Similar a TERN eMAST : Observations and terrestrial ecosystem models : Terrestrial Ecosystem Modelling and Scaling Infrastructure : ecosystem Modelling And Scaling infrasTructure (eMAST)
Similar a TERN eMAST : Observations and terrestrial ecosystem models : Terrestrial Ecosystem Modelling and Scaling Infrastructure : ecosystem Modelling And Scaling infrasTructure (eMAST)(20)
TERN eMAST : Observations and terrestrial ecosystem models : Terrestrial Ecosystem Modelling and Scaling Infrastructure : ecosystem Modelling And Scaling infrasTructure (eMAST)
1. ecosystem Modelling And Scaling
infrasTructure (eMAST)
Observations and terrestrial ecosystem models
Presentation by Brad Evans based on contributions by Colin
Prentice, Michael Hutchinson, Gab Abramowitz, Ben Evans,
Rhys Whitley, Daisy Duursman, Tim Pugh, Julie Pauwels
3. Research domain: Impacts of rising CO2
Thus the ecosystem modeller seeks to:
1. Understand the effects of CO2 increases on
ecosystems
2. Quantify negative feedbacks – the impact of
rising CO2, land surface warming and
extreme events on ecosystems
6CO2 + 6H20 C6H12O6 + 6O2
light energy
chlorophyll +
nutrients
4. IPCC Consensus: CO2 Fertilization
WUE
NPP
WUE =
GPP
ET
NPP = GPP - R
N & P
Land Surface Models
-> Coupled to Climate Models
Other approaches
5. Observations , models and policy…
(1) MORE
Observations
(2) BETTER
models are
developed
(3) Models
evaluated
against
observations
(4) EVEN
BETTER
Models
(5) BETTER
Policy
A viscous cycle
Synthesis
6. Unifying principles for ecosystem modellers
# 1: Observations, Models and Understanding:
Integration of empirical science and modelling
betters scientific understanding.
# 2: Transparency, Evaluation, Confidence :
Reproducible models, evaluated with observations,
enhance model efficacy.
# 3: Innovation, Standards, Simplicity: Continuous
innovation, use standards, mitigate unnecessary
complexity.
7. eMAST Observations and Models
Models
OzFlux
CO2 and water fluxes
Plot Networks
Vegetation Observations
via AeKos and Others
AusCover
Remote Sensing –
Satellite, in-situ & Obs.
Bureau of
Meteorology and
Geoscience
Australia
Land
Surface
Models
Soils
Properties of soil
dap.nci.org.au
geonetwork
TERN TDDP
tern.org.au
RDSI VM’s
raijin@nci
INTERSECT
NeCTAR
PALS
EVALUATION
NeCTAR
Virtual
Labs
8. eMAST Delivers in 2014-2015 : 1 of 3
Simple land surface process models
• eMAST R-Package: MQ & ANU Bioclimate indices and surface processes
• eMAST Earth System Model Connex (C++ & FORTRAN): MQ & ANU
Bioclimate indices and surface processes coupled to ACCESS and other
Earth System Models
• ePiSaT R-Package: Continental Gross Primary Production (data model
fusion)
• Community R-Packages: Hutchinson Drought & BoM Heatwave – in kind
from Ivan Hanigan (ANU)
• pyeMAST: Python version of eMAST tools including big data services
(connectivity with SPEDDEXES).
Statistical land surface models
• Data Assimilation: Ensemble Kalman Filter coupled to process based land
surface model (Renzullo, CSIRO)
• Fubaar: Machine learning land surface model (in-kind MQ – Keenan)
Open Source !
Tools
9. eMAST Delivers in 2014-2015 : 2 of 3
Observation assimilation into Models
• eMAST Ecosystem Model Parameters Database (EMP DB).
• NCAR’s Data Assimilation Research Testbed (DART)
• DART-CESM : In collaboration with NEON, Inc. (USA)
• DART-CABLE : In collaboration with the NCI, NCAR and CSIRO
• Assimilation of : fluxes, leaf properties, plot network observations
Modelled Data discovery and ACCESS Tools
• SPEDDEXES: A community based solution to (a) publishing big data (b)
sharing big data (c ) discovering big data and (d) programmatic access to
big data on Australia’s eResearch infrastructure.
• SPEDDEXES@NeCTAR-VL’s: Collaborative extension of the SPEDDEXES tools
to the NeCTAR Virtual Laboratories – embedding in the Climate and
Weather Laboratory
Benchmarking and Evaluation
• eMAST@PALS : Development of the PALS system for eMAST and TERN data
streams
• eMAST BENCH : International collaboration on benchmarking
Tools
10. eMAST Delivers in 2014-2015: 3 of 3
NEXT Generation of Ecosystem Models
• ARC DP on Australian Tropical Savanna’s : Past Present and Future:
Enhancing ecosystem models for Tropical Savanna’s
• ARC DP on the Next Generation of Ecosystem Models: Using plant trait
observations to inform a new approach to ecosystem modelling.
• GePiSaT: Global version of the ePiSaT model (eMAST and Imperial College
of London)
• CAMELS: Coupling ACCESS with Models of Ecosystems and the Land
Surface: Next generation approach to ecosystem and land surface
modelling
Datasets from eMAST
• ANUClimate: A extension of past methods for gridding Climate and
Weather for the Australian continent .
• eMAST Bioclimate
• eMAST Land Surface Modelling
Tools & Data
11. Climate and Bioclimate data
Res. 0.01 degrees (nominally 1km) T, P, R + and 50 + indices
12. : New approach for Big Data
It is no longer practical, let alone affordable, to
continue to do data-intensive ecosystem science
in the copy-and-work paradigm, a new approach
to working with Big Data is required.
Think about network data access, not file downloads
…
Cross-disciplinary use of file formats and services
…
Open-source server technology and file formats
…
Work with big data in a high performance facility
13. Big Data : eMAST’s collections
10
100
1000
10000 5419
1928
326
176 140
DataVolumes(TB)
Scientific Data for Research (NCI RDSI node)
by 2015
14. Three eMAST projects
1. Observations: The Ecosystem Model Parameters
Database
2. Models: Ecosystem Production in Space and Time
3. Observations in Models: CABLE-DART Data
assimilation on the NCI
15. Observations
The Ecosystem Model Parameters Database
• Originally setup to generate continental
scale surfaces of leaf properties
(nitrogen, phosphorus etc) using ANN’s
• Adapted in April 2014 for use with Data
assimilation
• Focal point for ecosystem scientists and
plot networks to contribute observations
for use in models
EMP DB
Example One
19. eMAST : Data assimilation
Collaborative ‘Community’ approach: Work with international experts (Fox –
NEON and Hoar – NCAR) and local champions Renzullo (CSIRO) and Evans. Open
to community participation (Wang, Haverd and Trudinger CSIRO)
22. Ecosystem Production in Space and Time
Example Three
ePiSaT
Data filtering:
Removal of outliers
etc.. Gap filling of
PAR (PPFD) for GPP
1
3
1R =
Assimilation
Amax = - 2
Efficiency
Φ =
2
2
3
Amax *
FC =
Rectangular
Hyperbole
3 parameter
1 2 3
Respiration
Quantum
R -
Φ I
Amax +Φ I
23. How does gross primary
productivity (GPP) vary in space
and time across Australia?
How can we ‘simply’ estimate GPP
across Australia?
What data does TERN provide that
might be useful for addressing this
research question?
Ecosystem Production in Space and Time
ePiSaT
24. Choose the ePiSaT model from
emast.org.au
TDDP or
SPEDDEXES
Obtain OzFlux data via the
TERN/ OzFlux portals
Run the ePiSaT model –
generate estimates of
ecosystem parameters,
evaluate them
Obtain climate (eMAST) and
satellite data (AusCover) to
scale the ePiSaT parameters
Produce continental scale
estimates of GPP and evaluate
them
Ecosystem Production in Space and Time
ePiSaT
25. This project is supported by the Australian National Data Service (ANDS). ANDS is supported by the Australian Government through the National Collaborative Research Infrastructure Strategy
Program and the Education Investment Fund (EIF) Super Science Initiative. For more information visit the ANDS website ands.org.au and Research Data Australia services.ands.org.au.
Opensource solution - Open servers technology, open data formats. NetCDF - cross-diciplinary use i.e. meteorology, hydrology & modellers - climate, land surface and remote sensing.
Value added services such as aggregation services