Database Reference
In-Depth Information
12.5.2 Climate Modeling
The Earth System Grid (ESG) project
32
,
70
provides infrastructure to support
the next generation of climate modeling research. The ESG allows climate sci-
entists to discover and access important climate modeling datasets, including
the Parallel Climate Model (PCM), the Community Climate System Model
(CCSM), as well as datasets from the Intergovernmental Panel on Climate
Change (IPCC) 4th Assessment Report (AR4).
71
The original infrastructure for ESG included two data portals, one at the
National Center for Atmospheric Research and one at Lawrence Livermore
National Laboratory. Each of these ESG portals had an associated metadata
catalog that was implemented using a relational database. The next genera-
tion of the ESG features a federated architecture, with data nodes at many
sites around the world publishing data through several ESG data portals or
gateways
. This data publication will include the extraction and publication
of metadata attributes. The bulk of the metadata for this version of the ESG
architecture will be stored in a relational database with a different schema
from the previous generation of ESG. Metadata for research and discovery
will also be harvested into an RDF triple store from the multiple federated
ESG gateway sites; this will allow users to execute global searches on the full
ESG holdings via the triple store.
The metadata model for the latest generation of the ESG includes metadata
classes that describe a climate experiment, model, horizontal grid, standard
name, and project as well as a set of data-related classes.
72
The Experiment
class includes a specification of the input conditions of a climate model ex-
periment. The Model class describes the configuration of a numerical climate
model. The Horizontal Grid class specifies a discretization of the earth's sur-
face that is used in climate models. Standard Names describe scientific quan-
tities or parameters generated by a climate model run, such as air pressure,
atmospheric water content, direction of sea ice velocity, and so forth. A Project
is an organizational activity that generates datasets. Data objects in the ESG
metadata model have one of four types: dataset, file, variable, or aggregation.
A
dataset
is a collection of data generated by some activity, such as a project,
a simulation run, or a collection of runs.
Files
are usually in the self-describing
netCDF
73
data format. A
variable
is a data array with n dimensions, where
the array is associated with a dataset. An
aggregation
is a service that provides
a view of a dataset as a single netCDF file and that can perform statistical
summaries over variable values, such as “monthly means.”
There are several ways that ESG users search for and select datasets based
on metadata attributes.
74
In one scenario, users can search for datasets based
on metadata attributes using a simple Google-style text search over all the
metadata associated with ESG datasets. The ESG gateway also presents users
with a set of search terms and possible values, as illustrated in Figure 12.3;
these terms include the experiment, model, domain, grid, variables, temporal
frequency, and dataset. An ESG user may also access an ESG portal using
