Environmental remote sensing data used at VTT
Tuomas Hä[emailprotected]
+358 20 722-6282
VTT TECHNICAL RESEARCH CENTRE OF FINLAND
2
VTT Technical Research Centre of Finland
VTT ISthe biggest multi-technological applied research organisation in Northern Europe
VTT HASmultidisciplinary R&D from electronics to building technologyclients and partners: industrial and business enterprises, public sector, universities and research institutes
VTT CREATESnew technology and science-based innovations in co-operation with domestic and foreign partners
Turnover 232 M€Personnel 2,74076% with higher academic degree5,730 customersEstablished 1942 VTT has been granted ISO9001:2000 certificate.
VTT TECHNICAL RESEARCH CENTRE OF FINLAND
3
Environmental remote sensing data use in the nutshell
Which data• All kind of imagery from Earth's surface• From ground-based cameras to AVHRR• Optical & radar
Use• Value adding chain development including• Preprocessing methods• Analysis methods• Software development• System development
For whom• Forest industry, consultants, European Space Agency,
Finnish and foreign governmental agencies and foreign private sector
Applications• Sustainable use of natural resources, mitigation of
environmental change impact and environmental disasters. Cost savings, more effective operation, timely information.
VTT TECHNICAL RESEARCH CENTRE OF FINLAND
4
Environmentmonitoring
service
Environmentmonitoring
service
Environmentmonitoring
service
Environmentmonitoring
serviceEnvironmentMonitoring
Service
• Analysis• Processing• Product Tailoring
Media
Satellite and airborne data
In situ measurements
Data providers
Models• Biomass• Weather• Ice• Smoke• Oil
GIS DBs ContextContent
Connecting people and data
VTT TECHNICAL RESEARCH CENTRE OF FINLAND
5
Calibrated mosaic from Europein one-kilometer resolution using 63 weather satellite images
VTT TECHNICAL RESEARCH CENTRE OF FINLAND
6
VTT TECHNICAL RESEARCH CENTRE OF FINLAND
7
Location of forest types in red / NIR spectral domain
Figure 7. Target variable values in the spectral space in the Temperate and Borealstratum.
VTT TECHNICAL RESEARCH CENTRE OF FINLAND
8
• Modis 1B-level data from summers 2000 and 2001
• 27 images• Ground
resolution 250 meters
• Size of the mosaic about 1400 km by 1900 km
• red channel = red reflectance
• green channel = near-infrared reflectance
• Landsat ETM+ image from 6th of July 2001 from Suonenjoki and stem volume values from Suonenjoki plot data
• Landsat ETM+ image from 10th
of June 2000 from Puumala and stem volume values from Puumala plot data
VTT TECHNICAL RESEARCH CENTRE OF FINLAND
9
Stem volume models estimated from the Suonenjoki data sets
Coniferous trees Broad-leaved trees
VTT TECHNICAL RESEARCH CENTRE OF FINLAND
10
Total stem volume in Finnish forestry districts
50
100
150
200
250
300
350
400
S-Both
S-Sav
oHäm
e-Uus
Centra
l Fin
Kainuu
Kymi
SW FinLa
pland
N-Kare
liaN-B
othN-S
avo
Pirkan
maaCoa
stÅlan
d
mill
. m3
EstimateStatistics
VTT TECHNICAL RESEARCH CENTRE OF FINLAND
11
JERS radarimage mosaic (preliminary)from Eurasia
Glo
bal B
orea
l For
est M
appi
ng P
roje
ct -
JAX
A
VTT TECHNICAL RESEARCH CENTRE OF FINLAND
12
Effect of topographic normalization
SAR image without (left) and with (right) utilization of the DEM in radiometric and geometric correction
VTT TECHNICAL RESEARCH CENTRE OF FINLAND
13
• Osa mosaiikista perspektiivikuvana
• Korkeutta liioiteltu 5-kertaisesti
• Perspektiivikuva tehty GRASS-GIS ohjelmistolla
• Lokan tekoaltaan itäpäästä kohti Sokostin seudun tuntureita
• Kuvat huhtikuulta 2007
• ALOS/Palsar data © JAXA, METI 2007
Esimerkki orto-oikaistusta Polarimetrisesta tutkakuvamosaiikista
VTT TECHNICAL RESEARCH CENTRE OF FINLAND
14
Dual-Pol 12.6.2007/Heinavesi: Stem Volume Correlation/HV
• Whole dataset: r = 0.79
• 0…150 m3/ha: r = 0.93
• Saturation around 150 m3/ha
• One obvious clear-cut stand (harvested March 2007, ground March 2007)
VTT TECHNICAL RESEARCH CENTRE OF FINLAND
15
Example of classification result
May: PolSARpro supervised ALOS/Palsar data © JAXA, METI 2007
VTT TECHNICAL RESEARCH CENTRE OF FINLAND
16
ALOS/Palsar in the detection of clear cuts
R=20070612G=20070728B=20070912Yellow = cut
28.7.-12.9.Red = cut
12.6.-28.7.ALOS/Palsar
data © JAXA, METI 2007
VTT TECHNICAL RESEARCH CENTRE OF FINLAND
17
Forest cover change in French Guiana1992-2003-2007using ESA radar
VTT TECHNICAL RESEARCH CENTRE OF FINLAND
18
Natural disaster monitoring workstation
Web application infrastructure– Satellite imagery analysis– Other information– Multi-user tool with different user roles
VTT TECHNICAL RESEARCH CENTRE OF FINLAND
19
Growing stock volume
287 m3/ha
0 m3/ha
Ikonos false color image of Suonenjoki test area estimate plotted on Ikonos PAN channel
VTT TECHNICAL RESEARCH CENTRE OF FINLAND
20
Stem diameter
28 cm
0 cm
Ikonos false color image of Suonenjoki test area estimate plotted on Ikonos PAN channel
VTT TECHNICAL RESEARCH CENTRE OF FINLAND
21
Stem number / ha
2941 stems/ha
0 stems/ha
Ikonos false color image of Suonenjoki test area estimate plotted on Ikonos PAN channel
VTT TECHNICAL RESEARCH CENTRE OF FINLAND
22
Proportion of broadleaved trees from stem volume
100 %
0 %
Ikonos false color image of Suonenjoki test area estimate plotted on Ikonos PAN channel
VTT TECHNICAL RESEARCH CENTRE OF FINLAND
23
Accuracy at plot level
0 100 200 300 400 5000
50
100
150
200
250
300
350
400
450
500Estimated stem vol vs. gnd data, FS1-BL-111203
stem volume [m3/ha]
stem
vol
est
imat
e [m
3/ha
]
R = 0.85
0 5 10 15 20 25 30 350
5
10
15
20
25
30
35Estimated stem diameter vs. gnd data, FS1-VOL-101203a
stem diameter [cm]
stem
dia
met
er e
stim
ate
[cm
]
R = 0.90
VTT TECHNICAL RESEARCH CENTRE OF FINLAND
24
Tree species study - tree detection
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
1
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
GF
1
a) on aerial image b) on IKONOS PAN channel
Detected trees on Plot No: 2
25 m
Spruce (validation data)Pine (validation data)Birch (validation data)Trees located on Ikonos image
VTT TECHNICAL RESEARCH CENTRE OF FINLAND
25
Forest Monitoring – Tree Species Map
Pine-%
Deciduous-%Spruce -%
Tree species proportions calculatedfor a 1 km x 1 km Ikonos-2 image
Average segment size 1 ha
VTT TECHNICAL RESEARCH CENTRE OF FINLAND
26
1996
VTT TECHNICAL RESEARCH CENTRE OF FINLAND
27
1997
VTT TECHNICAL RESEARCH CENTRE OF FINLAND
28
1998
VTT TECHNICAL RESEARCH CENTRE OF FINLAND
29
1999
VTT TECHNICAL RESEARCH CENTRE OF FINLAND
30
2000
VTT TECHNICAL RESEARCH CENTRE OF FINLAND
31
2001
VTT TECHNICAL RESEARCH CENTRE OF FINLAND
32
2002
VTT TECHNICAL RESEARCH CENTRE OF FINLAND
33
2003
VTT TECHNICAL RESEARCH CENTRE OF FINLAND
34
2004
VTT TECHNICAL RESEARCH CENTRE OF FINLAND
35
2005
VTT TECHNICAL RESEARCH CENTRE OF FINLAND
36
2006
VTT TECHNICAL RESEARCH CENTRE OF FINLAND
37
2007
VTT TECHNICAL RESEARCH CENTRE OF FINLAND
38
Forest Monitoring - SegmentationHierarchical Pyramidas segmentation result
Ikonos-2 image , 1.4 km x 1.0 km
The mean size of the segments is in the three segmentations:
White borders: 1 ha
White and yellow borders: 0.5 ha
White, yellow and green borders: 0.3 ha
VTT TECHNICAL RESEARCH CENTRE OF FINLAND
39
Image series from Enontekiö 28.8. - 28.9.2007Daily images recorded at high noon
Challenges
• Illumination intensity variation -especially sunny images
• Incident light colour balancevariations + trend
• Camera colour balance variations• Target movement (wind)
Web camera images
VTT TECHNICAL RESEARCH CENTRE OF FINLAND
40
ForSe - Forest and Season MonitoringWeb Cameras In Automatic Autumn Colour Monitoring
A time series of web camera images and corresponding feature images from Enontekiöautumn 2007
VTT TECHNICAL RESEARCH CENTRE OF FINLAND
41
SCI comparison with reference data/Oulanka site
Yellowing starts
Falling starts
Leaves yellowed (> 50%)
Leaves fallen (> 50%)
Phenology dates (Metla) marked with blue arrows
245 250 255 260 265 2700
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
DOY
Season colour development/Oulanka- Autumn 2007
SC
I
greenlight greenyellowbrownfallen
The leaves' colour development forOulanka site
• 9 ROIs (1 ref + 9 for classification)• classifier training with Enontekiö sitedata (see slide 13)• no instances in class 'red' (not plotted)
The ROIs shown on Oulanka site imagetaken 16.9.2007
VTT TECHNICAL RESEARCH CENTRE OF FINLAND
42
Lehdet täysikasvuisia
Kellastuminen alkaa
Variseminen alkaa
Lehdet kellastuneet
Lehdet varisseet
Kartan värien selitykset
Monitoring of seasonal changes
Autumn coloring in September-October 2005
VTT TECHNICAL RESEARCH CENTRE OF FINLAND
43
Kioto+ (post-Kyoto) Satellite Mission in Brief
Proposed near-polar orbiting satellite monitoring system • Super high-resolution (0.5 m) optical images• Statistical sampling principle => conceptually comparable to field
measurements• Individual trees visible• Land cover class/class change can be defined without any
uncertainty
The images are permanent documents about the situation during image acquisition
Three major uses:1. Statistical data on global, regional and national forest and land
cover with confidence interval (target area 25 000 km2 and larger)2. Training and validation data for "wall-to-wall" satellite imagery3. Augmentation of field measurements
VTT TECHNICAL RESEARCH CENTRE OF FINLAND
44
Sampling principle of Kioto+
QuickBird-satellite image from Southern FinlandResolution 60 cm ©Digitalglobe
15 km4 km