructure development. This science team will articulaand addresses well-posed science questions, first using the loctest bed CBEO:T and then the CBEO network node CBEO:The team’s main objective is to provide a “grounding” for tcyberinfrastructure developments, a task that previous expeences have shown to be crucial. With this approach, the domascientists take the lead in the development of the CBEO raththan relying on cyberinfrastructure experts to dictate the creatioof these types of information systems. Specific efforts are devotto the following activities:1. Examine the Hagy et al. �2004� finding �Fig. 3� in deta

Using the CBEO to explore anoxia development, using dnamic �velocity-directed� multivariate interpolation schemto produce O2 sources and sinks as variables. Explore tstatistical significance of the perceived shift in loading: hpoxia relationships.

2. Perform a rigorous comparison of the 15 years of modelinoutput data and the integrated observational data set. Thisin contrast to previous analyses that used only the fixed sttion data �Fig. 3�a��. Focus on calibrations of source and sinterms as well as state variables.

3. More carefully compare the predicted versus observed hpoxic volume �Fig. 3�b��. Use the integrated observationdata set to seek reasons why the model fails to describe iterannual variations. Examine miscalibration�s� under itemas potential cause�s� of problems.

4. Analyze the flux of organic matter and dissolved oxygfrom the shallow water regions of the Bay using the intgrated data set and the hydrodynamic model transport �vlocity and diffusion� data.

5. Use the integrated data set to make improved estimatesprimary production �Cerco and Noel 2004�, a fundamentcomponent of the models. Compare results with the intgrated model-derived data.

JOURN

The major goal of the CBEO project is to demonstrate that proerly developed and deployed cyberinfrastructure can transforscience by allowing researchers and managers to use data to eplore questions in totally new ways. For example, it appears thmany of the fundamental processes governing coastal hypoxia anot yet fully understood or adequately incorporated into curremodels for the Bay. The CBEO prototype aims to demonstrathat improved access and use of existing observational data, dastreams, and modeling results can provide important new insighinto these issues, and also facilitate future initiatives of researand education. We believe that this prototype is most effectivedeveloped by an approach that allows environmental scientisand engineers to explore science and education questions at“test-bed” level, and also supports parallel activities by computscientists to develop the necessary cyberinfrastructure network fbetter shared use at regional and national scales. Such paraldevelopment in connected activities allows each group to procewithout “waiting” for developments by the other but, more important, it also allows the development of interim results by bogroups that can be used to inform the overall process. It is ohope that the approaches and tools developed in this prototywill also benefit a wide variety of other environmental researsystems where improved integration of disparate kinds of mesurements and modeling results are needed.

Acknowledgments

The CBEO project is funded by the National Science Foundatithrough Grant Number 0618986. The CBEO team would also lito thank Gary Schenk from the EPA Chesapeake Bay ResearProgram and Carl Cerco from the U.S. Army Corps of Engineefor making available large numerical modeling data sets as was for their continued invaluable input to this project.

References

Baru, C. �2004�. “The GEON grid software architecture.” Proc., 24Annual ESRI Int. User Conf., Aug. 9–13, 2004, San Diego, ESRRedlands, Calif.

Bermudez, L., and Piasecki, M. �2006�. “Community metadata profifor the semantic web.” Geoinformatica J., 10 �March�, 159–176.

Boesch, D. F., Brinsfield, R. B., and Magnien, R. E. �2001�. “ChesapeaBay eutrophication: Scientific understanding, ecosystem restoratioand challenges for agriculture.” J. Environ. Qual., 30�2�, 303–320

Cerco, C. F. �1995�. “Simulation of long-term trends in Chesapeake Beutrophication.” J. Environ. Eng., 121�4�, 298–310.

Cerco, C. F., and Cole, T. �1993�. “3-dimensional eutrophication modelChesapeake Bay.” J. Environ. Eng., 119�6�, 1006–1025.

Cerco, C. F., and Noel, M. R. �2004�. “Process-based primary productimodeling in Chesapeake Bay.” Mar. Ecol.: Prog. Ser., 282, 45–58

Chamberlin, D. �1998�. A complete guide to DB2 universal databaMorgan Kaufmann, San Francisco.

Cloern, J. E. �2001�. “Our evolving conceptual model of the coaseutrophication problem.” Mar. Ecol.: Prog. Ser., 210, 223–253.

Collaborative Large-Scale Engineering Analysis Network for Enviromental Research �CLEANER�. �2007�. “Water and environmentalsearch systems network.” �www.watersnet.org� �Apr. 2007�.

Consortium of Universities for the Advancement of HydrologSciences, Inc. �CUAHSI�. �2007�. “Hydrologic information system�www.cuahsi.org/his� �Apr. 2007�.


Cooper, S. R., and Brush, G. S. �1991�. “Long-term history of Chesa-

w.ke

ons-p.

ef.

er-a-

n.

b.

an.”

/

innt

ofte

vi-

Stic

nd

ta-g/

H.of

ticnd

ri-3,

R.in

ry

ndticof

er/

ic7.ngta

kyon

Malone, T. C., Boynton, W., Horton, T., and Stevenson, C. �1993�.ngd-

8�.o-8,

he

�.of

ss-m.

ngrk,

n,1,

ans-

w.

T:t.”

cer.”

a-ss

ot-

ra-n.

heone-

or

w.

w.

ofE

orn.ris

ssra-

c-t.,

peake Bay anoxia.” Science, 254�5024�, 992–996.CRC. �2008�. “Chesapeake community modeling program.” �www

chesapeake.or/ChesCommModel.html� �July 2008�, ChesapeaResearch Consortium �CRC�, Edgewater, Md.

Cronin, W. B., and Pritchard, E. W. �1975�. “Additional statisticsthe dimensions of the Chesapeake Bay and its tributaries: Crossection widths and segment volumes per meter depth.” Special ReNo. 42, Chesapeake Bay Institute, The Johns Hopkins Univ., R75-3, Baltimore.

Cuker, B. E. �2003�. “Minorities at sea together �MAST�: A model intdisciplinary program for minority college students.” Current, J. Mrine Education, 18, 45–51.

Diaz, R. J. �2001�. “Overview of hypoxia around the world.” J. EnviroQual., 30�2�, 275–281.

Ecological Metadata Language �EML�. �2003�. �http://knecoinformatics.org/software/eml/� �July 2008�.

Federal Geographic Data Committee �FGDC�. �2007�. “North Americprofile �NAP� of the ISO 19115 metadata standard�http://www.fgdc.gov/standards/projects/incits-l1-standards-projectsNAP-Metadata� �Oct. 2007�.

Hagy, J. D., Boynton, W. R., and Keefe, C. W. �2004�. “HypoxiaChesapeake Bay, 1950–2001: Long-term change in relation to nutrieloading and river flow.” Estuaries, 27�4�, 634–658.

Harding, L. W., Itseire, E. C., and Esaias, W. E. �1994�. “Estimatesphytoplankton biomass in the Chesapeake Bay from aircraft remosensing of chlorophyll concentrations, 1989–92.” Remote Sens. Enron., 49�1�, 41–56.

Harding, L. W., Magnuson, A., and Mallonee, M. E. �2005�. “SeaWiFretrievals of chlorophyll in Chesapeake Bay and the mid-Atlanbight.” Estuarine Coastal Shelf Sci., 62�1–2�, 75–94.

Howarth, R., et al. �2000�. “Nutrient pollution of coastal rivers, bays, aseas.” Issues Ecol., 7, 1–15.

International Standards Organization �ISO�. �2007�. “Geographic medata 19115:2003 Technical Committee 211.” �http://www.iso.oriso/iso_catalogue/catalogue_tc/catalogue_detail.htm?csnumber�26020� �Oct. 2007�.

Johnson, B. H., Kim, K. W., Heath, R. E., Hshieh, B. B., and Butler,L. �1993�. “Validation of a three-dimensional hydrodynamic modelChesapeake Bay.” J. Hydraul. Eng., 119�1�, 2–20.

Kemp, W. M., et al. �2004� “Habitat requirements for submerged aquavegetation in Chesapeake Bay: Water quality, light regime, aphysical-chemical factors.” Estuaries, 27�3�, 363–377.

Kemp, W. M., et al. �2005�. “Eutrophication of Chesapeake Bay: Histocal trends and ecological interactions.” Mar. Ecol.: Prog. Ser., 301–29.

Kemp, W. M., Smith, E. M., Marvin-DiPasquale, M., and Boynton, W.�1997� “Organic carbon-balance and net ecosystem metabolismChesapeake Bay.” Mar. Ecol.: Prog. Ser., 150, 229–248.

LEADPortal. �2008�. Linked environments for atmospheric discove�LEAD�, �http://lead.ou.edu/� �Jan. 2006�.

Ludäscher, B., Lin, K., Jaeger-Frank, E., Bowers, S., Altinas, I., aBaru, C. �2006�. “Enhancing the GEON cyberinfrastructure: Semanregistration, discovery, mediation, and workflows.” Internal Reportthe Geosciences Network, �http://daks.ucdavis.edu/~ludaesch/Papgeon-sms.pdf� �July 2008�.

Maidment, D. R., Zaslavsky, I., and Horsburgh, J. �2006�. “Hydrologdata access using web services.” Southwest Hydrology, 5�3�, 16–1

Malik, T., Burns, R., and Chaudary, A. �2005�. “Bypass caching: Makiscientific databases good network citizens.” Proc., Int. Conf. on DaEngineering.

Malik, T., Szalay, A. S., Budavri, A. S., and Thakar, A. R. �2003�. “SQuery. A webservice approach to federate databases.” Proc., Conf.Innovative Data Systems Research.


“Nutrient loading to surface waters: Chesapeake case study.” Keepipace with science and engineering, M. F. Uman, ed., National Acaemy Press, Washington, D.C., 8–38.

Malone, T. C., Crocker, L. H., Pike, S. E., and Wendler, B. W. �198“Influences of river flow on the dynamics of phytoplankton prduction in a partially stratified estuary.” Mar. Ecol.: Prog. Ser., 4235–249.

Maryland Dept. of Natural Resources �MD DNR�. �2005�. “Eyes on tbay.” �www.eyesonthebay.net� �Jan. 15, 2006�.

myGEON. �2006�. �http://www.geongrid.org/mygeon.html� �Jan. 2006National Science Foundation �NSF�. �2005�. “Cyberinfrastructure

environmental observatories: Prototype systems to address crocutting needs �CEO:P�.” �http://www.nsf.gov/funding/pgm_sumjsp?pims_id�13517�.

NEON. �2007�. “Transforming ecological research, and forecastiecological change.” National Ecological Observatory Netwo�http://www.neoninc.org� �Sept. 2007�.

Nixon, S. W. �1995�. “Coastal marine eutrophication: A definitiosocial causes, and future concerns.” Ophelia. Int. J. Marine Biol., 4199–219.

OCEAN.us. �2005�. “Data management and communications pl�DMAC� for research and operational integrated ocean observing sytems �IOOS�.” �http://dmac.ocean.us/dacsc/imp_plan.jsp�.

OOI. �2008�. Oceans observatories initiative (OOI), �http://wwoceanleadership.org/ocean_observing� �July 2008�.

Pacific Marine Environmental Laboratory �PMEL�. �2005�. “FERREAn interactive computer visualization and analysis environmen�http://ferret.pmel.noaa.gov/Ferret/� �Jan. 2006�.

Rabalais, N. N., Turner, R. E., and Scavia, D. �2002�. “Beyond scieninto policy: Gulf of Mexico hypoxia and the Mississippi RiveBioScience, 52�2�, 129–142.

Roman, M., Zhang, X., McGilliard, C., and Boicourt, W. �2005�. “Sesonal and annual variability in the spatial patterns of plankton biomain Chesapeake Bay.” Limnol. Oceanogr., 50�2�, 480–492.

Rosenberg, R. �1990�. “Negative oxygen trends in Swedish coastal btom waters.” Mar. Pollution Bull., 21�7�, 335–339.

SDSC. �2006�. “The Geosciences network, GEON: Building cyberinfstructure for the GeoSciences.” �http://www.geongrid.org/� �Ja2006�.

Szalay, A. S., et al. �2002�. “The SDSS Skyserver: Public access to tSloan Digital Sky Server data.” Proc., ACM SIGMOD Int. Conf.Management of Data, Association for Computing Machinery’s Spcial Interest Group on Management of Data.

Ueng, S.-K, and Wang, S.-C. �2005�. “Interpolation and visualization fadvected scalar fields.” Visualization, IEEE, Piscataway, N.J.

UNIDATA. �2008�. “Integrated data viewer �IDV�.” �http://wwunidata.ucar.edu/software/metapps/� �July 2008�.

USEPA. �2007�. “Chesapeake Bay program CIMS.” �http://wwchesapeakebay.net� �Nov. 26, 2007�.

Yang, Q., and Parvin, B. �2003�. “High-resolution reconstructionsparse data from dense low-resolution spatio-temporal data.” IEETrans. Image Process., 12�6�, 671–677.

Zaslavsky, I., et al. �2005a�. “Grid-enabled mediation services fgeospatial information.” Next generation geospatial informatioFrom digital image analysis to spatiotemporal databases, P. Agouand A. Croitoru, eds., Taylor & Francis, London, 15–24.

Zaslavsky, I., Memon, A., and Memon, G. �2005b�. “Integration acroheterogeneous spatial data and applications within a large cyberinfstructure project.” GIS Devel., 9�5�, 28–31.

Zhang, Y., and Akambhamettu, C. �2003�. “On 3-D scene flow and struture recovery from multiview image sequences.” IEEE Trans. SysMan, Cybern., Part B: Cybern., 33�4�, 592–606.

2008

Prototype System for Multidisciplinary Shared Cyberinfrastructure:...

Documents

Transcript of Prototype System for Multidisciplinary Shared Cyberinfrastructure:...