Remote Sensing Technology for Scalable Information Networks Douglas G. Goodin Kansas State...

Post on 11-Jan-2016

212 views 0 download

Tags:

Transcript of Remote Sensing Technology for Scalable Information Networks Douglas G. Goodin Kansas State...

Remote Sensing Technology for Remote Sensing Technology for Scalable Information NetworksScalable Information Networks

Douglas G. GoodinKansas State University

Geoffrey M. HenebryUniversity of Nebraska - Lincoln

Ecological Remote Sensing enables recurrent observation…

What is the role of remote sensing in ecological research?

…at vast but variable spatial extents…

…at multiple spatial scales…

Konza Prairie – 4 m resolution Konza Prairie – 1000 m resolution

Konza

…and provides regional context

*Konza

Elements of Remote Sensing

Remote Sensing Technology is…

Hardware – sensors, computers, storage, distribution networks

Software – commercial, public domain,

user-created

“Wetware”– scientists, data managers

What are the Elements of Remote Sensing Technology (from an ecological perspective)?

Orbital, airborne, near-ground sensor systems Ranges of spatial, temporal, & spectral

resolutions

System for data acquisition, processing,

distribution, & archiving Algorithms to retrieve biogeophysical variables Theory for interpretation & prediction

Observed

Phenomenon

Spectral Region

Biogeophysical Variables

Representative

Sensors

Ranges of Resolutions

Solar Reflectance

Visible,

Near-IR,

Mid-IR

Albedo

fPAR

Land Cover

NPP

AVHRR SeaWiFS

MODIS MERIS

TM/ETM+ ALI IKONOS

AVIRIS MASTER

1 m – 1 km

<1 d – 18 d

1–228 bands

Terrestrial Emission

Mid-IR,

Thermal-IR,

Microwaves

Surface temperature

Surface moisture

SMMR SSM/I

AVHRR MODIS

ASTER TIMS

25 m - 25 km

<1 d – 3 d

1 – 50+ bands

Anthropogenic Radiation

RADAR,

LIDAR,

[SONAR]

Surface roughness

Soil moisture

Terrain

RADARSAT ASAR

JERS SIR-C

VCL LVIS

8 m – 150 m

18 d

<10 bands

Types of Earth Observing Sensors

Orbital Remote Sensing Systems

Landsat

US – Private/Gov’t

Moderate spatial resolution

1972-Present

IKONOS

US – Private

1999 – present

Very fine spatial resolution (1-4m)

NOAA – Polar Orbiter

US Government

Coarse spatial resolution, global coverage

1982 - Present

RADARSAT

Canada – Gov’t/private

Imaging radar

1996 - Present

Terra/EO-1“Next-Generation” – Earth Observation

• Multi-instrument platform

• Multispectral, hyperspectral

Coordinated observationWith Landsat - 7

Aircraft Sensing Systems

• Flexible mission planning• Selectable spatial resolution• High cost (?)

AVIRIS

• US Gov’t (NASA)

• Hyperspectral (224 bands)

• Multiple Aircraft (ER-2, Twin Otter)

Other Aircraft Systems

•Multiple (light) aircraft platforms

•(Relatively) modest cost

•Researcher control!

Close Range Remote Sensing

•A wide variety of multi/hyper spectral instruments

•Not just “ground truth”

•Researcher control

TheData

Pyramid

Coordinated Observation at Multiple Scales

What are the Elements of Remote Sensing Technology (from an Ecological perspective)?

Orbital, airborne, near-ground sensor systems Ranges of spatial, temporal, & spectral

resolutions System for data acquisition, processing,

distribution, & archiving Algorithms to retrieve biogeophysical variables Theory for interpretation & prediction

Observed

Phenomenon

Spectral Region

Biogeophysical Variables

Representative

Sensors

Ranges of Resolutions

Solar Reflectance

Visible,

Near-IR,

Mid-IR

Albedo

fPAR

Land Cover

NPP

AVHRR SeaWiFS

MODIS MERIS

TM/ETM+ ALI IKONOS

AVIRIS MASTER

1 m – 1 km

<1 d – 18 d

1–228 bands

Terrestrial Emission

Mid-IR,

Thermal-IR,

Microwaves

Surface temperature

Surface moisture

SMMR SSM/I

AVHRR MODIS

ASTER TIMS

25 m - 25 km

<1 d – 3 d

1 – 50+ bands

Anthropogenic Radiation

RADAR,

LIDAR,

[SONAR]

Surface roughness

Soil moisture

Terrain

RADARSAT ASAR

JERS SIR-C

VCL LVIS

8 m – 150 m

18 d

<10 bands

Types of Earth Observing Sensors

Spatial Resolution

Coarse FineModerate

Spectral Resolution

Panchromatic: 1 spectral band - very broad

Multispectral: 4-10 spectral bands - broad

Superspectral: 10-30 spectral bands - variable

Hyperspectral: >30 spectral bands - narrow

The challenge of hyperspectra is to reduce dense, voluminous, redundant data into a compact, effective suite of superspectral bands and indices for retrieval of biogeophysical fields.

What are the Elements of Remote Sensing Technology (from an Ecological perspective)?

Orbital, airborne, near-ground sensor systems Ranges of spatial, temporal, & spectral

resolutions System for data acquisition, processing,

distribution, & archiving Algorithms to retrieve biogeophysical variables Theory for interpretation & prediction

Acquisition

Processing

Distribution/Storage

Data Handling System - Hardware

Data analysis system – linkages are critical

Archiving/Distribution

Researchers/Groups

The MODIS systemAn example

What are the Elements of Remote Sensing Technology (from an Ecological perspective)?

Orbital, airborne, near-ground sensor systems Ranges of spatial, temporal, & spectral

resolutions System for data acquisition, processing,

distribution, & archiving Algorithms to retrieve biogeophysical variables Theory for interpretation & prediction

NDVI = (NIR - Red)/(NIR + Red)

R = f(,) sin cos d d

0 = [((i=1..N)xi2)/N] * [(C/k) * (sin )/(sin ref)]

Retrieval of Biogeophysical Quantities & Indices

EVI =2.5*(NIR-Red)/(L+NIR+C1*Red-C2*Blue)

Calibration to derive physical quantities: an engineering problem

Does the instrument give the correct physical data?

Is the instrument’s range & sensitivity appropriate for the application?

Cross-sensor calibration

Calibration to derive ecological quantities: a scientific problem

Can the sensor data yield ecologically relevant relationships?

NOT ground “truth” – ground level observation RESCALING

Empirical relationships are site & time specific but reflectance, emission, and backscattering are interactions not intrinsic properties of observable entities

Calibration to derive ecological quantities: a scientific problem

Top-down vs. bottom-up modeling perspectives

Model invertibility

Model robustness

(4) June 1998 sampling

NDVI = 0.1226(ln{total aboveground biomass}) - 0.3171

r2 = 0.6075

0

0.1

0.2

0.3

0.4

0.5

0.6

0 1 2 3 4 5 6 7 8

total aboveground biomass ln(g/m2)

ND

VI

Moss-Annual

Not Moss-Annual

Linear (Not Moss-Annual)

Empirical Model – Top down

Analytical Models – Bottom up

What are the Elements of Remote Sensing Technology (from an Ecological perspective)?

Orbital, airborne, near-ground sensor systems Ranges of spatial, temporal, & spectral

resolutions System for data acquisition, processing,

distribution, & archiving Algorithms to retrieve biogeophysical variables Theory for interpretation & prediction

To enable ecological forecasting, we need monitoring strategies for

change detection: perceiving the differences

change quantification: measuring the magnitudes of the differences

change assessment: determining whether the differences are significant

change attribution: identifying or inferring the proximate cause of the change

Observations

Ground segmentAcquisition, processing,

storage, & archiving

Ground segmentAcquisition, processing,

storage, & archiving

Retrieval of biogeophysical variables

Spatio-Spectral-Temporal

analysisDefinitions of nominal trajectories and

estimates of uncertainty

Assimilation of current observational datastreams

Change detection Change quantification

Change attribution Change assessment

Ecological Questions &Hypotheses

Information for Ecological Forecasting

Tuning the macroscope of remote sensing to support ecological inference requires an integrated and sustained

approach to technology & theory

ACKNOWLEDGMENTS

DGG acknowledges support from NASA EPSCoR subcontract 12860.

GMH acknowledges support from NSF #9696229/0196445 & #0131937.