LiDAR based sampling for subtle change, developments, and status

Transcription

1 LiDAR based sampling for subtle change, developments, and status Erik Næsset Norwegian University of Life Sciences, Norway

2 Conclusions: 1. LiDAR is an extremely precise tool for measuring forest canopy structure at any geographical scale from fractions of individual trees to regions and nations 2. LiDAR has thus a significant role to play in change detection and monitoring 3. LiDAR is probably the most precise remote sensing technique available for forest assessment

3 Properties and potentials of airborne LiDAR data DEPARTMENT OF ECOLOGY AND NATURAL RESOURCE MANAGEMENT

4 Change and development at different geographical scales: 1. Applying LiDAR for sampling at regional and national levels (~10,000-1,000,000 km 2 ). Biomass

5 Change and development at different geographical scales: 1. Applying LiDAR for sampling at regional and national levels (~10,000-1,000,000 km 2 ). Biomass 2. Wall-to-wall inventory and monitoring at district level (~100-10,000 km 2 ). Resolution: ~100 m 2 Applications: forest resources biomass defoliation gap dynamics 100 m

6 Change and development at different geographical scales: 1. Applying LiDAR for sampling at regional and national levels (~10,000-1,000,000 km 2 ). Biomass 2. Wall-to-wall inventory and monitoring at district level (~100-10,000 km 2 ). forest resources biomass defoliation gap dynamics 3. Single-tree level biomass recruitment, growth, mortality defoliation Tree #14 Foliage area: 44 m 2 Crown volume:126 m 3

7 Change and development at different geographical scales: 1. Applying LiDAR for sampling at regional and national levels (~10,000-1,000,000 km 2 ). Biomass 2. Wall-to-wall inventory and monitoring at district level (~100-10,000 km 2 ). forest resources biomass defoliation gap dynamics 3. Single-tree level biomass recruitment, growth, mortality defolition

8 Recruitment, growth, mortality of small trees above tree line Laserdata: 7.8 p/m 2 Field data: - Point-centered quad sampling - Plot size: 25 m radius - Up to 4 trees selected per quadrant (<1 m, 1-2 m, 2-3 m, >3 m) Tree #157 H=0.67 m h max =0.11 m

9 Recruitment, growth, mortality of small trees above tree line Probability of tree identification (h max >0 m) Probability of tree identification Spruce 0.5 Pine 0.4 Birch Crown diameter (m)

10 Recruitment, growth, mortality of small trees above tree line Accuracy of tree heights measured with laser Laser-derived tree height (m) Spruce Pine Birch 1:1 line Field-measured tree height (m)

11 Recruitment, growth, mortality of small trees above tree line Points of discussion: 1. Commission and omission errors 2. Inventory versus monitoring Tree #153 Obj. #217 Obj. #329 H=0.39 m h max =0.06 m H=0.35 m h max =0.13 m H=0.51 m h max =0.45 m

12 Recruitment, growth, mortality of small trees above tree line Norway, N (1,500 km) Laser scanner data summer 2006 Repeated acquisition 2010/11 Hundreds of sub-alpine/alpine gradients Pulse density: 5-10 p/m 2 Swath width: m

13 Recruitment, growth, mortality of small trees above tree line PCQ samples of small trees on 50 sites along the 1,500 km transect Re-measured after 5 years Recruitment, growth, mortality Calibration of LiDAR-based change detection LiDAR-based change detection

14 Tree line delineation according to FAO definitions DEPARTMENT OF ECOLOGY AND NATURAL RESOURCE MANAGEMENT

15 Estimating changes in mountain forest carbon pools Objectives: Evaluate the accuracy of airborne LiDAR for above ground biomass estimation in an alpineforest ecotone Evaluate the accuracy of estimated change (growth) in biomass over a 4 yr time span Data: Laser scanner data early summer 2005 and late summer 2008 Field data from 55 plots (growth and yield experiments) 6 different species Field plots measured in 2005 and re-measured in 2008

16 Estimating changes in mountain forest carbon pools Laser data calibration DEPARTMENT OF ECOLOGY AND NATURAL RESOURCE MANAGEMENT

17 Estimating changes in mountain forest carbon pools Preliminary results 2005 data Above-ground biomass: Range: tons/ha Mean: 11.9 tons/ha Biomass LiDAR-relationships: One species (n=28) lnbio=b 0 +b 1 lnh mean +b 2 lnd 0 R 2 =0.95

18 Regional changes in mountain forest carbon pools Norway, N (1,500 km) LIDAR data: Laser scanner data summer 2006 Repeated acquisition 2010/11 Field data: Hundreds of permanent NFI plots Permanent NFI plots and additional plots in the mountain forest established on 50 sites along the transect

19 Regional changes in mountain forest carbon pools Mountain forest plots DEPARTMENT OF ECOLOGY AND NATURAL RESOURCE MANAGEMENT

20 Defoliation: LiDAR for mapping of forest damage Pine sawfly (Neodiprion sertifer), Norway, 2005 DEPARTMENT OF ECOLOGY AND NATURAL RESOURCE MANAGEMENT

21 Defoliation: LiDAR for mapping of forest damage Idea: To use change in LAI as an indicator of defoliation Data: LiDAR data from throughout the growing season LICOR LAI-2000 field measurements Results: LAI = * ln(n a /N b ), R 2 =0.93 LAI LAI Time point 1: May 2: July 3: August LIDAR: ln(n a /N b )

22 Defoliation: LiDAR for mapping of forest damage Heavy Light No damage LiDAR-predicted change in LAI Change in NDVI, MODIS data

23 Biomass estimation in dense forests The saturation problem Estimated biomass (tons/ha) LiDAR TROPICAL Drake et al. (2003) BOREAL Næsset & Gobakken (2008) CONIFER Nelson et al. (2007) CONIFER Hall et al. (2005) CONIFER Lefsky et al. (2002) CONIFER Means et al. (1999) CONIFER Lefsky et al. (1999) DECIDUOUS Lefsky et al. (2002) RaDAR Optical True biomass (tons/ha)

24 Influence of sensor and flight specifications Different instruments flying altitudes footprint sizes pulse energy pulse width will influence on the data. Need for repeated ground calibration in monitoing.