Remote sensing of biophysical parameters: linking field, airborne and continental scale data_Johansen
1. Remote Sensing of Biophysical
Parameters: Linking Field, Airborne
and Continental Scale Data
Presentation by: Kasper Johansen1,2 and Stuart Phinn1,2
1The University of Queensland (k.johansen@uq.edu.au)
2 Joint Remote Sensing Research Program
2. Outline
1. Introduction: Australian Supersites Network and
AusCover
2. AusCover Activities and Products
3. AusCover Data Collection
4. Linking Field, Airborne and Satellite Image Data
3. Plot locations of the ASN
Australian Supersites Network (ASN)
and AusCover
Plot locations of AusCover
5. AusCover Activities and Products
A nationally consistent approach to deliver, calibrate and validate satellite
image based data sets designed for Australian conditions.
6. Field-based Measurements Satellite Based MeasurementsAirborne Measurements
AusCover Activities and Products:
AusCover Field and Airborne Campaigns
Time-Series Measurements
7. Field and Airborne Data Collection Design
• Each site represents dominant
biome or environment with
high conservation value
• Homogenous 5 x 5 km sites
• Capture variability within site
• Rapid sites (TLS, hemispherical
photos, GPS)
• Field data collection
influenced by accessibility and
type of environment
8. • Homogenous plots
• Ground, mid-storey and
over-storey
• Basal area
• Vegetation structure
• LAI (LAI-2200)
• Hemispherical photos
• Terrestrial Laser Scanning
• Leaf samples and species
ID
• Phenology photos
Field and Airborne Data Collection Design
27. AusCover Products
• The vertically-projected fraction of long-term, persistent green
vegetation (nominally woody vegetation) cover
28. Landsat Time-Series of Persistent Green-
Vegetation Fraction for Australia
SOURCE DESCRIPTION
QLD DSITIA Fractional-cover field sites
ABARES Fractional-cover field sites
NSW OEH SPOT-5 Image-interpretation
NT Bushfires DBH field sites
NT NRETAS Fractional-cover field sites
ACRIS Locations of low-foliage scrub
WA Woody-vegetation sampling sites
QLD Herbarium Biomass field sites
• Field data from 800 sites collected using nationally consistent protocol, RMSE of 11%
• Decision tree classifier based on robust regression statistics used to classify each pixel
as persistent or non-persistent green vegetation
29. Landsat Time-Series of Persistent Green-
Vegetation Fraction for Australia
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
0 1000 2000 3000 4000
day
greenfraction
min
max
mask
Non-PGV
30. Persistent Green Vegetation Fraction –
Airborne LiDAR Validation
• Collation of Riegl LMS-Q560 and Riegl LMS-Q680i waveform LiDAR datasets
captured within the temporal extent of the product (2000-2010)
• Woody Foliage Projective Cover estimates from field calibration of LiDAR Pgap
• Comparison with Landsat persistent green extent and cover fractions
31. Persistent Green Vegetation Fraction –
Airborne LiDAR Validation
• The mean RMSE and Pearson correlation coefficient across all flight
paths were 0.13±0.076 and 0.71±0.15, respectively.
• Herbaceous and woody understorey not included in LiDAR FPC (<0.5m)
33. Future Work & Conclusions
Future Work
• Research linking both ASN and AusCover field, airborne and satellite
image data and flux measurements (e.g. for productivity, biomass,
time-series change, vegetation structure and phenology assessment)
• Expand data collection over time
Conclusions
• Unique data sets available for selected supersites across Australia
• Will support research and ecosystem science in Australia in selected
biomes
• Working with state and federal government agencies and researchers
associated with AusCover and TERN enabled this work
34. Acknowledgements
AGENCY PEOPLE
ABARES Jasmine Rickards
NT Bushfires Andrew Edwards
NT NRETAS Nick Cuff
ACRIS / CSIRO Gary Bastin, Matt Bradform
WA DEC Graeme Behn
Airborne Research Australia Jorg Hacker
Monash Jason Beringer
CDU Stefan Maier
QLD Herbarium
NSW Office of Environment and Heritage Tim Danaher
JCU Mike Liddell
Individual AusCover nodes
TERN in general
35. Remote Sensing of Biophysical
Parameters: Linking Field, Airborne
and Continental Scale Data
Presentation by: Kasper Johansen1,2 and Stuart Phinn1,2
1The University of Queensland (k.johansen@uq.edu.au)
2 Joint Remote Sensing Research Program