institute for sustainable earth and environmental software isees

Institute for Sustainable Earth and Environmental Software

ISEES

Matthew B. Jones

National Center for Ecological Analysis and Synthesis (NCEAS)

University of California Santa Barbara

ISEES Sustainability and Adoption WorkshopSeptember 10-11, 2013

Introductions

• Your 140 character twitter intro

Science and Synthesis

• Synthesis critical to advancing science• Merger of synthesis with experimental and

observational science

Ocean Health Index (OHI)O

cean H

ealt

h Ind

ex

Halp

ern

et

al. 2

01

2

Software in the science lifecycle

From Reichman, Jones, and Schildhauer; doi:10.1126/science.1197962

Software for the Earth, Life, and Environmental Sciences

• Statistical analysis– e.g., R, SAS, Matlab, Systat, Excel, etc.

• One-off models (by students, faculty, etc.)

• Custom analytics (e.g., Primer, MetaWin, MaxEnt)

• Modeling frameworks (e.g., ROMS)

• Community models (e.g., Century, Community Climate Model)

• Workflows (Kepler, VisTrails, …)

• Computing engines (e.g., Sun Grid Engine, Amazon ECS)

• Data management (DataONE, Metacat, DataUp)

• Service computing (Blast, WMS, WFS, …)

Software challenges

• Wide range of software types• Code Complexity and Quality• Reproducibility• Systems integration• Development and maintenance are labor

intensive– NSF not set up for infrastructure/maintenance

• Software lifetime long compared to hardware• Under-appreciated value

ISEES Vision

• Massively accelerate science– (Earth, environmental, and life science)

– Enable collaboration and integration across disciplines

– Invent, develop, integrate, mature, and sustain software• used throughout the scientific lifecycle

Determining needs

• What needs to be improved?

• What challenges do we face?

• How do we solve these?

Any solution must…

• Provide value to participants in their reputation economy

• Enable participants, not compete with them

Can an Institute build it “for them”?

• No. Must empower community– Scaling/leverage– Creativity– Knowledge of domain

• Community driven initiative– Model after synthesis centers– Link to community initiatives such as ESIP

ISEES Steering Committee

• Matthew Jones (Cyberinfrastructure)

• Lee Allison (Geology)

• Daniel Ames (Hydrology)

• Bruce Caron (Collaboration)

• Scott Collins (Ecology)

• Patricia Cruse (Library)

• Peter Fox (CI & Semantics)

• Stephanie Hampton (Ecology)

• Chris Mattmann (JPL; Apache)

• Carol Meyer (ESIP Community)

• William Michener (DataONE)

• James Regetz (Analytics)

• Mark Schildhauer (Semantics)

Strategic planning approach

Two tracks

Stakeholders and community

Community Engagement, Sustainability, and Governance

Stakeholders andcommunity

Sustainability

Structure and Governance

Workforce Development

Knowledge and skills: content gaps

Mechanisms for education

StrategicRecommendations

Collaborative Space

• Document sharing and wiki– https://projects.nceas.ucsb.edu/isees/projects/soft

ware/

• Etherpad collaborative editing– https://epad.nceas.ucsb.edu/

• Username/pw– Username: isees– Password: swforscience

ISEES Science Drivers Workshop

• Outcomes– Science challenges limited by software– Functional areas for ISEES

Burrows et al. 2011. Science 334:652-655

Fresh water availability

Ecosystems

Human society

Water dimension

Biological dimension

Human dimension

MeanExtremesUncertainty

MeanExtremesUncertainty

MeanExtremesUncertainty

Time

Allo

catio

n

Now

Visualization

Scenarioprescription

Data resources• CUAHSI HIS• World water

online• GEOSS

• DataONE• NASA/ESA/other• NEON• EarthCube• NSW/WMO/other

• CoCoRaHS• Water managers• Army Corps• Social media

Data types• Precipitation• Atmos. H2O• Groundwater• Reservoir storage• River discharge

• Water quality• Soil moisture• Other climate• LC/LU• Built

infrastructure

• Economic• Population• Ag/irrigation• Sap flux/tower ET• Human use• Physical hydrology

New

dat

a in

itiati

ves

Data management• Selection• Provenance• Rectification

Scenario support• Simulation• Historical• Social science

Earth system models (CSDMS)• CESM• ESMICs

Data fusion• Spatial

statistics• Assimilation

Data ingestion

ExperimentationFeedback analysis

Community input and refinement

TheoryAlgorithmsParameterizations

How will coupled human and biophysical systems shape and be shaped by water availability?

Sources

Transport

Recipient Systems

Resistor

OutputVisualization

ScenariosDecision-Support tool

Clim

ate

chan

ge (m

odel

out

put)

Hyd

rolo

gica

l mod

ifica

tions

Popu

latio

n ch

ange

(sce

nario

s)La

nd u

se a

nd c

over

cha

nge

(mod

els,

obs

erva

tions

)

Archive, provenance, other considerations

Q: What are the controls, impacts, and societal responses to atmosphere–land–water transfer of pollutants, and how will they change under multiple, global-change stressors?

• Modularity: main program with modules (off/on in parameter file)• Flexible I/O:

• OPeNDAP (Open-source Project for Network Data Access protocol); •Storage: flexible output (netcdf, ASCII formats) and data archive system

• Existing pollutant transport models • CMAQ annual deposition Community Modeling and Analysis System (CMAS) Center

• SPARROW water quality model USGS• NASA models of aerosol movement• SMS and Delta3D for sediment transports• CMS for CDOM transport and oil-spills

• Landscape and habitat models (USGS, WRI)

Software Needs for Data & Model Output Synthesis

Perturbations of IC (climate and land-use scenarios)• New transport models: Coupled atmospheric-ocean transport models

- High Performance Computing with multi-processors and MPI capabilities - multi-scale nesting capabilities - hind-cast and near-real time capabilities- stochastic capabilities & ensemble simulations to formulate uncertainties

Output & Visualization Needs•user interface, interactive scenarios •connectivity module linking sources to recipients: where the pollution comes from?•Matlab 2 & 3D animations•R - statistical package

Spatially and temporally predict carbon storage &

flux globally at 1km scales to 2300

What can ISEES do for you?

• Computation training for early career and mid and senior scientists (14)

• Assimilation and QA/QC tools for heterogeneous data (13)• Provide a collaborative environment for ecologists, computing

scientists, social scientists, etc. (10)• Develop dynamic, flexible visualization tools (9)• Support for software maintenance and sustainability,

including software building blocks (e.g., modules) (9)

What can ISEES do for you?

• Improved tools for capturing decisions and workflows in collaborative research projects (6)

• Software discovery: One-stop shopping for finding and characterizing software and models -- focus on users (6)

• Provide consultants, collaborators for software, CS, for researchers (6)

• Community hub for standards convergence (4)• Facilitate merging of disparate software tools (3)• Develop user-friendly interfaces to existing models (3)• Provide a framework for multiscale, coupled modeling

systems (2)

What can ISEES do for you?

• Make high performance computing available to the average ecologist and environmental scientist (2)

• Software to help with uncertainty and error propagation in spatial models (2)

• Provide web-based software services, i.e. ability to run analyses on ISEES servers via accessible interfaces (2)

• Software vetting (check software being developed in-house) (1)

• Help me contribute to community software (1)• Taxonomy scrubbing software (1)• Improved model intercomparison (1)

Software Lifecycle and Components Workshop

• Goal: Envision a model for ISEES that enables efficient, reproducible, scalable, and impactful environmental science– Identify *functions* that ISEES would be ideally suited to

perform or coordinate– Provide recommendations to the ISEES steering committee

for our strategic plan

• Contribute to a paper outlining this vision for ISEES• Stimulate amazing and fun discussions here and later

about software in science

ISEES Software lifecycle model

Figure by M. B. Jones, NCEAS

Functional areas for ISEES

• Participants identified 4 functions

– Community building– Training and Advocacy– Consulting Services– Infrastructural Services

Training and Advocacy

• Advocate for the benefit of Open Source• On-site and remote courses on sw

development for research sw developers• Develop the policies that incentivize sharing

and collaboration (licensing, attribution)• Provide best practices for sw development

aimed at scientific community

Community Building

• Operate community based support services for software

• SW dating game -- match science with sw experts -- e.g. dba services, ux services

• Support collaboration groups focus on sw projects

Consulting Services

• Assisting groups in software:– Design and Architecture, Hardening, Maintenance,

Preservation

• Consultation and mentoring services– licensing, testing as a service, engineering, cost

modeling

• Certify papers as reproducible

Infrastructural services

• Software discovery services• Software review and certification program• Provide software use and quality metrics• Actively survey the community

Science & Software collaboration

• Science-directed hackathon-style working groups– 8-12 member working groups– co-directed by scientist and developer – address one or more software problems that

impede a grand challenge science question

• Advisory board equally comprised of scientists/developers to select winning proposals

• Option for 1-2 developers to continue on problem between working groups

Questions?

• http://isees.nceas.ucsb.edu/

• http://www.nceas.ucsb.edu/ecoinfo/