Presentation is loading. Please wait.

Presentation is loading. Please wait.

Abteilung Biometrie und Informatik SMC Spring Meeting 2007, Vancouver, WA. Mapping Forest Characteristics across the Landscape using Sample Plot and Airborne.

Similar presentations


Presentation on theme: "Abteilung Biometrie und Informatik SMC Spring Meeting 2007, Vancouver, WA. Mapping Forest Characteristics across the Landscape using Sample Plot and Airborne."— Presentation transcript:

1 Abteilung Biometrie und Informatik SMC Spring Meeting 2007, Vancouver, WA. Mapping Forest Characteristics across the Landscape using Sample Plot and Airborne Laser Scanning Data Johannes Breidenbach Forest Research Institute of Baden-Württemberg (Forstliche Versuchs- und Forschungsanstalt Baden-Württemberg)

2 Abteilung Biometrie und Informatik SMC Spring Meeting 2007, Vancouver, WA. Outline Introduction Lidar Methods Plotwise estimation of –Lorey’s height –Volume, Biomass –Diameter distributions

3 Abteilung Biometrie und Informatik SMC Spring Meeting 2007, Vancouver, WA. Systematic sample plot inventory Introduction

4 Abteilung Biometrie und Informatik SMC Spring Meeting 2007, Vancouver, WA. r = 3 m; > 10 cm r = 6 m; > 15 cm r =12 m; > 30 cm Introduction r = 2 m; > 7 cm

5 Abteilung Biometrie und Informatik SMC Spring Meeting 2007, Vancouver, WA. Mapping of forest characteristics with Lidar data Introduction 1.Calibration of regression models (Lidar+sample plots) 2.Application of regression models (Lidar only)

6 Abteilung Biometrie und Informatik SMC Spring Meeting 2007, Vancouver, WA. Mapping of forest characteristics with Lidar data Introduction 1.Calibration of regression models (Lidar+sample plots) 2.Application of regression models (sample plots)

7 Abteilung Biometrie und Informatik SMC Spring Meeting 2007, Vancouver, WA. Outline Introduction Lidar Methods Plotwise estimation of –Lorey’s height –Volume, Biomass –Diameter distributions

8 Abteilung Biometrie und Informatik SMC Spring Meeting 2007, Vancouver, WA. Light detection and ranging (Lidar) Source: TopoSys Lidar

9 Abteilung Biometrie und Informatik SMC Spring Meeting 2007, Vancouver, WA. Digital Terrain Model Lidar

10 Abteilung Biometrie und Informatik SMC Spring Meeting 2007, Vancouver, WA. Digital Surface Model Lidar

11 Abteilung Biometrie und Informatik SMC Spring Meeting 2007, Vancouver, WA. Outline Introduction Lidar Methods Plotwise estimation of –Lorey’s height –Volume, Biomass –Diameter distributions

12 Abteilung Biometrie und Informatik SMC Spring Meeting 2007, Vancouver, WA. Two Approaches to analyse Lidar Data High resolution data (> 1 beam/m²) –Single tree approach –By now: mainly scientific studies Low resolution data (< 1 beam/m²) –Plotwise approach –Robust method Methods

13 Abteilung Biometrie und Informatik SMC Spring Meeting 2007, Vancouver, WA. Methods

14 Abteilung Biometrie und Informatik SMC Spring Meeting 2007, Vancouver, WA. Min Max Median 3. Quartile 1. Quartile Lidar vegetationheight [m] Methods Height metrics d5 d4d3d2 d1 Density metrics Mean

15 Abteilung Biometrie und Informatik SMC Spring Meeting 2007, Vancouver, WA. Additional covariates Coniferous proportion (CP) = (CHM F - CHM L )/ CHM F > 0.3 Crown cover (CC) = CHM F > 1 m Methods

16 Abteilung Biometrie und Informatik SMC Spring Meeting 2007, Vancouver, WA. Outline Introduction Lidar Methods Plotwise estimation of –Lorey’s height –Volume, Biomass –Diameter distributions

17 Abteilung Biometrie und Informatik SMC Spring Meeting 2007, Vancouver, WA. Incidents effecting height estimates and Lidar metrics CP = Crown shape Stem density (TPH) Slope  Density metrics couldn‘t substitute the influence of slope and TPH Lorey‘s height

18 Abteilung Biometrie und Informatik SMC Spring Meeting 2007, Vancouver, WA. Lorey‘s height

19 Abteilung Biometrie und Informatik SMC Spring Meeting 2007, Vancouver, WA. Outline Introduction Lidar Methods Plotwise estimation of –Lorey’s height –Volume, Biomass –Diameter distributions

20 Abteilung Biometrie und Informatik SMC Spring Meeting 2007, Vancouver, WA. E(μ) = Mean.L + Mean.L * CP + Mean.L * CC + d7 E(σ) = Mean.L R² 0.6; RMSE 120 m³/ha; 60 Mg/ha (~28%) Estimation of Volume and Biomass Vol./Biomass

21 Abteilung Biometrie und Informatik SMC Spring Meeting 2007, Vancouver, WA. Nature reserve Ridis Area: 31,5 ha Estimated biomass (inventory): 8491 t, 300 t/ha, 8% Error 1 Estimated biomass (Lidar): 8681 t, 276 t/ha, 2% Error 1 Area: 18,5 ha Estimated biomass (inventory): 3796 t, 186 t/ha, 17% Error 2 Estimated biomass (Lidar): 4082 t, 221 t/ha, 3% Error 2 Area: 1 ha Estimated biomass (Lidar): 329 t, 10% Error 1,2 difference not significant Vol./Biomass

22 Abteilung Biometrie und Informatik SMC Spring Meeting 2007, Vancouver, WA. Estimations using mixed effects models Vol./Biomass

23 Abteilung Biometrie und Informatik SMC Spring Meeting 2007, Vancouver, WA. Vol./Biomass Bias decreases, variance increases, hardly any change in RMSE

24 Abteilung Biometrie und Informatik SMC Spring Meeting 2007, Vancouver, WA. Outline Introduction Lidar Methods Plotwise estimation of –Lorey’s height –Volume, Biomass –Diameter distributions

25 Abteilung Biometrie und Informatik SMC Spring Meeting 2007, Vancouver, WA. Estimation of diameter distributions Lidar vegetationheight [m] DBH DBH [cm]

26 Abteilung Biometrie und Informatik SMC Spring Meeting 2007, Vancouver, WA. DBH

27 Abteilung Biometrie und Informatik SMC Spring Meeting 2007, Vancouver, WA. DBH

28 Abteilung Biometrie und Informatik SMC Spring Meeting 2007, Vancouver, WA. Future work Influence of registration errors Integration into existing systems and procedures (IRIS)

29 Abteilung Biometrie und Informatik SMC Spring Meeting 2007, Vancouver, WA. Thank you for your attention!

30 Abteilung Biometrie und Informatik SMC Spring Meeting 2007, Vancouver, WA. Two Approaches to analyse Lidar Data High resolution data (> 1 beam/m²) Low resolution data (< 1 beam/m²) Methods

31 Abteilung Biometrie und Informatik SMC Spring Meeting 2007, Vancouver, WA. Two Approaches to analyse Lidar Data High resolution data (  1 beam/m²) –Suited for single tree approach = delineation of single crowns –Statistical relations between Crown properties (e.g. crown diameter), tree height (laser derived) and Volume, diameter (terrestrial, single tree related ground truthing)  Conifer dominated, open forests, high costs for georeferencing of single trees  By now primarily for scientific purposes Methods

32 Abteilung Biometrie und Informatik SMC Spring Meeting 2007, Vancouver, WA. High resolution data Methods

33 Abteilung Biometrie und Informatik SMC Spring Meeting 2007, Vancouver, WA. Two Approaches to analyse Lidar Data Low resolution data (  1 beam/m²) –Plot level approach –Satistical relations between Vegetation height distributions (laser derived) and Sample plot volume, plot mean tree height, plot biomass, plot diameter structure (terrestrial, e.g. FFI)  Robust method  Use of available standard inventory data for calibration Methods

34 Abteilung Biometrie und Informatik SMC Spring Meeting 2007, Vancouver, WA. Material ~2000 inventory sample plots –Volume 420..1200 m³/ha –Biomass 220..550 Mg/ha –Lorey‘s height 30..42 m –Spruce, Beech, Maple and others ~150 km² Lidar data –Optech ALTM 1225 –1.5 m point spacing (low density) –Leaf-off condition Material

35 Abteilung Biometrie und Informatik SMC Spring Meeting 2007, Vancouver, WA.

36 Abteilung Biometrie und Informatik SMC Spring Meeting 2007, Vancouver, WA. Results log(μ) = β μ0 + β μ1 · 3rd Qu + β μ2 · 1st Qu · 3rd Qu log(σ) = β σ0 + β σ1 · 1st Qu + β σ1 · 1st Qu · 3rd Qu DBH


Download ppt "Abteilung Biometrie und Informatik SMC Spring Meeting 2007, Vancouver, WA. Mapping Forest Characteristics across the Landscape using Sample Plot and Airborne."

Similar presentations


Ads by Google