![Hier klicken, um den Treffer aus der Auswahl zu entfernen](images/unchecked.gif) |
Titel |
Modelling Climate Change Effects on Species Distribution using a Cloud Computing Infrastructure |
VerfasserIn |
Valerio Angelini, Paolo Mazzetti, Michele Pierri |
Konferenz |
EGU General Assembly 2011
|
Medientyp |
Artikel
|
Sprache |
Englisch
|
Digitales Dokument |
PDF |
Erschienen |
In: GRA - Volume 13 (2011) |
Datensatznummer |
250048607
|
|
|
|
Zusammenfassung |
Climate change threatens to commit 15-37% of species to extinction by 2050 accelerating a
mass extinction precipitated by widespread land use changes. The need to assess these
impacts and recommend solutions to policy-makers is correspondingly acute and has been
highlighted by the Fourth Assessment Report of the Intergovernmental Panel for Climate
Change (IPCC, 2007) [1]. Such analyses require robust infrastructures capable of
integrating huge volumes of data from biodiversity archives, satellite remote sensing,
and climate change data. The main requirements for such infrastructures are: a)
the support of multidisciplinary interoperability in order to harmonize services
interfaces and data models; and b) high scalability to enable running of complex models
with different climate change and species distribution scenarios. The first issue is
currently addressed by several initiatives aiming to provide advanced infrastructures for
geospatial and Earth Science resource sharing, such as projects funded under the 7th
Framework Programme, standardization activities from international bodies like ISO and
OGC, European and global initiatives like INSPIRE, GMES, GEOSS. The high
scalability could be provided, in principle, by the Distributed Computing Infrastructures
(DCIs) designed and developed in the last decade. The integration of geospatial
infrastructures and DCIs has been investigated by several initiatives. For example,
recently, the implementation of standard OGC services on top of different Grid
middlewares has been analyzed in the context of several initiatives such as FP6/FP7
Projects (e.g. CYCLOPS, DEGREE, ENVIROGRIDS, etc.), standardization bodies
activities (e.g. OGF-OGC MoU, OWS Phase 6) and international working groups (e.g.
G-OWS).
We present an experimentation done integrating the architectural solution proposed by the
Climate Change and Biodiversity Working Groups of the GEOSS Architecture
Implementation Pilot Phase 2 and Phase 3 (AIP-2, AIP-3) [2], with a Cloud Computing
Infrastructure. In this experimentation the system is based on open standards hiding all the
complexity of the distributed computation to the user. The Ecological Niche Model (ENM)
processing implemented using OpenModeller [3], an open-source niche modelling
project, is exposed through an OGC WPS 1.0 standard interface. The process is able
to retrieve the requested input datasets from external sources publishing standard
interfaces, such as OGC WCS 1.1 for climate change environmental layers and GBIF
access service for species occurrences. The WPS server we implemented makes
use of a batch system to launch and control the execution of the required jobs on
a remote pool of computing nodes. The computing nodes run on a mainstream
Infrastructure-as-a-Service (IaaS) Cloud Computing platform (Amazon EC2/S3).
An additional software component is able to re-size the working nodes pool by
dynamically instantiating or destroying the dedicated computing instances on the
Cloud infrastructure. The desired size of the pool is constantly balanced basing
on the length of the batch system job queue. To be able to run multiple instances
of the ENM process we developed an additional WPS that, accepting ranges of
different parameters launches multiple instances of the model by computing all
the required combinations of the parameter values. Finally, we developed a light
AJAX GUI to launch and follow the execution workflow and then visualise the
results.
Tests performed on the developed prototype demonstrated the possibility to run ENMs
calculations and projections maintaining the execution time almost independent of spatial
resolution, spatial coverage, number of scenarios, etc. exploiting the capabilities of
on-demand allocation of computing power and storage space.
In conclusion, the solution proposed in our experimentation presents various
advantages:
Interoperability, through the adoption of open standards for resource sharing.
Scalability, through the adoption of IaaS solutions, enabling the typical scientific
use cases with possible peaks of intense computing resources utilisation.
Cost-effectiveness, since for the entire prototype development and the
experimentation we paid approximately 50-¬Â for about 700 CPU hours.
References
[1] S. Nativi, P.Mazzetti, H. Saarenmaa, J. Kerr, É. Ó Tuama, “Biodiversity and climate
change use scenarios framework for the GEOSS interoperability pilot process”, Ecological
Informatics 4 (2009) 23-33.
[2] GEOSS Architecture Implementation Pilot, available at http://www.ogcnetwork.net/AIpilot
[3] OpenModeller Home Page, available at http://openmodeller.sourceforge.net/ |
|
|
|
|
|