1. Scientific Workflows for the Sensor Web ICT for Earth Observation Anwar Vahed

2. www.meraka.org.za September 2008 / 2 Topics The Sensor Web Scientific workflows Managing knowledge: The need for SW4SW SW4SW issues  Representation & structure  Frameworks & tools Open questions

3. www.meraka.org.za September 2008 / 3 The Sensor Web vision Sensor Web Middleware

4. www.meraka.org.za September 2008 / 4 The promise of the Sensor Web Data democracy: Improved ease of (controlled) access to sensor data resources Leverage sensor assets and integrate EO data with domain data Provide near real-time EO information, Alerts for dynamic (transcient) phenomena But will researchers use it?

5. www.meraka.org.za September 2008 / 5 Scientific Workflows “Traditional” workflows:  Stateful and acyclic  Record and share knowledge Scientific workflows: (A cyber research environment)  Stateless and cyclic  Generate knowledge APIs and toolkits  Kepler, Taverna, Gridbus,...  Bioinformatics, Cheminformatics Scientists & workflows Record & share Learn Build & refine Run

6. www.meraka.org.za September 2008 / 6 The Sensor Web Case Bring the sensor web to the scientist: The Empirical Sensor Web: Open, distributed environment for scientific exploration, knowledge generation and management An enabling environment that focus on science questions, not data plumbing Simplify complex processes for the domain scientist, Automate repetitive tasks Knowledge management: Public/private spaces for knowledge storing, sharing Adaptive architectures: Processing/Service & Data Grids, self-organisation, learning workflows

7. www.meraka.org.za September 2008 / 7 A Scenario

8. www.meraka.org.za September 2008 / 8 Cholera research case Research team of:  Microbiologists, Computer scientists, Statisticians, Mathematical modellers, Remote sensing & GIS experts, Oceanographers, Biochemists, Climatologists, Water & environmental ecologists Wants to access data of:  Daily meteorological data (temp, rainfall and relative humidity) from weather stations, satellite, sea surface temperature, in- situ water temp, salinity, pH, nitrogen,… for sampling points And use it in:  Statistical software, signal processing software, Mathematical and computational intelligence software To undertake:  Analysis, Data Mining, Monitoring, Modelling and simulation To understand cholera better

9. www.meraka.org.za September 2008 / 9 Cholera research requirements Integrated “debugging” environment (IDE) to construct, automate/run/trace, reuse, nest, parameterised workflows Services and resources to support:  (Standardized) access to sensory data of various forms across internet  Tasking of sensors, steering of observations  Data provenance  Continuous monitoring and increased preparedness  Preprocessing and visualisation  Integration with non-EO data These are not very different for other domains!

10. www.meraka.org.za September 2008 / 10 Questions (1/2) How can we understand the needs of scientists when working with Sensor Web resources, and how can these be enabled by scientific workflows? What are appropriate formal representations for scientific workflows for the purposes of their machine and human manipulation in the Sensor Web? How can data and service (“soft”) resources be discovered and acquired for utilisation in scientific workflows? How can computational and communication (“hard”) resources be discovered and leveraged during execution of a scientific workflow?

11. www.meraka.org.za September 2008 / 11 Questions (2/2) To what extent can the quality of existing soft and hard resources be determined and how can this information be used to modify a workflow (a) before execution, (i.e., during composition), during execution (dynamically), and after execution (workflow refinement)? How can scientific workflows be managed (stored, shared and manipulated) to permit flexible levels of security (e.g., read, modify, execute) and accessibility (e.g., individual, private, group, community, public spaces) in the Sensor Web? How can scientific workflows be executed robustly in a Sensor Web environment, where resource access and quality are uncertain?

12. www.meraka.org.za September 2008 / 12 Inputs?