EcoTrends: Progress & new challenges

Report
EcoTrends: The
adventure continues
Christine Laney
IM Meeting, San Jose, CA
8/3/2007
Overview
• 2004/2005: Deb Peters (JRN, ARS) & Ariel Lugo (LUQ,
FS) dream of a book & start looking for data. EcoTrends
(formerly Trends) project was born.
• Editorial committee formed, collaborations built, funds
acquired.
• Goal: Create a platform for synthesis of long-term
ecological data from research stations across the world.
Editorial Committee:
Ariel Lugo [LUQ, USFS]
Scott Collins [SEV]
Debra Peters [JRN, ARS]
Mark Ohman [CCE]
Peter Groffman [HBR]
Christine Laney [JRN]
Tim Kratz [NTL]
Morgan Grove [BES, USFS]
Charley Driscoll [HBR]
Bob Waide [LUQ, LNO]
Charlene d’Avanzo [Hampshire College, Education Representative]
Overview
• Original request: any presummarized long term data
and/or graphs of data
• Exploration of the results (data, graphs & metadata) &
questions from correspondents pointed to a need for a
more directed request
• Specification of dataset length, themes and variables of
interest, minimal metadata, etc.
• Increasingly clear that an organizational system was
needed to handle the input.
• Needed to design a way to make these products and
their sources, easily available online -> stage 1 & 2 ->
spaghetti & linguini -> PASTA
Overview
• 2006/2007: 50
research programs,
• 5 networks
• Wide variety of data
collection, storage &
accessibility methods,
metadata practices, &
communication
systems
Overview
• Current request (in general): raw long
term data, not presummarized
• Utilize many data sources other than
participant websites.
• Target for deliverables: Early 2008
The average day in the EcoTrends cubicle
• Collect source data from websites, data managers, PIs, etc.
• Ask questions
• Generate R scripts that read in the data, aggregate it to
monthly or yearly values, with simple statistical measures,
store in standard format with names that can be parsed to
populate eml files
• Generate R scripts to plot the data (book format)
• Ask questions
• Database information about the datasets and relationships
between data files, scripts, and graphs.
• Create supporting book content (maps, graphs that condense
the information, etc)
• Ask questions
• Make graphs available for critique
• Reduce, reuse, recycle, redo & revise
• Build new collaborations
Where are we?
Wet Deposition of Calcium (mg/L)
Eastern Forests
Data imported into geodatabase, joined
with existing site table and mapped onto
a template
Average monthly air temperature
Northwestern forests
Average monthly air temperature
anomalies
Northwestern forests
Walter Climate Diagrams
Northwestern Forests: moisture deficit occurs in the middle of summer
Biotic data: challenging!
•Nested sampling structures
•Site-specific methods (plot sizes, sampling
structure, etc) and ecosystem types
• Site-specific lists of species and species
codes (& non-species codes) - metadata
must be extremely complete.
•Comments/notes sometimes more difficult
to standardize
•Methods of data derivation often not
included in metadata (e.g., conversion of
species-specific volume & cover to
biomass)
•Less long-term data available than
thought throughout the LTER system
EcoTrends website: http://www.ecotrends.info
•Initial look and
feel developed and
online (Editorial
committee, Electric
Sage, Inc.)
•LNO developing
services (PASTA,
Metacat) to make
datasets and
graphs available
The EcoTrends Socioeconomic Catalog
http://coweeta.ecology.uga.edu/trends/catalog_trends_base2.php
•Ted Gragson
(CWT), Morgan
Grove (HBR),
Chris Boone (CAP)
•23 LTER sites
(missing MCM,
MCR, PAL)
•Population, Urban
population,
employment data
•See Barrie Collins
for more
information
Sevilleta
Successful case study: Grassland
Data Integration workshops
• Integrated GDI ANPP aggregate data from 4
LTER sites (JRN, KNZ, SEV, SGS) with
contextual climate and biogeochemical data
• Using existing data & analytic framework, rapidly
explored possible relationships between
variables
– Air temperature: min, mean, max, growing season, all
year
– Precipitation: growing season, non-growing season,
year totals
– Nitrate deposition in precipitation
– A combination of these factors led to the most error
being explained in the multiple regression models to
date (Carri LeRoy)
Some challenges for the future
• Keeping track of source data:
– URL & authentication issues
• Currently manually go through authentication
procedures to obtain data – passwords,
agreements, forms, etc.
– Data file formats & content
• Data should be read easily into common statistical
packages – R, SAS, MatLab, etc.
– No characters in a column that has mostly numeric data
– Data has standardized flagging symbols, abbreviations
or phrases in separate columns, not in the value column
– Comments field should be separate from flagging column
– Etc…
Some challenges for the future
– Metadata
• Generate EML where missing
• Improve content of EML to include as much
information about the research and about the data
files
• Be able to work with other forms of metadata
• Incorporating new datasets and new sites
(same issues as above)
Coming soon – P2ERLS at http://www.p2erls.net
•Search for long
term ecological
research
programs
belonging to any
participating
organization,
network, or
agency
•Very simple set
of criteria to join –
name of program,
latitude, longitude,
web address
•Linked to
EcoTrends and
LTER networks
Thanks!
• Jornada: Deb Peters, Ken Ramsey, Randy
Carr
• All IMs and PIs from participating sites
• LNO: Mark Servilla, Duane Costa, Inigo
San Gil
• EcoTrends editorial committee
• Grassland Data Integration team
• Many others…

similar documents