Tucson Arizona, May 16-17, 2006 1 A.The EO’s: Role of Modeling (Science/Models in the Context of...

11
Tucson Arizona, May 16-17, 20 06 1 A. The EO’s: Role of Modeling (Science/Models in the Context of the EO’s & Enabling Technologies) B. Mobilizing Theory in Principle: Cyber- Infrastructure C. “Wild Seeds”: Modeling in a Non-EO Context D. Reconciling Theory With Observation (Reconciling Theory with Data) E. EO’s and Model Applications for Environmental Stewardship (Model Applications & Matters Arising) F. Educational & Learning Opportunities Session Chair: Hoshin Gupta (University of Arizona) Grand Challenges of the Future for Environmental Modeling Environmental Observatories Modeling Workshop May 16-17, 2006, Tucson, Arizona Reconciling Theory With Observation (Reconciling Theory with Data)

Transcript of Tucson Arizona, May 16-17, 2006 1 A.The EO’s: Role of Modeling (Science/Models in the Context of...

Page 1: Tucson Arizona, May 16-17, 2006 1 A.The EO’s: Role of Modeling (Science/Models in the Context of the EO’s & Enabling Technologies) B.Mobilizing Theory.

Tucson Arizona, May 16-17, 2006

1

A. The EO’s: Role of Modeling (Science/Models in the Context of the EO’s & Enabling

Technologies)

B. Mobilizing Theory in Principle: Cyber-Infrastructure

C. “Wild Seeds”: Modeling in a Non-EO Context

D. Reconciling Theory With Observation(Reconciling Theory with Data)

E. EO’s and Model Applications for Environmental Stewardship(Model Applications & Matters Arising)

F. Educational & Learning Opportunities

A. The EO’s: Role of Modeling (Science/Models in the Context of the EO’s & Enabling

Technologies)

B. Mobilizing Theory in Principle: Cyber-Infrastructure

C. “Wild Seeds”: Modeling in a Non-EO Context

D. Reconciling Theory With Observation(Reconciling Theory with Data)

E. EO’s and Model Applications for Environmental Stewardship(Model Applications & Matters Arising)

F. Educational & Learning Opportunities

Session Chair: Hoshin Gupta (University of Arizona)

Grand Challenges of the Future for Environmental ModelingEnvironmental Observatories Modeling Workshop

May 16-17, 2006, Tucson, Arizona

Grand Challenges of the Future for Environmental ModelingEnvironmental Observatories Modeling Workshop

May 16-17, 2006, Tucson, Arizona

Reconciling Theory With Observation(Reconciling Theory with Data)Reconciling Theory With Observation(Reconciling Theory with Data)

Page 2: Tucson Arizona, May 16-17, 2006 1 A.The EO’s: Role of Modeling (Science/Models in the Context of the EO’s & Enabling Technologies) B.Mobilizing Theory.

Tucson Arizona, May 16-17, 2006

2

Grand Challenges

“What new technologies for observing, simulating, and telecommunicating will emerge over the next 5-10 years?

How will they change the grand challenges for modeling?

What will those challenges be?

How might those challenges be pursued?

Grand Challenges

“What new technologies for observing, simulating, and telecommunicating will emerge over the next 5-10 years?

How will they change the grand challenges for modeling?

What will those challenges be?

How might those challenges be pursued?

Certainly we can discuss how NSF can support ever more efficient ways (technologies) to help us do the modeling equivalent of “business as usual” …

But I hope we will spend some time on …What are the grand challenges? How might we approach them?

Page 3: Tucson Arizona, May 16-17, 2006 1 A.The EO’s: Role of Modeling (Science/Models in the Context of the EO’s & Enabling Technologies) B.Mobilizing Theory.

Tucson Arizona, May 16-17, 2006

3

Models are complex assemblies of multiple, constituent hypotheses… that must be tested … against the new streams of field data …

Working out novel ways of conducting these tests, will be a major scientific challenge associated with the EO’s

ACTSReconciling Theory With Observation

Theories <---> Models <---> ACTS <---> ObservationsTheories <---> Models <---> ACTS <---> Observations

How on earth are we going to take “large” models …

and large volumes of data …

juxtapose them …

and make sense of this juxtaposition ???

How on earth are we going to take “large” models …

and large volumes of data …

juxtapose them …

and make sense of this juxtaposition ???

Page 4: Tucson Arizona, May 16-17, 2006 1 A.The EO’s: Role of Modeling (Science/Models in the Context of the EO’s & Enabling Technologies) B.Mobilizing Theory.

Tucson Arizona, May 16-17, 2006

4

To be effective, an intervention must be introduced at the correct logical level

If a problem seems unsolvable … consider that you may have a meta-problem.

JOHN GALL(Systemantics: The Underground Text of Systems

Lore)

ACTSReconciling Theory With Observation

Page 5: Tucson Arizona, May 16-17, 2006 1 A.The EO’s: Role of Modeling (Science/Models in the Context of the EO’s & Enabling Technologies) B.Mobilizing Theory.

Tucson Arizona, May 16-17, 2006

5

DATA IS NOT INFORMATION !!!

DataBase

Information AContext A

Information BContext B

Prior InformationPrior Knowledge

Analysis

Page 6: Tucson Arizona, May 16-17, 2006 1 A.The EO’s: Role of Modeling (Science/Models in the Context of the EO’s & Enabling Technologies) B.Mobilizing Theory.

Tucson Arizona, May 16-17, 2006

6

DATA IS NOT INFORMATION !!!

Information is obtained by viewing data in a certain context (perceptual filtering).

There may be (are always) multiple plausible contexts …

Extraction -- “What kind of information does this data contain?”

“How do we extract it?”

SIGNAL (Data) ---> SIGNATURES (Information)

We should generally be interested in “minimal” representations of Information … … they indicate how much complexity we can resolve

RInfo << RData

Page 7: Tucson Arizona, May 16-17, 2006 1 A.The EO’s: Role of Modeling (Science/Models in the Context of the EO’s & Enabling Technologies) B.Mobilizing Theory.

Tucson Arizona, May 16-17, 2006

7

Confront Models with Information !!!

Perceptual

Conceptual

Symbolic

Computational

DATA

FITTINGVERIFICATIONVALIDATION

FITTINGVERIFICATIONVALIDATION

INFORMATION

PAR. ESTIMATIONEVALUATION

FAULT DETECTION

DIAGNOSIS

PAR. ESTIMATIONEVALUATION

FAULT DETECTION

DIAGNOSIS

? Assumptions ?

Page 8: Tucson Arizona, May 16-17, 2006 1 A.The EO’s: Role of Modeling (Science/Models in the Context of the EO’s & Enabling Technologies) B.Mobilizing Theory.

Tucson Arizona, May 16-17, 2006

8

Classical Approach to Model Evaluation

Is

I-S-OMeasurement Os

N-Dim Time Series

RD

Om

I-S-OSimulation

N-Dim Time Series

RD

ModelModel

Structure (Eqns)Parameters ()

R

SystemSystem

Form & FunctionInvariants

RS

-

C()

=Om- Os

Measure of Closeness(Single Criterion based on residuals)

R1

OO

timetime

t

t

Fundamentally WeakHas Little Diagnostic Power

Page 9: Tucson Arizona, May 16-17, 2006 1 A.The EO’s: Role of Modeling (Science/Models in the Context of the EO’s & Enabling Technologies) B.Mobilizing Theory.

Tucson Arizona, May 16-17, 2006

9

Diagnostic Approach to Model Evaluation

I-S-OData ModelModelI-S-O

Simulation

Is

SystemSystem

RC R?

Measures of Closeness

(Diagnostics)

1M - 1

S

cM - c

S

RD RD R

RS >> RD >> RC

Structure (Eqns)Parameters ()

Form & FunctionInvariants

RS

Model Generated “Data”(N-Dim Time Series)

Observed Data(N-Dim Time Series)

I-S-O Structures

I-S-O Indices (S)

RC

RC

Indices(Pattern

Properties)

Signatures(Pattern Extracts)

I-S-O Structures

I-S-O Indices (M)

RC

Signatures(Pattern Extracts)

Indices(Pattern

Properties)

Model Identification <--> Search for Isomorphisms

M()

S

1

2

Page 10: Tucson Arizona, May 16-17, 2006 1 A.The EO’s: Role of Modeling (Science/Models in the Context of the EO’s & Enabling Technologies) B.Mobilizing Theory.

Tucson Arizona, May 16-17, 2006

10

Grand Challenges

DATA (Large Volumes) ---> INFORMATION

DIAGNOSTICS APPROACH TO SYSTEMS IDENTIFICATION

Grand Challenges

DATA (Large Volumes) ---> INFORMATION

DIAGNOSTICS APPROACH TO SYSTEMS IDENTIFICATION

Page 11: Tucson Arizona, May 16-17, 2006 1 A.The EO’s: Role of Modeling (Science/Models in the Context of the EO’s & Enabling Technologies) B.Mobilizing Theory.

Tucson Arizona, May 16-17, 2006

11

Ed Rastetter - Marine Biological Lab, Woods HoleDeveloping and Testing Mechanistic Models of Terrestrial Carbon Cycling Using Time Series Data

Hans Graber - University of MiamiReal-Time Forecasting System of Winds, Waves and Surges in Tropical Cyclones

Three Breakout Sessions

THIS MORNING