hyperforest combining lidar and hyperspectral data to ... · remote sensing to support forest...

HYPERFOREST Combining LiDAR and hyperspectral data

to support forest management

Pieter Kempeneers (VITO)

Frieke Van Coillie (UG)

Kris Vandekerckhove (INBO)

F. Morsdorf (RSL)

Renato Cifuentes (KUL)

Franz Kai Ronellenfitsch (GLI)

28/11/2014 2 © 2012, VITO NV

Remote sensing in support of forest management

LiDAR

+ Forest structural properties, light conditions, tree detection

Hyperspectral

+ Tree species mapping, biophysical and chemical properties

- Poor tree delineation, forest structure

© Uni-trier 2011 © OSU 2010

28/11/2014 3 © 2012, VITO NV

Hyperspectral and Lidar data fusion

+

» Improved hyperspectral pre-processing

» Improved forest parameter retrieval

» Improved tree species mapping

Objectives

28/11/2014 4 © 2012, VITO NV

0 4 0 8 0 1 2 0 1 6 0 2 0 0 K i l o m e t e r s

N

E W

S

Flanders

Wallonia

Brussels Wijnendale forest

Aelmoeseneie forest

Kersselaerspleyn Study area

28/11/2014 5 © 2012, VITO NV

Study area

Kersselaerspleyn

• Low structural complexity

• Homogeneous old beech stands

• Limited admixture of oak

28/11/2014 6 © 2012, VITO NV

Study area

Wijnendale forest

• Medium structural complexity

• Mixed oak forest with maple, beech,

larch, hazel, …

28/11/2014 7 © 2012, VITO NV

Study area

Aelmoeseneie forest

• High structural complexity

• Mixed oak, beech, ash and larch stands

• Rich understorey

28/11/2014 8 © 2012, VITO NV

» Field measurements (FieldMap technology)

Data

Full dendrometric inventories

• Tree position • Species • DBH all trees > 5 cm diameter • Tree heights all trees upper

canopy

Canopy gaps • Hemispheric photos Tree vitality • On a selection of trees:

evaluation of discoloration and

leaf loss (international

methodology for vitality evaluation).

28/11/2014 9 © 2012, VITO NV


Data

Full dendrometric inventories

• Tree position • Species • DBH all trees > 5 cm diameter • Tree heights all trees upper

canopy

Canopy gaps • Hemispheric photos Tree vitality • On a selection of trees:

evaluation of discoloration and

leaf loss (international

methodology for vitality evaluation).

28/11/2014 10 © 2012, VITO NV


» LiDAR

» Terrestrial TLS (FARO LS 880HE; range: 70 m; wavelength: 785 nm)

Data

28/11/2014 11 © 2012, VITO NV


» LiDAR


» Airborne ALS (Riegl LMS Q560 full waveform; wavelength: 1560 nm;

point density > 10 points/m2)

Data

28/11/2014 12 © 2012, VITO NV


» LiDAR


» Airborne ALS (Riegl LMS Q560 full waveform; wavelength: 1560 nm;

point density > 10 points/m2)

» Hyperspectral data

Sensor Spectral resolution Spatial resolution

CASI 96 bands (368-1052 nm) 1 m

AHS 63 bands (452-2552 nm) 5 m

APEX 301 bands (375-2500 nm) 1.5 m

Data

Apex - Kersselaerspleyn

28/11/2014 13 © 2012, VITO NV

Work packages

28/11/2014 14 © 2012, VITO NV

50% leaves thinned

Upper canopy only

Full scene

Lower canopy only

Contribution of vegetation structure on spectral signature

28/11/2014 15 © 2012, VITO NV

Advanced HS image pre-processing

» Improving Dense Dark Vegetation mask for visibility estimation

» Additionally, shadow detection algorithm

28/11/2014 16 © 2012, VITO NV


» Spectral smoothing

28/11/2014 17 © 2012, VITO NV


» Improving geometric correction with LiDAR-based DSM

Geometric correction using traditional DTM

Geometric correction using LiDAR-based DSM

28/11/2014 18 © 2012, VITO NV


» Improving BRDF correction

28/11/2014 19 © 2012, VITO NV


» Improving BRDF correction

28/11/2014 20 © 2012, VITO NV


» HS image denoising

Band 1 original Band 1 denoised

28/11/2014 25 © 2012, VITO NV



Poplar Beech Copper beech

28/11/2014 26 © 2012, VITO NV

Improved forest canopy parameter retrieval

N1

N2

N3

N3

N4

N5

H0

H1

H2

H3

H4

H5

Canopy height model Height profiles

28/11/2014 27 © 2012, VITO NV


N1

N2

N3

N3

N4

N5

H0

H1

H2

H3

H4

H5

Fractional vegetation cover Height profiles

28/11/2014 30 © 2012, VITO NV

Improved tree species mapping

Hyperspectral data (quicklook) Tree height distribution

Insufficient information for accurate tree species classification

28/11/2014 31 © 2012, VITO NV

Accuracy assessment: no data fusion

Species VHM PHV APEX

PA UA PA UA PA UA

Beech 55 35 83 55 79 78

Ash 0 0 0 0 55 88

Larch 13 43 23 71 93 85

Poplar 68 61 81 76 100 99

Copper beech

0 0 0 0 27 82

Chestnut 0 0 0 0 72 64

Oak 37 34 64 53 53 71

Overall acc.

43 63 82

Kappa 0.24 0.50 0.76

Tree heights Tree structure Hyperspectral


28/11/2014 32 © 2012, VITO NV

Species Decision fusion

PA UA

Beech 96 96

Ash 91 98

Larch 99 99

Poplar 97 96

Copper beech

99 99

Chestnut 97 100

Oak 95 96

Overall acc.

96

Kappa 0.95


Accuracy assessment: data fusion

28/11/2014 34 © 2012, VITO NV

Project output

• Cifuentes La Mura, R., Van Der Zande, D., Van Beek, J., Farifteh, J., Coppin, P. (2013). Elements contributing to accuracy of canopy

structure assessment using terrestrial Lidar data in broad-leaved forests. ASPRS 2013 Annual Conference. ASPRS Annual Conference.

Baltimore, USA, 24-28 March 2013 (pp. 1-8) ASPRS.

• Cifuentes La Mura, R., Van Der Zande, D., Farifteh, J., Coppin, P. (2013). Effects of object reflectivity, sanning settings and resolution on

quality of terrestrial laser scanning data from a simulated forest canopy. SilviLaser 2013. SilviLaser 2013. Beijing, China, 9-11 October

2013.

• Cifuentes La Mura, R., Van Der Zande, D., Farifteh, J., Salas, C., Coppin, P. (2014). Effects of voxel size and sampling setup on the

estimation of forest canopy gap fraction from terrestrial laser scanning data. Agricultural and Forest Meteorology, 194, 230-240.

• Cifuentes La Mura, R., Van Der Zande, D., Salas, C., Farifteh, J., Coppin, P. (2014). Correction of erroneous LIDAR measurements in

artificial forest canopy experimental setups. Forests, 5 (7), 1565-1583.

• Cifuentes La Mura, R., Van der Zande, D., Farifteh, J., Tits, L. & Coppin, P. (2014). Modeling light distribution in forest canopies using

terrestrial LiDAR data in a ray-tracing environment. In preparation

• P.; Bertels, L.; Vreys, K.; Biesemans, J., "Geometric Errors of Remote Sensing Images Over Forest and Their Propagation to Bidirectional

Studies," Geoscience and Remote Sensing Letters, IEEE , vol.10, no.6, pp.1459,1463, Nov. 2013

• P. Kempeneers, F. Devriendt, F. Van Coillie, K. Vandekeckhove, F. Morsdorf, Combining LiDAR and hyperspectral remote sensing data

to improve information extraction for forestry, Silvilaser September 16-19, 2012, Vancouver, Canada

• Somers B., Pieter Kempeneers, Flore Devriendt, Frieke Vancoillie and Kris Vandekerkhove, APEX and LiDAR remote sensing supporting

fine-scale forest management, BruHyp Airborne Imaging Spectroscopy Workshop 2012 – Bruges, Belgium, 4 september 2012

• Somers B., P. Kempeneers, F. Devriendt, F. Vancoillie, K. Vandekerkhove &F. Morsdorf, Hyperforest: Advanced airborne hyperspectral

remote sensing to support forest management, Belgian Earth Observation Day 2012 – Bruges, Belgium, 5 september 2012.

• P. Kempeneers, K. Vandekerkhove, F. Devriendt, F. Van Coillie, Propagation of shadow effects on typical remote sensing applications in

forestry, proceedings Whispers 2013, June 2013, Gainsville USA.

• P. Kempeneers, F. Van Coillie, F. Devriendt, W. Liao, K. Vandekerkhove, Tree species mapping by combining hyperspectral with LiDAR

data, conference proceedings IGARSS 2014, July 2014, Quebec, Canada

• P. Kempeneers, F. Van Coillie, W. Liao, K. Vandekerkhove, Data fusion of airborne LiDAR and hyperspectral data for tree species

mapping, conference proceedings Forestsat 2014, November 2014, Riva del Garda, Italy

• Schneider, F. D.; Leiterer, R.; Morsdorf, F.; Gastellu-Etchegorry, J.-P.; Lauret, N.; Pfeifer, N. & Schaepman, M. E., Simulating imaging

spectrometer data: 3D forest modeling based on LiDAR and in situ data, Remote Sensing of Environment , 2014, 152, 235 - 250

28/11/2014 35 © 2012, VITO NV

• Schlerf, M., Rock, G., Lagueux, P., Ronellenfitsch, F., Gerhards, M., Hoffmann, L., & Udelhoven, T. (2012). A Hyperspectral Thermal

Infrared Imaging Instrument for Natural Resources Applications. Remote Sensing, 4(12), 3995-4009. doi: 10.3390/rs4123995

• Schlerf, M., & Atzberger, C. (2012). Vegetation Structure Retrieval in Beech and Spruce Forests Using Spectrodirectional Satellite Data.

IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 5(1), 8-17. doi: 10.1109/jstars.2012.2184268.

• Wandera L., Verhoef, W., Fauzi, A., Skidmore, A., Schlerf, M. (2014): Estimation of mangrove forest attributes using the Soil Leaf Canopy

(SLC) model applied to hyperspectral imagery. European Journal of Remote Sensing (under review).

• Wandera, L.N., W. Verhoef, A. Fauzi, M. Schlerf (2013). Chlorophyll estimation in mangrove forest for potential use as an indicator of

vegetation conditions, 6th Workshop on Remote Sensing of the Coastal Zone Matera, Italy.

• Wandera, L.N., W. Verhoef, A. Fauzi, M. Schlerf (2013). Chlorophyll estimation in mangrove forest as a potential indicator of vegetation

conditions, IECI 2013 conference, University of Trier.

• F. Devriendt, F. Van Coillie, R. De Wulf, L. Bertels, P. Kempeneers, K. Vandekerkhove and F. Morsdorf (2012). Classification of

unmanaged forest reserves in Flanders (Belgium) at the tree crown level using airborne hyperspectral and LiDAR data. Extended abstract

and oral presentation for GEOBIA 2012 conference, 7 – 9 May 2012, Rio de Janeiro, Brazil.

• F. Van Coillie, F. Devriendt and R. De Wulf (2012). Directional local filtering assisting individual tree analysis in closed forest canopies

using VHR optical and LiDAR data. Extended abstract and oral presentation for GEOBIA 2012 conference, 7 – 9 May 2012, Rio de

Janeiro, Brazil.

• F. Devriendt, F. Van Coillie, R. De Wulf, P. Kempeneers, K. Vandekerkhove and F. Morsdorf (2012). Synergy of hyperspectral & LiDAR

data for tree species mapping within an unmanaged closed forest reserve in Flanders, Belgium (2012). Abstract and poster for

FORESTSAT 2012 conference, 12 – 14 September 2012, Corvallo, Oregon, US.

• F. Devriendt, F. Van Coillie, R. De Wulf, P. Kempeneers, K. Vandekerkhove and F. Morsdorf (2013). Tree species mapping within

unmanaged closed forest reserves in Flanders (Belgium) using hyperspectral and LiDAR imagery to support forest management (2013).

Abstract for 8th EARSeL SIG Imaging Spectroscopy workshop, 8 – 10 April 2013, Nantes, France.

• F. Van Coillie, F. Devriendt, L.P.C. Verbeke and R. De Wulf (2013). Directional local filtering for stand density estimation in closed forest

canopies using VHR optical and LiDAR data. IEEE Geoscience and Remote Sensing Letters 10(4), 913-917.

• F. Devriendt, F. Van Coillie, R. De Wulf en K. Vandekerkhove (2012). Mogelijkheden van nieuwe aardobservatietechnieken voor het

bosbeheer in Vlaanderen. Bosrevue, 42, 12-16.

• M.L. Verdonck, F. Van Coillie, K. Vandekerkhove, P. Kempeneers en R. De Wulf (2014). Kan hyperspectrale remote sensing gebruikt

worden voor vitaliteitsbepaling van individuele bomen? Een verkennend onderzoek. Bosrevue (accepted)

• Van Coillie, F. M. B, Liao, W., Devriendt, F., Gautama, S., De Wulf, R. R., & Vandekerkhove, K., 2014, Effect of Hyperspectral Image

Denoising with PCA and Total Variation on Tree Species Mapping using Apex Data, Special-Thematic issue of the South-Eastern

European Journal of Earth Observation and Geomatics, 3 (2S), 281-286

Project output

28/11/2014 36 © 2012, VITO NV

HYPERFOREST project showed potential of LiDAR and HS data for forestry applications

» HS image pre-processing

» Forest parameter retrieval

» Species mapping and forest health assessment

In conclusion

hyperforest combining lidar and hyperspectral data to ... · remote sensing to support forest...

Documents