Woody Vegetation (2015)

New imaging processes to describe fundamental properties of forests for manager and state of environment reporting.

Completed in May 2015, the Woody Vegetation Project (2.07) produced tools and procedures to auto-generate landscape level woody vegetation features, such as spatial layers, from field and remote sensing woody vegetation data. The metrics are assessed to inform carbon accounting, biodiversity and ecosystem health, and fire management.

Download a copy of the project summary and outputs here. 

The open source tools have been developed for sustainable land management decision-making and monitoring, mapping and natural resource management activities.

Both PhD students completed and submitted their studies which once published will be available in the CRCSI library.

State and federal land managers have a mandate to map and report on Australian native woody vegetation. The achievements of this project are the ability to characterise woody vegetation ecosystems using automated feature generation using ground, airborne and satellite image and ranging data.

selection of data primitives

Data primitives in the context of the study area are defined as a set of landscape metrics that are functional descriptors of woody vegetation established through international policy and protocols. These are:

• The metric selected for international directives were identified
• A preliminary priority list was created through a survey of Australian and New Zealand stakeholders from education, research and industry sectors
• Final set of metrics was selected in a workshop held with Australian federal and state agency representatives.

The end users of this work in Australia determined the metrics needed to be:

• Scalable up to the landscape level 
• Easily utilised in Australian sclerophyll environments.

Please read more in the following presentations and publications:

1. Suarez et al., CRCSI 2.07 WS 2013
2. CRCSI 2.07 Deliverable 1
3. Axelsson et al., 2012

protocols and guidelines 

Protocols and guidelines for field and image data collection and processing have been provided. In some cases, those guidelines have been compiled from management agencies and research groups. In others, specific guidelines have been created when appropriate ones did not exist.

Python tools have been developed for assessing forest vertical structural complexity. The code is ForestLAS. A full list of supporting documents that form the protocols and guidelines can be found here: Protocols and Guidelines.

For direct access to the software and guidelines, please click here. 

ground-based assessment

In the ground-based assessment, a comparison of common sampling designs for Gap probability/Leaf Are Index ground truth measurements was carried out.

Please read more in the following presentations and publications:

1. Woodgate et al., 2013
2. Woodgate et al., GSR2 2012

A comparison of LAI ground-based assessment method performance was done over 11 sites along the South-East coast of Australia. Uncertainties computed from the method-to-method comparisons are higher than error threshold expected for EO products that use this data as validation.

Please read more in the following presentations and publications:

1. Woodgate et al., IGARSS 2013
2. Woodgate et al., CRCSI 2013
3. Woodgate et al., 2015
4. Woodgate et al., IGARSS 2013

One of the study sites has been reconstructed using 3D modelling to investigate LAI ground-based retrieval methods.

Results show the accurate measure of canopy component clumping and leaf angle distribution has a strong impact in the LAI retrieval accuracy.

Moreover, a new element has been added to the Gap fraction method for LAI estimation (e.g. Woody element angular distribution).

Please read more in the following presentations and publications:

1. Woodgate et al., CRCSI 2014
2. Woodgate et al., in preparation
3. Woodgate PhD thesis

Airborne-based Assessment

Airborne discrete return LiDAR imagery was used to derive canopy structural attributes: Height, canopy cover and number of canopy strata. Estimates over the three sites fully covered the existing range of variation in Victorian sclerophyll forests.

In order to guide land management agencies in the process of designing airborne data acquisition, a study of the return density requirements for the assessment of this metric was also performed and can be found in Wilkes et al., 2015.

Please read more in the following presentations and publications:

1. Wilkes et al., Forestsat 2014
2. Suarez et al., ANZIF 2015
3. Wilkes et al., CRCSI 2013
4. Wilkes et al., 2015 (PERS, in press)
5. Wilkes et al., submitted to MEE

satellite-based assessment

Ensemble regression methods were used to up-scale LiDAR-based estimates for large area assessment. Results show the methods assess the existing heterogeneity. Higher errors were found for highest and lowest estimate values as expected. Errors in canopy height ranged from 0 to 10 m and reached 7% in canopy cover estimation.

Please read more in the following presentations, publications and reports:

1. Suarez et al., ANZIF 2015
2. Suarez et al., (Draft)
3. Wilkes et al., submitted to MEE

landscape feature extraction

An automatic classification was performed using isodata tool where the output class number is undetermined (therefore applicable at different scales).

The classification followed by a removal of features under a minimum management unit area demonstrated useful for landscape feature generation.

This is an ongoing research but preliminary results can be found in the following presentations, publications and reports:

1. Suarez et al., ANZIF 2015
2. Suarez et al., (Draft)
3. CRCSI 2.07 newsletter
4. Suarez et al., (in preparation)

Project participants

Four organisations took part in this project with RMIT being the lead agency. The remaining agencies where: DELWP (Victoria), DSITIA (Queensland), and DPI (NSW). The project was supported by a research team of two project leaders, a post-doctoral fellow and two PhD students.

contacts

Prof Simon Jones

Project Leader

Email

 

Dr Andrew Haywood

Project Leader

Email

 

Dr Lola Suarez

Post-Doctoral Fellow

Email

 

Phil Wilkes

PhD Student

Email

 

Will Woodgate

PhD Student

Email