Interoperability among Earth Observations and Earth Science Models

1. Introduction

• EO data are very diverse, and ESMs in different Earth science domains are various

• Issues related to use EO data in ESMs

  • data collections

  • incompatible in format, projection, and resolutions

  • needed data don’t exist

• Technologies

  • Interoperability standards and specifications

  • Standard-based data access technologies

  • Web processing services

  • Geospatial processing models, workflow, production visualization

  • Workflow engine -> automatic processing, run models

2. OGC Activities and Standards for Model-Sensor interoperability and the Model Web

• Model web <-> Sensor web

  • Sensor web interoperability is based on open standards.

  • Model Interoperability- an evolution: OpenMI: dedicated tools, technological framework, Model-as-a-Service

• Interfaces simplify Model interactions

  • schedule model runs

  • source data and parameters are flexible and interactive

  • model output

• OpenMI -> data exchange, improve ability to model complex scenarios

• WPS for Model Access: Geoprocessing workflow

• Examples: Simulated flood event(WPS, OpenMI)

• Information Schemas for Models: for data format issues

• Questions:

  • OGC standards in Cloud-> there are already some works

3. Needs for Earth Observational Data from Coastal Oceanographic Modeling Perspective

• The coastal ocean is inter-disciplinary marine system: land, sea, atmosphere, underlying sediments…….

• Issues:

  • Irregular geometry

  • Intertidal wetlands

  • steep topography

  • conservation properties

• Three examples show the necessity to get the data driving the models

  • data integration

  • resolution

  • missing data -> data estimation

  • validation

• Current barriers in accessing Earth observation

  • data are not enough for modelers

• Questions:

  • Is there any collaborations between OGC and domain scientists?

    • Yes, some examples are given.

3. Realize effortlessly feeding FVCOM and CRM with multisource

• EO data are important for models, but there are huge observation data

• Levels of data processing (data transformation and provenance): level_1, 2,3, …, model input, model output (products)

• 60% of time to prepare data for models

• Modelers need: model-specific form as input; their own data format ….., but there are big gap between modelers and data

  • exactly matching; different dataform; scale issues

• Solutions : fine-grained existing interoperable web services (CSW, WMS,WCS,.....); workflow technology; gear up the related services in workflow; execute workflows automatically

• Framework: workflow combine models, data preprocessing and data inputs together.

• Examples: web-portal for order submission, workflow services in CyberConnector for data processing, Interactive standards

• Questions:

  • Diversity of models, how to provide the needed models if the models are new?

    • The services are abstracted, and they can be organized to generate a new complex workflow as a new model.

4. Geospatial workflows for reusing standard Web services

• BPEL as the base workflow language

• Event-triggering mechanism

  • Petri net: wait until all are ready

• Workflow types: sequence, loop, Parallel, multi-choice

• BPELPOwer: web-based, support multiple protocols; chain and manage workflow services

• Geospatial Workflows: data-intensive, location-sensitive, data customization

• Special examples: virtual workflow to help find the interesting data from the existing models

• Enterprise Service bus (Apache Synapse); quality and provenance; move programs -> security issues.

• Question: How to control and evaluate the errors in the workflow?

  • Do not have answers, but it is really a good questions.