Download presentation
Presentation is loading. Please wait.
Published byNathaniel Rich Modified over 9 years ago
1
DataModel When data and model are in isolation We are getting …
2
Soil carbon modeled in CMIP5 vs. HWSD
3
DataModel Prediction Integrated Data-Model Approaches to Carbon Cycle Research
4
Mike Kuperberg, DOE program officer: DOE's perspective on the data-model integration Yiqi Luo, University of Oklahoma: Challenges and opportunities in data-model integration Anthony Walker, ORNL: Benchmark analysis of models against data from FACE experiments Sasha Hararuk, University of Oklahoma: Evaluation and improvement of global land models against soil carbon data. Robert Cook, ORNL: Ecoinformatics and cyberinfrastructure to promote data-model integration Panelists
5
A new philosophy of research Modeling activities guide field research activities, which in turn informs modeling activities. This cyclical processing of information should maximize the financial and scientific investments and result in high quality predictive models
6
Field researchModeling Scientific inquiry Process thinkingData Gain best knowledge from imperfect data and imperfect models ?
7
1. Benchmarking: Data used to evaluate model performance 2.Data assimilation: Multiple streams of data ingested into model to improve its performance 3. Parameterization: Data used to parameterize models 4.Process representation: New algorithms to represent processes instead of a black-box approach Experiment results model
8
Benchmarking
9
Problem: If an incomplete set of variables are used for benchmarking, …benchmarking can give false confidence when models predict with some accuracy but for the wrong reasons
10
How do CLM-CASA’ and CABLE simulate Soil C? IGBP-DIS data
11
Data assimilation to improve soil C simulation by two global models: IGBP-DIS data
12
Changes in temporal dynamics: CLM-CASA’ 5,270 Pg 5,780 Pg 11,100 Pg
13
3/1/2012 5 th NSF RCN FORECAST meeting Forecast of Resource and Environmental Change using data Assimilation Science and Technology
14
1. Model strengths and deficiencies: Effective communication from modeling to experimental communities? 2. Data model products: What are the data model products directly useful for model improvement? 3. Infrastructure: Data assimilation techniques, data model products, cyberinfrastructure, visualization, and analytic tools. 4. Possible national center(s): Infrastructure development and coordination of activities Strategies to promote experiment- model interactions
15
DAAC, CDIACNCAR CLM Field researchModeling Scientific inquiry Process thinkingData
16
16
17
Users can access observational data and convert to their specified format, spatial resolution, spatial extent, and temporal extent. Pilot Study: Integrate Observations with Models using “Access Broker” Original Observational Data FTP/HTTP/… SCRIP (regrid) Data Process Customized Observational Data Request for Data 17 Original MODIS Data MODIS Web Service Model-Data Comparison Framework Data Assimilation Framework Stefano Nativi et al.
18
Field researchModeling Scientific inquiry Process thinkingData National center for experiment-model integration (NCEMI)
Similar presentations
© 2024 SlidePlayer.com. Inc.
All rights reserved.