Comments on: Interactions Between Data, Observations, and Modeling

1
Comments on Interactions Between Data,
Observations, and Modeling
Sydney Levitus, NODC/NOAA CCSP Workshop Washington
, D.C. December 5, 2002
Panel on Grand Challenges in Modeling,
Observations, and Information Systems
2
What is the strength of the CCSP Strategic Plan?

• Recognition that in order to improve the climate
  record we need to
• extend the climate record back in time using
  paleoclimatic data,
• extend the climate record back in time by
  rescuing historical instrumental data,
• extend the climate record forward in time by
  the improvement of existing observation systems
  and by developing new climate system observing
  systems,
• develop biological and ecological data
  databases,
• promote international cooperation,
• provide open access to data,
• provide for adequate data management.
• This is not a motherhood statement. The climate
  system is complex and has
• biological and chemical components as well as
  physical components.

3
Uses of climate system instrumental databases

• Some specific uses for climate system
  instrumental databases and products derived from
  such databases
• 1) Diagnostic studies describing the earth's
  climate system
• 2) Boundary and Initial conditions for numerical
  models
• Data assimilation studies
• 4) Verification for climate simulation modeling
  runs
• 5) Ground or Sea Truth" for satellite
  measurements"
• 6) Establishing climatologies for for
  paleoclimatic studies (e.g., fields of
  temperature and salinity for CLIMAP).

4
Climate system modeling and diagnostic
requirements

• Global, integrated, comprehensive, scientifically
  quality-controlled, well-documented environmental
  databases in one format with electronic access.
• Gridded global products based on well-documented
  environmental databases with all analysis
  procedures documented.

5
Desirable Database Characteristics for Modelers
and Climate System Diagnosticians (1)

• Global- The phenomenon we are trying to
  understand are global in nature
• Integrated- Users want one-stop shopping. As
  many relevant variables as possible should
  be in the database
• Comprehensive- Users want the most complete
  databases possible, both in time and space

6
Desirable Database Characteristics for Modelers
and Climate System Diagnosticians (2)

• Scientifically quality-controlled-
• Users may not have time to quality control the
  databases they use. This task requires the
  efforts of a dedicated team of scientists
• Well-documented-
• Users must have complete documentation of the
  database they are using
• One format-
• Users do not want to deal with multiple formats
  if they can help it
• Electronic access to database and documentation.

7
Four examples of frequently used climate-system
databases

• Comprehensive Ocean-Atmosphere Data Sets
• Surface marine meteorological data, mainly from
  merchant ships. Heterogeneous.
• 2) International Satellite Cloud Climatology
  Project
• Satellite cloud data. Homogenous
• 3) World Ocean Database (e.g., 1994, 1998 ,2001)
• Ocean profile-plankton data from research
  ships, ship-of-opportunity
  programs, navy ships, buoys,
• Heterogeneous.
• 4) I.G.Y. ocean profile data
• Ocean profile-plankton data from research
  ships.
• Heterogeneous

8
World Ocean Database 2001 An example of an
integrated, heterogenous database

• temperature earths heat balance
• salinity earths freshwater balance
• oxygen biogeochemical cycles
• nutrients biogeochemical cycles, heat balance
• chlorophyll biogeochemical cycles, heat
  balance
• plankton biogeochemical cycles, heat balance.

9
World Ocean Database 2001(a heterogeneous
database)

• Data from 55,897 cruises
• Data from 3057 ships and other platforms
• Data from 489 institutes
• Data from 112 countries.

10
Utility of NODC Profile Data Based on Scientific
Citations
based on a search of the Scientific Citation
Index as of December 2001
11
OSD cast data acquired through the GODAR Project
for 1900 1991compared to NODC archive holdings
as of 1991
12
Desired characteristics of gridded products

• Similar to databases
• 1) Users want global products based on
  well-documented databases.
• As climate-system models achieve higher spatial
  resolutions the modeling community requires
• gridded products of higher space and time
  resolution.
• 3) Gridding procedures must be completely
  documented.

13
Effect of horizontal resolution ( 1 versus ¼
grid)
(Features with wavelengths less than 5?x are
greatly reduced in amplitude in each figure)
Temperature at z 200m depth
1 grid
¼ grid
14
What does it take to build the global databases
and gridded products required by the climate
modeling and diagnostic communities?
As climate system databases increase in size more
capable computer hardware and software are
important for database development and
access but just as important are human resources
and sustained project support. Climate system
databases require research teams that use the
databases by producing products from the
databases. This is the final step of quality
control. Data processing (e.g., quality control)
and stewardship require scientific expertise
built up over time. Climate System Database
development needs to be viewed as long-term,
ongoing projects similar to numerical model
development.
15
International contributions to database
development

• Many global climate system databases can not be
  developed by a single country.
• International data exchange is greatly
  facilitated by
• intergovernmental data exchange systems such as
  the
• International Ocean Data and Information
  Exchange System (IODE) of the Intergovernmental
  Oceanographic Commission (IOC)
• 2) non-governmental data exchange systems such as
  
• the World Data Centers of the International
  Council of Scientific Unions (ICSU).
• These systems need to be supported through
• a) direct financial support,
• b) support for exchange of data managers and
  scientists,
• c) meeting support,
• d) capacity transfer to less developed countries.

16
Need for data management by various institutional
arrangements
1) Not all data can or should be managed by
national data centers although these centers may
act as archives for all data. 2) There is a
need for Project Data Management. 3) Data
expertise often lies with the P.I.s who made the
original measurements. 4) However, Project Data
Management needs to be adhere to the same
standards and policies as national data centers.
17
WOD01 chlorophyll data
18
WOA01 Zooplankton biomass