4 nov 2016 · RSS Remote Sensing Solutions GmbH 2016 Forest height monitoring and aboveground biomass variability in Indonesia's tropical forests
Studying deforestation has been an important topic in forestry research Especially, canopy classification using remotely sensed
Canopy spectral invariants for remote sensing and model applications Dong Huang sition of the solutions of some basic radiative transfer sub-
1 oct 2020 · Sample PhenoCam images of the canopy in a US tower site in 2006 on B: an Remote sensing can be part of the solution, as it can monitor
41763_3FSiegert1.pdf
© RSS Remote Sensing Solutions GmbH 2016
Forest height monitoring and
aboveground biomass variability in
Indonesia's tropical forests
Florian Siegert, Sandra Lohberger, Kristina Konecny, Matthias Stängel,
Werner Wiedemann, Uwe Ballhorn, Peter Navratil
GFOI R&D and GOFC-GOLD Land Cover
Science Meeting
The Hague, The Netherlands
31 October ² 4 November, 2016
© RSS Remote Sensing Solutions GmbH 2016
Tropical lowland forest in Indonesia
Most species-rich ecosystem in SE Asia
Mainly dipterocarps (60% endemic)
Forest height: 45 m, max: 70 m
© RSS Remote Sensing Solutions GmbH 2016
Land cover and land use 2013 (MoEF based on Landsat)
Lowland Dipterocarp forest
Montane Dipterocarp forest
Peat swamp forest
Heath forest (kerangas)
Mangrove forest
© RSS Remote Sensing Solutions GmbH 2016
High AGB variability (data from GlobBiomass)
Lowland Dipterocarp forest
AGB 200 -800 t/ha
Montane Dipterocarp forest
AGB 150 -300 t/ha
Peat swamp forest
AGB 10 -350 t/ha
Heath forest (kerangas)
AGB 5 -200 t/ha
Mangrove forest
AGB 70 -200 t/ha
© RSS Remote Sensing Solutions GmbH 2016
Low pole 10-100 t/ha
Medium tall
100-200 t/ha
Photo: F. Siegert
Tall
200-350 t/ha
Peat swamp forest
Natural variability of AGB
© RSS Remote Sensing Solutions GmbH 2016
Human induced variability of AGB
Intensive managed logging
Illegal logging
Disturbance by wildfires
Shifting cultivation
© RSS Remote Sensing Solutions GmbH 2016
Methodological approach (CARL research)
Field
inventory
LiDAR
Modelling
Modelling
SAR
Forest height and
biomass estimation
Large scale
biomass estimations
Drones
© RSS Remote Sensing Solutions GmbH 2016
Forest inventories and allometric modelling
Nested circular plots
Forest type - tree species ² DBH ²
(tree height)
Estimation of biomass and carbon
per ha using allometric models (Chave et al. 2005/2015)
© RSS Remote Sensing Solutions GmbH 2016
LiDAR acquisition and data products
3D point
cloud
Aerial image
True ortho image
© RSS Remote Sensing Solutions GmbH 2016
60 m
LiDAR acquisition and data products
© RSS Remote Sensing Solutions GmbH 2016
Digital terrain model
Lowland Dipterocarp Peat Swamp Forest
LiDAR acquisition and data products
Digital surface model
© RSS Remote Sensing Solutions GmbH 2016
SUMATRA
Overview of field and LiDAR data
Central Kalimantan
²165 field inventory plots
²6.000 km² full-coverage LiDAR
West Kalimantan (Kapuas Hulu)
²84 field inventory plots
²1200 km LiDAR transects
East Kalimantan (Berau & Malinau)
²78 field inventory plots
²1500 km LiDAR transects
South Sumatra
²115 field inventory and
biodiversity plots
²1850 km of LiDAR transects
Data collected by:
Bioclime
© RSS Remote Sensing Solutions GmbH 2016
LiDAR based aboveground biomass model
AGB model was created at 5m
spatial resolution (i.e. each pixel represents an area of 0.1ha)
Cell values were scaled to represent
aboveground biomass in tons per hectare
High aboveground biomass
variability within classes could be identified (e.g. Primary Dryland
Forest)
Areas with the highest
aboveground biomass were located around the Kerinci Sebelat National Park
Lowland
Dipterocarp
Peat swamp
forest
Mangrove
© RSS Remote Sensing Solutions GmbH 2016
Forest types in Lidar 3D point clouds
Lowland Dipterocarp
Peat swamp forest
Mangrove
© RSS Remote Sensing Solutions GmbH 2016
Forest degradation in Lidar 3D point clouds
Lowland Dipterocarp Peat swamp forest
undisturbed logged
Severely degraded by
recurrent fire
© RSS Remote Sensing Solutions GmbH 2016
LiDAR point cloud
Aerial photo
Illegal logging
Logging in Lidar 3D point clouds
© RSS Remote Sensing Solutions GmbH 2016
Representative samples of Lidar and field plots
Terrain
height
Canopy
height Tree height
© RSS Remote Sensing Solutions GmbH 2016
Canopy
Height
70 m
0 m 025m
Ø Terrain
Height:
40.0 m
AGB:
373.7 t/ha
Pristine Lowland Dipterocarp Forest
© RSS Remote Sensing Solutions GmbH 2016
Canopy
Height
70 m
0 m 025m
Plot ID:
01_06
Ø Terrain
Height:
196.5 m
AGB:
273.2 t/ha
Disturbed Lowland Dipterocarp Forest
© RSS Remote Sensing Solutions GmbH 2016
Canopy
Height
70 m
0 m 025m
Plot ID:
01_02
Ø Terrain
Height:
434.2 m
AGB:
374.8 t/ha
Sub-montane Dipterocarp Forest
© RSS Remote Sensing Solutions GmbH 2016
Canopy
Height
70 m
0 m 025m
Plot ID:
01_01
Ø Terrain
Height:
325.6 m
AGB:
166.3 t/ha
Disturbed Sub-montane Dipterocarp Forest
© RSS Remote Sensing Solutions GmbH 2016
LiDAR height metrics for AGB estimation
LiDAR height metrics:
Centroid Height (CH)
Quadratic mean canopy height (QMCH)
Published: Jubanski et al. 2013, Englhart et al. 2013
R2=0.84
© RSS Remote Sensing Solutions GmbH 2016
LiDAR transects show high AGB variability
Undisturbed
Lowland Dipterocarp
Logged in
2008/2009
Logged in
2011/2012
RapidEye
18.09.2012
© RSS Remote Sensing Solutions GmbH 2016
0 750
AGB [t/ha]
545 t/ha
300 t/ha
220 t/ha
Undisturbed
Lowland Dipterocarp
Logged in
2008/2009
Logged in
2011/2012
LiDAR transects show high AGB variability
© RSS Remote Sensing Solutions GmbH 2016
LiDAR transects to estimate AGB variability
Recent logging activities (2011/12)
RapidEye 18.09.2012
0 750
AGB [t/ha]
RapidEye 06.02.2014
LiDAR transects show high AGB variability
© RSS Remote Sensing Solutions GmbH 2016
Mean AGB:
155.7 t/ha
Orthophoto (10/2012) RapidEye (09/2012) AGB Model (10/2012) CHM (10/2012)
0 750 t/ha
0 65 m
LiDAR transects show high AGB variability
© RSS Remote Sensing Solutions GmbH 2016
RapidEye [4-5-3]
18.09.2012
0 750
AGB [t/ha]
LiDAR transects show high AGB variability
© RSS Remote Sensing Solutions GmbH 2016
Mean AGB:
176.8 t/ha
Trees up to 45 meters
tall Orthophoto (10/2012) RapidEye (09/2012) AGB Model (10/2012) CHM (10/2012)
0 750 t/ha
0 65 m
LiDAR transects show high AGB variability
© RSS Remote Sensing Solutions GmbH 2016
RapidEye [4-5-3]
18.09.2012
0 750
AGB [t/ha]
LiDAR transects show high AGB variability
© RSS Remote Sensing Solutions GmbH 2016
Mean AGB:
444.9 t/ha
Trees up to 55 meters
tall Orthophoto (10/2012) RapidEye (09/2012) AGB Model (10/2012) CHM (10/2012)
0 750 t/ha
0 65 m
LiDAR transects show high AGB variability
© RSS Remote Sensing Solutions GmbH 2016
22.05.2009
02.10.2010
21.06.2010
29.07.2012
RapidEye time series showing the progress of selective logging and fast regrowth.
Monitoring forest degradation
© RSS Remote Sensing Solutions GmbH 2016
CHM 2007 CHM 2011 DTM 2011
Monitoring forest degradation by repeated LiDAR
© RSS Remote Sensing Solutions GmbH 2016
CHM 2011
Monitoring forest degradation by repeated LiDAR
© RSS Remote Sensing Solutions GmbH 2016
Moitoring AGB change (decrease)
LiDAR 2007
LiDAR 2011
Rapideye 2009
Rapideye 2011
Englhart S., Jubanski J. & Siegert F. (2013). Quantifying dynamics in tropical peat swamp forest biomass with multi-temporal LiDAR datasets. Remote Sensing, 5, 2368²2388
© RSS Remote Sensing Solutions GmbH 2016
Orthophoto
Average CH
height: 15.5 m
Average CH
height: 17.3 m Gap:
Average height: 8.9 m
AGB=227 t/ha
Average height: 12.9 m
AGB=248 t/ha
Monitoring AGB change (increase)
© RSS Remote Sensing Solutions GmbH 2016
Methodological approach (CARL research)
Field
inventory
LiDAR
Modelling
Modelling
SAR
Forest height and
biomass estimation
Large scale
biomass estimations
Drones
© RSS Remote Sensing Solutions GmbH 2016
Drone data acquisition
https://vimeo.com/rssgmbh
© RSS Remote Sensing Solutions GmbH 2016
Canopy height models from drones and LiDAR
LiDAR 3D point cloud
Aerial photo and point cloud
acquired by drone
© RSS Remote Sensing Solutions GmbH 2016
SAR Interferometry ² a promising methodology
InSAR has the potential to measure tree height, which then can be used to estimate AGB in tropical forests
Research project:
Pol-InSAR 4 AGB ²Forest height retrieval and AGB estimation using polarimetric SAR Interferometry (TerraSAR-X, TanDEM-X, RADARSAT-2, Sentinel-1)
²Accuracy analysis
²Development of operational methods
funded by
© RSS Remote Sensing Solutions GmbH 2016
Methodological approach (CARL research)
Field
inventory
LiDAR
Modelling
Modellig
SAR
Forest height and
biomass estimation Large scale biomass estimations
Drone
© RSS Remote Sensing Solutions GmbH 2016
GlobBIOMASS - regional AGB estimation
ALOS PALSAR mosaic
²25 m spatial resolution
²acquired in 2009
²HH, HV polarization
²Acquired during dry season
(May-October)
SRTM
²30m spatial resolution
Additional data
²LiDAR AGB estimations
© RSS Remote Sensing Solutions GmbH 2016
R²=0.65
RMSE=86 t/ha
Multiple linear regression
model based on -backscatter -ratios -textures
GlobBIOMASS - regional AGB estimation
© RSS Remote Sensing Solutions GmbH 2016
AGB estimation up to 200 t/ha showing variability in lower and higher AGB ranges Change assessment and emission estimation feasible
GlobBIOMASS - regional AGB estimation
Logged forests
Logged forests
© RSS Remote Sensing Solutions GmbH 2016
AGB & carbon stock (CARL pre-operational)
Photo: F. Siegert
Field
inventory
LiDAR data acquisition
RapidEye imagery
Stratification
Modeling
Upscaling
Land cover classification
Carbon stock
map
© RSS Remote Sensing Solutions GmbH 2016
Determination of local AGB values
²Intersection of AGB model with land cover classification ²Different forest types and degradation stages ²Descriptive statistics for each class: Minimum, Maximum, Mean, Standard deviation
© RSS Remote Sensing Solutions GmbH 2016
CARL Framework Implementation
CARL Definition Supporting information Practicalities 1
Research
Basic principles
observed and technology concept formulated. Lowest level of readiness. Scientific research begins to be translated into applied research and development (R&D). Applications are speculative, and there may be no proof or detailed analysis to support the assumptions. Published research exists that identifies the principles that underlie the technology. Promising case studies exist.
2 Analytical and
experimental function (calibration) and/or proof of concept (validation). Active R&D is initiated. This includes analytical and laboratory studies to measure parameters of interest. Method still needs to be tested as repeatable or applicable in the REDD+ MRV context. Validation data may not be available. 3
Pre-operational
Demonstration in small-
scale environment. The basic technological components are integrated with reasonably realistic supporting elements so they can be tested in a small-scale environment. End-to-end processing demonstrated. Data processing methods have been documented in peer review publications. Methods have been assessed for applicability in different forest monitoring contexts.
4 Demonstration in a
larger-scale environment. Representative model or prototype method (near the desired performance), which is well beyond that of level 3, is tested in a larger-scale environment. Processing workflows and methods demonstrated in large scale processing systems or national programs. Basic data may not necessarily yet be available for routine monitoring. 5
Operational
Demonstration in an
operational environment (national level).
Represents the end of true method development.
Technology/method has been proven to work in its
final form. Application of the technology/method in its final form with at least one MRV cycle completed. Product/technology/method fits within a context or system, optimised to meet monitoring requirements with documented uncertainties. Processing workflows and methods are robust and repeatable. Methods are demonstrated to be applicable in forest monitoring contexts. Documented peer reviewed guidance is available. Core data is available for monitoring at relevant spatial and temporal resolutions, and scales. Includes a clear pathway for continuous improvement.
6 Actual technology
proven through successful use in operational environments. Application of the technology/method in its final form with at least two MRV cycle completed. Method implemented in country(ies) as part of national environmental and forest monitoring programs. TTE Technical Assessment or Technical Analysis process, or equivalent, completed successfully. Advantages and trade-offs of the method are well documented, notably in terms of adequacy, technical complexity, performance, and costs. Includes a clear pathway for continuous improvement.
© RSS Remote Sensing Solutions GmbH 2016
7OMQN \RX IRU \RXU MPPHQPLRQ¬
© RSS Remote Sensing Solutions GmbH 2016
Comparison with other AGB maps
SAR AGB model Baccini Saatchi
LiDAR
Central
Kalimantan
© RSS Remote Sensing Solutions GmbH 2016
With FORCLIME Germany supports
Indonesia's efforts to reduce
greenhouse gas emissions from the forestry sector, and to implement sustainable forest management.
Advice REDD+ and forest
development at national, provincial and district levels
Technical advice on the
implementation of REDD+
Funding: 28 Mio Euro
© RSS Remote Sensing Solutions GmbH 2016
Local aboveground biomass values
Adapted from local district level