An overview of Lidar remote sensing of forests

An overview of Lidar remote sensing of forests C. Véga French Institute of Pondicherry

Outline q Principle and History q Systems and Platform q Data processing / Forestry q Airborne discrete Lidar q Terrestrial Lidar e Forests - National Workshop on Info Systems for Decision Making in Forestry 9, 10 & 11 th May 2013 Bangalore

What is Lidar ? q LIght Detection And Ranging or Laser Scanning q Active remote sensing measuring distance to target based on « time of flight » R = range t = time C = light speed ©Calypso, CNES, 2006 e Forests - National Workshop on Info Systems for Decision Making in Forestry 9, 10 & 11 th May 2013 Bangalore

History q Sixties : Airborne laser for measuring flight altitude q Seventies – Eighties : Airborne profiling systems (topography and forestry) q Nineties: Scanning systems with GPS and INS -> Georeferencing q 2000 ongoing : Industrial development – costs reduction e Forests - National Workshop on Info Systems for Decision Making in Forestry 9, 10 & 11 th May 2013 Bangalore

Systems q Full-waveform systems Record the complete range of energy reflected by surfaces q Discrete systems Record 1 up to N returns by emitted pulse q Scanning > 300 k. Hz Precision : 1 m xy; 0. 1 m z e Forests - National Workshop on Info Systems for Decision Making in Forestry 9, 10 & 11 th May 2013 Bangalore

Platforms ALTITUDE 500 - 1000 km SATELLITES (GLAS- 600 km / CALIOP- 705 km) High Altitude Planes (SLICER) 8 - 12 km 1200 - 3500 m 100 - 1000 m Mean Altitude Planes HELICOPTERS Low Altitude (corridor mapping 50 -150 m) Ground or Terrestrial Lidar e Forests - National Workshop on Info Systems for Decision Making in Forestry 9, 10 & 11 th May 2013 Bangalore

Data acquisition q Small Footprint Airborne Lidar e Forests - National Workshop on Info Systems for Decision Making in Forestry 9, 10 & 11 th May 2013 Bangalore

Data Visualisation q Small Footprint Airborne Lidar 833 m 890 m Draix, France) e Forests - National Workshop on Info Systems for Decision Making in Forestry 9, 10 & 11 th May 2013 Bangalore

Data Visualisation q Small Footprint Airborne Lidar e Forests - National Workshop on Info Systems for Decision Making in Forestry 9, 10 & 11 th May 2013 Bangalore

Data Visualisation q Terrestrial Lidar e Forests - National Workshop on Info Systems for Decision Making in Forestry 9, 10 & 11 th May 2013 Bangalore

Point cloud Processing q Discrete Airborne Laser Scanning (ALS) q Small Scale parameter estimation -> Plot Level q Large Scale parameter estimation -> Tree Level q Terrestrial Laser Scanning (TLS) q Stem characterization q Tree architecture e Forests - National Workshop on Info Systems for Decision Making in Forestry 9, 10 & 11 th May 2013 Bangalore

Preprocessing Raw point cloud DTM 833 m 890 m 21 m 0 m Normalized point cloud = Raw - DTM e Forests - National Workshop on Info Systems for Decision Making in Forestry 9, 10 & 11 th May 2013 Bangalore

Forest Parameters q Estimating Field parameters from Lidar parameters Calibration Field = Function (Lidar) Inversion q Multiplicative models q Stepwise approach e Forests - National Workshop on Info Systems for Decision Making in Forestry 9, 10 & 11 th May 2013 Bangalore

Small Scale Mapping q Volume estimation (Naesset, 2005) Volume estimated per grid cell Summed by stand -> mean/ha Photo interpretation Grid Lidar Field Plots Terrain + Lidar e Forests - National Workshop on Info Systems for Decision Making in Forestry 9, 10 & 11 th May 2013 Bangalore

Large Scale Mapping q Tree-based approaches - Segmentation methods q Local maxima extraction on raster + polygon fitting (Popescu et al. , 2003, 2004) e Forests - National Workshop on Info Systems for Decision Making in Forestry 9, 10 & 11 th May 2013 Bangalore

Large scale mapping q Tree-based approaches - Segmentation methods q Direct segmentation of the point cloud Lateral view Top view e Forests - National Workshop on Info Systems for Decision Making in Forestry 9, 10 & 11 th May 2013 Bangalore

Individual tree approaches q Direct estimation of tree density and tree parameters q Improving equations for volume and biomass (height and crown dimension) q Crown dimension explain better AGB (Popescu 2003) q Stem to stem management -> thinning e Forests - National Workshop on Info Systems for Decision Making in Forestry 9, 10 & 11 th May 2013 Bangalore

Terrestrial lidar q Limited to small surfaces (Plots) q Very high density (mm) q Utility for allometry, tree architecture and forest modeling e Forests - National Workshop on Info Systems for Decision Making in Forestry 9, 10 & 11 th May 2013 Bangalore

Terrestrial Lidar q Stem Characterization § Automatic Stem Extraction (PCA- Hough) (Bac et al. 2013) e Forests - National Workshop on Info Systems for Decision Making in Forestry 9, 10 & 11 th May 2013 Bangalore

Terrestrial Lidar (Bac et al. 2013) e Forests - National Workshop on Info Systems for Decision Making in Forestry 9, 10 & 11 th May 2013 Bangalore

Terrestrial Lidar q Tree architecture § L-Architect (Côté et al. 2011) e Forests - National Workshop on Info Systems for Decision Making in Forestry 9, 10 & 11 th May 2013 Bangalore

Potential for Indian Forestry q Measuring biomass -> issue in complex tropical forests q Conventional remote sensing -> signal saturation at low AGB q Lidar q Directly related to forest structure q No saturation with AGB q Best current data for plot and landscape estimation of forest parameters q Utility for calibrating texture indices from satellites images for ABG estimations at regional level e Forests - National Workshop on Info Systems for Decision Making in Forestry 9, 10 & 11 th May 2013 Bangalore

Variety of applications… Geomorphology Archeology Habitat Mapping Bird Erosion / Flooding Angkor ruins under the forest canopy (Chase and al. , 2010) e Forests - National Workshop on Info Systems for Decision Making in Forestry 9, 10 & 11 th May 2013 Bangalore

Thank you !

Forest Parameter Estimation q Plot-based Approach N Lidar Plots Statistical descriptors N Field Plots Regression analysis Validation Large scale mapping e Forests - National Workshop on Info Systems for Decision Making in Forestry 9, 10 & 11 th May 2013 Bangalore

Point Classification q Example for an ALS system recording 2 returns q Issue: Point penetration within canopy First Return Vegetation Last Return Ground e Forests - National Workshop on Info Systems for Decision Making in Forestry 9, 10 & 11 th May 2013 Bangalore

Point Classification Unique return = Ground (First= Last) e Forests - National Workshop on Info Systems for Decision Making in Forestry 9, 10 & 11 th May 2013 Bangalore

Point Classification First Return Vegetation Last Return Vegetation e Forests - National Workshop on Info Systems for Decision Making in Forestry 9, 10 & 11 th May 2013 Bangalore

Point Classification q Classification algorithms : extracting ground points q Lot of approaches and algorithms q Best one Iterative Tin – Delauney triangulation 3 D points Local minima Initial TIN Surface TIN Densification Angle & Distance ©F. Bretar, 2006 Axelsson (1999) e Forests - National Workshop on Info Systems for Decision Making in Forestry 9, 10 & 11 th May 2013 Bangalore

The Big Picture Forest tpye Biomass Texture DART Images (AMAP – CESBIO) Architecture Allometry Porosity Dynamic Fonctionning Height A-Lidar Model T-Lidar dbh Structure Biomass Dynamics