australian government - data.daff.gov.audata.daff.gov.au/data/...mapping-workshop_proc_11.pdf ·...

Workshop proceedings Canberra 23–24 November 2009

Jane Stewart, Jasmine Rickards, Vivienne Bordas, Lucy Randall and Richard Thackway

March 2011

Ground cover monitoring for Australia – establishing a

nationally coordinated approach to ground

cover mapping

Australian Bureau of Agricultural and Resource Economics and Sciences

Australian Government

www. aba r e s . g ov. a u

S c i e n c e a n d e c o n o m i c s f o r d e c i s i o n - m a k e r s

ii

© Commonwealth of Australia 2011

This work is copyright. The Copyright Act 1968 permits fair dealing for study, research, news reporting, criticism or review. Selected passages, tables or diagrams may be reproduced for such purposes provided acknowledgment of the source is included. Major extracts or the entire document may not be reproduced by any process without the written permission of the Executive Director, Australian Bureau of Agricultural and Resource Economics and Sciences (ABARES).

The Australian Government acting through ABARES has exercised due care and skill in the preparation and compilation of the information and data set out in this publication. Notwithstanding, ABARES, its employees and advisers disclaim all liability, including liability for negligence, for any loss, damage, injury, expense or cost incurred by any person as a result of accessing, using or relying upon any of the information or data set out in this publication to the maximum extent permitted by law.

ISBN 978-1-921448-89-8

Stewart, JB, Rickards, JE, Bordas, VM, Randall LA and Thackway, RM 2011, Ground cover monitoring for Australia – establishing a coordinated approach to ground cover mapping: Workshop proceedings Canberra 23–24 November 2009, ABARES, Canberra, March.

Australian Bureau of Agricultural and Resource Economics and Sciences Postal address GPO Box 1563 Canberra ACT 2601 Australia Switchboard +61 2 6272 2010 Facsimile +61 2 6272 2001 Email [email protected] Web abares.gov.au

ABARES project 43093

The Australian Bureau of Agricultural and Resource Economics and Sciences (ABARES) was formed following the merger of the Australian Bureau of Agricultural and Resource Economics (ABARE) and the Bureau of Rural Sciences (BRS) in 2010–11.

AcknowledgmentsThe authors thank all participants who presented and/or attended the Canberra workshop (23–24 November 2009), particularly for their support for a nationally coordinated approach to ground cover monitoring. This group has continued to contribute its expertise through on-line discussions, teleconferences and, now, project delivery. The activities established as a result of the workshop recommendations reflect the collaboration between State, Territory and Australian Government agencies undertaking ground cover monitoring. The authors thank John Leys and Michele Barson for their contribution to Appendix C and Tim McVicar for his contribution to Appendix D. The workshop and this report were funded by the Australian Government Department of Agriculture, Fisheries and Forestry.

Cover image: Chenopod shrublands, southern rangelands, South Australia (Rob Lesslie, ABARES).

iii

ContentsSummary 1

Acronyms and initialisms 2

1 Introduction 3

2 Options for monitoring ground cover using remote sensing 6

3 Methods to calibrate and validate remotely sensed ground cover 10

4 A coordinated approach for ground cover mapping 12

5 Working group objectives 13

Working group 1: Fractional cover mapping operational for the nation 13Working group 2: Upscaling 13Working group 3: Field data collection 14Working group 4: Monitoring land management practices with remote sensing 14

6 Workshop recommendations 15

Working group 1: Fractional cover mapping operational for the nation 15Working group 2: Upscaling 15Working group 3: Field data collection 17Working group 4: Monitoring land management practices with remote sensing 19

7 Implementing the workshop recommendations 21

Appendixes A Workshop presentations 26

B Workshop participants 27

C Remote sensing specifications for monitoring ground cover nationally 29

D Remotely sensed examples of upscaling 34

E SLATS generic site data form 36

References 42

Ground cover monitoring for Australia ABARES Workshop proceedings

iv

Tables1 Key tasks to deliver an operational remotely sensed fractional cover nationally

(2010–2013) 22

2 Key tasks in the validation of remotely sensed fractional cover (2010–2013) 24

Figures1 The importance of ground cover in minimising (a) water erosion and (b) wind erosion 4

2 Theoretical basis of the Guerschman et al. (2009) algorithm considers the spectral characteristics of photosynthetic vegetation (chlorophyll at 600–800 nanometres), non-photosynthetic vegetation (cellulose and lignin at 2000–2200 nanometres) and bare soil quantified by the Normalised Difference Vegetation Index (NDVI) and the cellulose absorption index (CAI). 7

3 The four working groups and their relationship to each other for delivery of nationally coordinated ground cover mapping 12

4 Layout of transects used in Queensland’s Statewide Landcover and Trees Study (SLATS) modified discrete point sampling method for measuring fractional cover in (a) pastoral environments and (b) agricultural crops sown in lines 17

5 A remotely sensed upscaling method to spatially extend data beyond the isolated field measurement sites 34

Boxes1 Cover and remote sensing 5

2 Characterisation of satellite remote sensing systems 6

3 Upscaling 14

4 Calibration and validation of remotely sensed data 18

Maps1 Results from the 2007–08 ARM Survey showing graziers with a minimum ground

cover target 4

2 Mean fractional vegetation cover for 2000–2009 8

3 Dry season ground cover and foliage projective cover (FPC) for the Fox Creek sub-catchment within the Burdekin catchment, Queensland, in 2002 using Landsat and MODIS imagery 9

4 Frequency of flag classes for 2000–2008 in the MODIS-based fractional cover of Guerschman et al. (2009) indicating algorithm performance 16

5 Existing monitoring sites indicating those that have measured ground cover using the SLATS method 20

6 Agricultural non-woody areas of the rangelands are where the MODIS-based fractional cover of Guerschman et al. (2009) is required first for monitoring 21

1

SummaryGround cover is the vegetation, biological crusts and stones in contact with the soil surface. It has a significant impact on the amount of soil redistributed or lost through wind and water erosion, on the biomass which contributes to soil carbon levels and the ability of vegetation to respond to rain after drought. Ground cover can be monitored directly through remote sensing.

A workshop was held in Canberra (23–24 November 2009) to reach consensus on a nationally coordinated approach to ground cover mapping. Different approaches were presented for estimating ground cover using remote sensing. The issues surrounding validation of such remotely sensed products were also discussed. Discussions were guided by the user specifications for this product.

A first step in this coordinated approach was the establishment of four working groups from the workshop participants. The working groups are:

1 fractional cover mapping operational for the nation—using the Moderate Resolution Imaging Spectroradiometer (MODIS) at 500-metre spatial resolution as a 16-day composite

2 upscaling—from field data to high/medium resolution, small extent imagery to low resolution, large area imagery

3 field data collection—for validation of fractional cover derived from remote sensing

4 monitoring land management practices with remote sensing.

Collectively, the four working groups will advise on the delivery of a validated fractional cover product for Australia. This product will be used:

• to monitor ground cover levels• as a key input to wind and water erosion modelling to predict rates of soil loss• to monitor the impact of different management practices on ground cover levels and soil

erosion risk.

The method of Guerschman et al. (2009) has been selected as the approach to implement nationally, with initial focus on making this MODIS-based fractional cover product operational. Essential to the use of this product is the accuracy of its ground cover estimates. The collection of field measurements and the use of upscaling techniques will underpin the validation and improvement of this fractional cover product.

The working groups documented the workshop discussion and have recommended a way forward. The effort required to implement the suggested tasks has been estimated and will be used to inform the development of a project plan. Progress on the project is available on the Caring for our Country website: http://www.nrm.gov.au/publications/factsheets/ground-cover-factsheet.html.


2

Acronyms and initialisms

ABARES Australian Bureau of Agricultural and Resource Economics and Sciences

ABS Australian Bureau of Statistics

ACLUMP Australian Collaborative Land Use and Management Program

ACRIS Australian Collaborative Rangeland Information System

ALUM Australian Land Use and Management (Classification)

ARM Agricultural Resource Management (Survey)

AVHRR Advanced Very High Resolution Radiometer

BG / BS Bare ground / bare soil

BRDF Bidirectional reflectance distribution function

BRS Bureau of Rural Sciences

CAI Cellulose absorption index

CEMSYS Computational Environmental Management System

CEOS Committee on Earth Observation Satellites

CMIS CSIRO Mathematics, Informatics and Statistics

CLW CSIRO Land and Water

CMA Catchment Management Authority

CSIRO Commonwealth Scientific and Industrial Research Organisation

DAFF Department of Agriculture, Fisheries and Forestry (Australian Government)

DECCW Department of Environment, Climate Change and Water (New South Wales)

DERM Department of Environment and Resource Management (Queensland)

EVI Enhanced vegetation index

GCI Ground cover index

GA Geoscience Australia

GRDC Grains Research and Development Corporation

IDL Interactive Data Language

LAI Leaf area index

LUMIS Land Use and Management Information System

MCAS-S Multi-Criteria Analysis Shell for Spatial decision support

MODIS Moderate Resolution Imaging Spectroradiometer

NBAR Nadir (directly below sensor) BRDF-Adjusted Reflectance

NCLUMI National Committee for Land Use and Management Information

NCST National Committee on Soil and Terrain

NDVI Normalized Difference Vegetation Index

NRM Natural resource management

NLWRA National Land and Water Resources Audit

NPV Non-photosynthetic vegetation

PV Photosynthetic vegetation

RUSLE Revised Universal Soil Loss Equation

SLATS Statewide Landcover And Trees Study (Queensland)

TERN Terrestrial Ecosystem Research Network

USGS United States Geological Survey

1

3

A target of the Australian Government’s Caring for our Country program is to improve the condition of the soil resource through adoption of better land management practices. Ground cover is a good indicator of management and its impact on soil condition (figure 1). Through the Australian Bureau of Statistics’ Agricultural Resource Management (ARM) Survey, data are collected on ground cover management such as tillage and stubble practices, how ground cover levels are monitored and minimum ground cover level targets for grazed paddocks (for example, map 1). These data are used for intermediate outcome reporting for Caring for our Country on sustainable practices. A suitable remotely sensed product of ground cover would enable ground cover levels to be directly monitored consistently for Australia over months, seasons and years. Such a product would give trends in ground cover levels complementing the ARM Survey and as an input to wind and water erosion, soil carbon and acidification models enabling forecasting. Ideally, this remotely sensed product would distinguish between the living and dry/dead vegetation and bare soil—a fractional ground cover product.

A workshop was held in Canberra (23–24 November 2009) to reach consensus on a nationally coordinated approach to ground cover mapping. Appendix A lists the presentations given to inform discussion on different methods and validation of such remotely sensed products. Workshop participants represented potential collaborators and users of a remotely sensed ground cover product (Appendix B). With limited resourcing, a collaborative approach will be important for achieving success. The adopted approach and delivered national product must thus seek to meet partners’ needs—those of the Australian Government, state and territory governments and others.

Specifications of a remotely sensed product for monitoring ground cover nationally were provided to workshop participants to focus discussions. These specifications are given in Appendix C. In summary, a national fractional cover product is required at monthly intervals (at 500 to 1000-metre resolution) supported by a yearly woody layer and a yearly or seasonal medium (30-metre) resolution layer. A high priority to validate the selected remotely sensed product/s is a network of reference sites where the method used to measure fractional cover is sensor-independent. A staged approach is required to deliver a national remotely sensed ground cover product, considering user needs, image acquisition and storage costs and availability of existing suitable or adaptable methodologies.

Introduction


4

The importance of ground cover in minimising (a) water erosion and (b) wind erosion

0

0

20

40

60

80

100

0

20

40

60

80

100

20 40 60 80 100

(a) water erosion (b) wind erosion

ground cover (%)ground cover (%)

aver

age

annu

al s

oil l

oss

(t/h

a)

soil

tran

spor

t rat

e (g

/m/s

)

1

0 20 40 60 80 100

Note: Threshold levels are indicated as green for low risk, orange for medium risk and red for high risk of soil erosion.Source: Barson and Leys, workshop presentation.

Results from the 2007–08 ARM Survey showing graziers with a minimumground cover targetm

ap

1

yes – have a target

graziers with a minimum ground cover target (%)

no – do not have a targetnumber of graziers reported in an NRM region1,123

Note: Sixty-nine per cent of graziers monitored ground cover in 2007–08. Fifty-seven per cent of these graziers had a minimum ground cover target of 40 per cent to more than 80 per cent (ABS 2009).Source: Barson and Leys, workshop presentation.

5


box 1 Cover and remote sensing

Land cover is the physical surface of the earth, including various combinations of vegetation types, soils, exposed rocks and water bodies. Land cover classes may be discriminated by characteristic patterns using remote sensing. Land cover is distinct from land use. Land use is how humans use the land, for example for urban and agricultural land uses.

Fractional cover is the fraction of an area (usually a pixel for the purposes of remote sensing) that is covered by a specific cover type such as green or photosynthetic vegetation, non-photosynthetic vegetation (that is, stubble, senescent herbage and leaf litter) or bare soil/rock. Areas that have been burnt resulting in ash/blackened soil are considered as a bare soil cover type.

Ground cover is the vegetation (living and dead), biological crusts and stone that is in contact with the soil surface. Non-woody ground cover such as crops, grass, forbs and chenopod-type shrubs may change monthly, making this component a good indicator of land management performance (Leys et al. 2009). Ground cover is a sub-component of land cover and, from a remote sensing perspective, is the fractional cover of the non-woody understorey.

6

2Box 1 defines ground cover and fractional cover in terms of remote sensing. A remotely sensed fractional cover product considers all vegetative cover—woody and non-woody at the different strata—ground, mid, upper and emergent (as defined in NCST 2009). To achieve a ground cover product, the fractional cover for non-woody vegetation within the ground strata (usually less than 2 metres tall) is of interest only. This requires a woody cover layer to mask the fractional cover product in those areas under woodlands and forests where it is difficult to extract ground cover. Such a woody layer is produced annually by some states for reporting of tree clearing and nationally for the National Carbon Accounting System by the Department of Climate Change and Energy Efficiency. Queensland Department of Environment and Resource Management (DERM) has recently tried unmixing Landsat pixels to provide cover estimates under canopies; however, this requires validation and further research.

box 2 Characterisation of satellite remote sensing systems

Different remote sensing satellite systems have diverse spatial, temporal and spectral resolutions.

Spatial resolution specifies the pixel size of satellite images covering the earth’s surface.

• High spatial resolution: 0.6–4 metres

• Medium spatial resolution: 4–30 metres (for example, Landsat)

• Low spatial resolution: 30 – >1000 metres (for example, MODIS)

Temporal resolution specifies the revisiting frequency of a satellite sensor for a specific location.

• High temporal resolution: < 24 hours – 3 days (for example, MODIS)

• Medium temporal resolution: 4–16 days (for example, Landsat)

• Low spatial resolution: > 16 days

Spectral resolution specifies the number of spectral bands in which the sensor can collect reflected radiance. Another important aspect of spectral resolution is the position of the bands in the electromagnetic spectrum.

• High spectral resolution: 16–220 bands

• Medium spectral resolution: 3–15 bands (for example, Landsat and MODIS for bands related to land properties)

• Low spectral resolution: < 3 bands

Owing to technical constraints, satellite remote sensing systems can only offer a high spatial resolution with a medium or low spectral resolution, and vice versa. For different applications it is necessary to find compromises between the different resolutions or to use alternative methods of data acquisition.Source: Satellite Imaging Corporation, http://www.satimagingcorp.com/characterization-of-satellite-remote-sensing-systems.html.

Options for monitoring ground cover using remote sensing

7


Photosynthetic and non-photosynthetic vegetation and bare soil have different spectral responses. These spectral responses can be used to select these components using remote sensing. Figure 2 illustrates how the algorithm of Guerschman et al. (2009) can extract the different fractional components. The ratio of MODIS band 7 to band 6 correlates negatively with the cellulose absorption index (CAI) and can be used with the Normalised Difference Vegetation Index (NDVI) to resolve the vegetation fractional cover with MODIS. The University of Adelaide also uses MODIS bands 6 and 7 to predict the cover rating from South Australia’s windscreen surveys. Both methods require independent field-based fractional cover data for calibration and validation of the derived MODIS-based products.

The algorithm of Guerschman et al. (2009) was developed for the northern savannas of Australia and applied nationally. The data are available as a 16-day composite on an 8-day phased production at 500-metre spatial resolution from 2000 onward (http://www- data.wron.csiro.au/remotesensing/MODIS/products/Guerschman_etal_RSE2009/; map 2).

The University of Adelaide’s index has been developed for South Australia’s agricultural lands and is not currently available nationally or publicly. Future developments include the use of fractional cover indices from relative spectral mixture analysis (Clarke et al. 2010).

Geoscience Australia (GA) and ABARES have developed a 250-metre Dynamic Land Cover Map (Lymburner et al. 2010) for Australia, which is due for public release in early 2011. This characterises the enhanced vegetation index time series to obtain statistics, phenology and seasonality factors on photosynthetic vegetation. GA aims to link the fractional vegetation cover product of Guerschman et al. (2009) to the Dynamic Land Cover Map to enable interpretation of trends in vegetative cover (separating photosynthetic and non-photosynthetic vegetation) under different land cover types.

DERM uses Landsat (and MODIS) to provide an annual ground cover index for all of Queensland (map 3). A Landsat-based (30-metre) woody vegetation cover product is also produced annually. DERM is moving toward a Landsat ground cover product which separates

Theoretical basis of the Guerschman et al. (2009)algorithm considers the spectral characteristics ofphotosynthetic vegetation (chlorophyll at 600–800nanometres), non-photosynthetic vegetation(cellulose and lignin at 2000–2200 nanometres)and bare soil quanti�ed by the Normalised Di�erenceVegetation Index (NDVI) and the celluloseabsorption index (CAI)

non-photosynthetic vegetation

photosyntheticvegetation

bare soil

CA

I

NDVI

2

Source: Guerschman, workshop presentation.


8

the three fractional components (photosynthetic vegetation, non-photosynthetic vegetation and bare soil), and is currently considering the accuracy and operational efficiencies of three approaches: (i) a regression-based bare ground prediction, (ii) multiple logistic regression algorithms and (iii) a spectral unmixing approach (Schmidt et al. 2010a). With free access to the United States Geological Survey Landsat image archive back to 2000, Queensland is acquiring imagery to enable time series analysis (for example, Scarth et al. 2010) within years (of up to 20 images per year) as well as their current annual end of dry season time series back to before 2000.

Airborne hyperspectral data are useful for mapping non-photosynthetic vegetation and the types of bare soil/rock, as they are not influenced by variation in soil moisture as multispectral sensors such as Landsat and MODIS are. Airborne hyperspectral data are available now but at a cost of approximately $15 million for a 15-metre baseline map of Australia. From 2013, hyperspectral soil products will become routinely accessible to Australia. CSIRO Exploration & Mining’s 10-year vision is web-accessible hyperspectral mineral maps of the Australian continent for baseline mapping and monitoring.

Mean fractional vegetation cover for 2000–2009

map

2

100%NPV

100%BS

100%PV

Note: PV = photosynthetic vegetation, NPV = non-photosynthetic vegetation, BS = bare soil.Source: Guerschman, workshop presentation.

9


Points to consider for monitoring ground cover using remote sensing are:

• climate is a major cover driver• the spatial arrangement of cover—often linked to management factors— is lost with

increasing pixel size• soil moisture has an effect on remotely sensed ground cover estimates—particularly

multispectral (that is, Landsat and MODIS) estimates of non-photosynthetic vegetation versus bare soil

• atmospheric correction is essential for time series remotely sensed data• remotely sensed estimates need to be validated with accurate on-ground measurement.

Box 2 summarises how different remote sensing systems are characterised by their spatial, temporal and spectral resolutions, and indicates where the current options (Landsat and MODIS) used for large extent monitoring of ground cover align.

Dry season ground cover and foliage projective cover (FPC) for the FoxCreek sub-catchment within the Burdekin catchment, Queensland,in 2002 using Landsat and MODIS imagery

map

3Fox CreekCatchment 2002– Landsat

Fox CreekCatchment 2002– MODIS

property boundarieshigh ground cover %low ground cover %above 20% FPC

Note: At low (< 40 per cent) ground cover levels information is lost at the 500-metre spatial resolution of MODIS.Source: Schmidt, workshop presentation.

10

3A remotely sensed fractional ground cover product requires field data to validate the reliability of the product. Various methods are available to collect fractional cover in the field and the accuracy of these methods can and should be assessed. For example, the methods used for grassland curing assessments (that is, the non-photosynthetic vegetation fraction) by the Bushfire Cooperative Research Centre have an accuracy of +/– 10 per cent for the Levy Rod method and +/– 20 per cent for visual estimates. DERM uses a modified discrete point transect sampling method in their SLATS program (Scarth et al. 2006; Schmidt et al. 2010b) (see figure 4). Automated in situ sensors are also another option for collecting multi-spectral data that can be converted to the three fractional ground cover components.

The Terrestrial Ecosystem Research Network (TERN) is currently establishing a national network of sites for calibration and validation of remotely sensed products supported by its AusCover nodes. Measurements proposed for these sites are:

• canopy multispectral reflectance (nadir or bidirectional)• leaf spectra (reflectance and transmittance)• background nadir spectral reflectance (soil and litter)• fraction of areal vegetation cover• vegetation crown allometry (height, width, gap)• phenology (green-up, mature, and senescent stages)• vegetation composition (either by species or structural type)• moisture status• fraction of non-photosynthesising vegetation• meteorological data.

Collaboration with the TERN initiative will provide access to high spectral reflectance measurements at known sites and, potentially, the required measurements for validating the selected remotely sensed ground cover product. Agreement on standards for measurements at sites will be important for effective linkage between the TERN and this national ground cover project.

TERN will draw on lessons learnt from international validation programs and encourage the use of:

• representative sites• homogenous or constantly mixed and large (preferably at least 5 kilometres by 5 kilometres)

sites

Methods to calibrate and validate remotely sensed ground cover

11


Methods to calibrate and validate remotely sensed ground cover • a hierarchy from highly instrumented supersites to networks of spatially extensive higher

order plots • high spatial resolution images to bridge to coarser resolution images• autonomous highly instrumented supersites for calibration and validation.

Box 4 provides internationally recognised definitions for calibration and validation.

Some initial validation of the Guerschman et al. (2010) MODIS-based fractional cover product have been undertaken by Schmidt et al. (2010b) and G Bastin ([CSIRO] 2009, pers. comm., September). Analysis of the algorithm’s performance using the flag classes (see map 4) also indicates where the product is currently working and where improvement is required. These types of analyses can assist with establishing sites for calibrating, validating and ultimately improving the selected remotely sensed product to provide ground cover estimates of known reliability.

12

4To coordinate national delivery of ground cover mapping, four working groups were established from the workshop participants. The working groups are:

1 fractional cover mapping operational for the nation—initially using MODIS

2 upscaling—from field data to high/medium resolution, small extent imagery to low resolution, large area imagery

3 field data collection—for validation of fractional cover derived from remote sensing

4 monitoring land management practices with remote sensing.

Collectively, the four working groups will advise on the delivery of a validated fractional cover product for Australia. Figure 3 illustrates the interconnections between the four working groups. Appendix B shows the membership to each of the working groups by workshop participants.

The four working groups and their relationship to each other for delivery of nationally coordinatedground cover mapping

1

2

3 4

MODIS-based fractionalcover mapping

operational nationally

upscaling

field data collection monitoring landmanagement practices

3

A coordinated approach for ground cover mapping

13

5The overarching objectives of the working groups are:

1 validation of an operational fractional cover product

2 ongoing improvement of this product

3 analysis of ground cover estimates from this product.

Each working group’s strategy will have as its priority an operational product available in 2010, with ongoing validation and improvement of that product over the term of Caring for our Country (2010–2013). The data collected for the current product validation ideally should be non-sensor specific to allow for production of a finer resolution product when feasible. This should also apply to the algorithms used in upscaling and production of the fractional cover.

The working groups will seek to be actively involved in the progress of related activities under the Terrestrial Ecosystem Research Network (TERN) AusCover (that is, related to calibration and validation of their land products) and identify potential opportunities for co-investment.

Working group 1: Fractional cover mapping operational for the nationThe purpose of working group 1 is to make remotely sensed fractional cover products for Australia operational. The initial focus will be the MODIS-based fractional cover product of Guerschman et al. (2009). This working group will seek to improve the existing product by:

• establishing where the existing algorithm does and does not work to inform stratification for locating sites to obtain field data

• using field data to further train the algorithm and validate the product• revising the Interactive Data Language (IDL) algorithm.

Working group 2: UpscalingThe purpose of working group 2 is to develop a strategy for upscaling (box 3) from field data to remotely sensed imagery to produce a MODIS-based fractional cover product.

This involves:

• determining the technical details of the optimal upscaling algorithm• advising on the sampling framework for location of field sites• advising on the timing of field data collection to be sufficiently co-incident with sensor/s

overpass.

Working group objectives


14

box 3 Upscaling

Upscaling is a way of taking information based in observations that have a small spatial (or temporal) domain and extending that domain to a larger geographic area (or time period). Crossing a resolution (or scale) boundary is a key feature of upscaling and is what distinguishes upscaling from aggregation (which is the process of combining data from smaller sub-regions). The dates of field work and image acquisition cannot be too far apart or else the characteristic being measured and estimated remotely may have changed (T McVicar [CSIRO] 2009, pers. comm., December).

Examples of upscaling techniques of vegetation characteristics relevant to this project include McVicar et al. (1996a,b), Milne et al. (2007) and Guerschman et al. (2009). These techniques are outlined in Appendix D.

Working group 3: Field data collectionThe purpose of working group 3 is to validate remotely sensed fractional cover products for the purpose of monitoring ground cover levels.

This involves:

• agreeing on field methods for measuring ground cover (national standard)• prioritising where to locate sites to measure ground cover• agreeing on the timing and frequency of measurement at a site• coordinating parties collecting field data, managing data and providing feedback.

Working group 4: Monitoring land management practices with remote sensingThe purpose of working group 4 is to develop a monitoring framework for ground cover management practices using remotely sensed fractional cover. In the mixed farming or cropping zone, this includes:

• advising on the sampling framework for location of field sites so key management practices (and land uses) are considered

• advising on the frequency and timing of field data collection of ground cover levels to capture the sequence of management practices for a site

• developing the metrics to interpret fractional cover time series for selected management practices (for example, timing and length of cultivated fallow, removal or retention of crop residues, timing and number of tillage passes, grazing pressure and practice, perennial pasture content, and conversion of continuously low productivity areas to perennial vegetation).

This working group will build on and seek to implement the recommendations of Stewart and Rickards (2010), focusing on selected sites in the cropping/mixed farming zone to complement the data collected through the Australian Bureau of Statistics’ Agricultural Resource Management Survey.

15

6Workshop recommendations

The key recommendations from the workshop are listed below, under the four working groups.

Working group 1: Fractional cover mapping operational for the nation• Use the method of Guerschman et al. (2009) to produce a fractional cover product

for Australia as 500-metre 16-day composites of bare soil, photosynthetic and non-photosynthetic vegetative cover derived from MODIS imagery.

• Automate the delivery of this fractional cover product to allow updating every four days.• Provide caveats on the use and limitations of the dataset.• Seek to improve the dataset with further validation using:

– flag analysis to identify areas where validation data are required (for example, map 4)– existing suitable site data available from state agencies (in particular the Queensland

Department of Environment and Resource Management (DERM) and NSW Department of Environment, Climate Change and Water (DECCW))

– new site data collected by state agencies, Geoscience Australia (GA), CSIRO and others (for example, the Grains Research and Development Corporation (GRDC)).

• Upscale from site data to Landsat and then to MODIS imagery. This is a more cost-effective and feasible way to collect field data and allows development of two ground cover products in tandem.

• Determine product accuracy (aiming for ±10–20 per cent error). Need to consider:

– improving discrimination of low ground cover levels—those less than 50 per cent– estimating area at risk—whether overestimation or underestimation is better when

targeting regions.

Working group 2: UpscalingA successful upscaling strategy should be ensured through implementing the following protocols.

• Ensure field sites are accurately located using differential GPS technology.• Ensure field data are collected using Queensland’s Statewide Landcover And Trees Study

(SLATS) modified discrete point sampling method along 100-metre transects:

– for pastoral environments using the star-shaped transect approach of Scarth et al. (2006)– for intensive agricultural environments with linearly sowed crops using the cross-transect

method of Schmidt et al. (2010b) (figure 4).• Ensure field crews are trained by DERM in the use of the SLATS modified discrete point

sampling method—this is the agreed minimum national specifications for field data.


16

• Ensure field data are stored in a spatially enabled relational database that is compatible with the current Queensland PostGIS database (http://postgis.refractions.net/) to quickly generate a national database.

• Locate field sites in both homogenous herbaceous land covers and in more heterogeneous tree-shrub-grass mixed systems. Within a field site the land cover needs to be fairly homogenous and representative of wider conditions.

• Design the field work strategy to cover a wide range of soil colours (for example, white kaolinite clays, black cracking clays, red desert sands and yellow earths). Focus on collecting data for the various natural and agricultural systems in jurisdictions other than Queensland (because of their extensive SLATS field sites), with the exception of the black clays of central-western Queensland.

• Assess the impact of soil colour to improve the MODIS fractional cover product based on Guerschman et al. (2009) by having coincident at selected sites:

– field collection using the SLATS modified discrete point sampling method– field level spectral radiometric observations—using a portable spectroradiometer

(available from Analytical Spectral Devices Inc.)– Hyperion imagery—with the exact location specified before tasking this pointable

hyperspectral sensor.

Frequency of �ag classes for 2000–2008 in the MODIS-based fractional cover of Guerschman et al. (2009) indicating algorithm performance

map

4

class 0 class 1

class 2 class 255

less than 20%20% – 40%40% – 60%60% – 80%more than 80%

Note: Class 0: pixel processed normally, unmixing unconstrained; Class 1: unmixing partially constrained, fractional coverestimated; Class 2: unmixing constrained and fractional cover not estimated; Class 255: MODIS NBAR data not available.Source: Randall, workshop presentation.

17


Working group 3: Field data collectionIn addition to the recommendations of the upscaling working group:• Adopt the definitions of the Committee on Earth Observation Satellites (CEOS) Working

Group on Calibration and Validation for validation and their validation hierarchy (box 4).• Establish the relationship between monitoring ground cover obliquely and from overhead.

This may allow use of existing obliquely viewed field measurements of ground cover (that is, windscreen surveys). Forward (2009) indicates there are more than 8130 sites from windscreen surveys in cropping and rangelands environments representing most jurisdictions that potentially could be used, also providing historical data (for example, South Australian data back to 1999–2000, and Victorian Mallee and Western Australian data back to 2007).

• Undertake a review of ‘human variability’ and its effect on the measurement of ground cover in the field for both rangelands and cropping. There is significant literature addressing this issue, particularly for the rangelands. This should underpin the adoption of the selected field method/s and its accuracy.

• Make minor modification to the SLATS generic site data form (Appendix E) to align with national standards for soil and land surveys (NCST 2009).

• Develop a statistically robust sampling framework that is representative of land uses and soil colours/types. The MCAS-S software tool (BRS 2009) could be used to prioritise where to collect ground cover measurements within this sampling framework.

Layout of transects used in Queensland’s Statewide Landcover and Trees Study (SLATS) modi�ed discrete point sampling method for measuring fractional coverin (a) pastoral environments and (b) agricultural crops sown in lines

60°

100m 100m

N

4(a) star-shaped transect (b) cross-shaped transect

45°


18

• Adopt a stratified approach for the sampling framework. Determine whether one approach is suitable nationally or different approaches can be accommodated considering existing approaches. Examples include:

– soil condition monitoring that is based on monitoring regions (broad landscape divisions based on physiographic regions), monitoring units (a combination of major soil types and land uses) and monitoring sites (locations within units where repeat sampling occurs using a 25-metre grid approach)

– the Victorian primary production landscapes– stratification for the rangelands (under TERN).

• Establish a central registry of monitoring sites from all disciplines. This would require a champion organisation to bring together the disciplines and activities. The initial focus of this registry would be the sites identified in existing state monitoring programs that are relevant to ground cover and in particular aligned with the SLATS methodology (such an assessment has been undertaken by NCLUMI through the Australian Collaborative Land Use and Management Program (ACLUMP)—map 5).

box 4 Calibration and validation of remotely sensed data

Calibration is the process of quantitatively defining the system responses to known, controlled signal outputs (that is, sensor radiometric and geometric performance, and stability).

Vicarious calibration refers to techniques that make use of natural or artificial sites on the earth’s surface for the post-launch calibration of sensors.

Training refers to the process of developing and refining an algorithm designed to produce a specific data product from the system outputs.

Validation is the process of assessing, by independent means, the quality of the data products derived from the system outputs (that is, comparison and scaling of remotely sensed derived biophysical products versus in situ measurements). Validation is undertaken in four stages:

• Stage 1 validation: Product accuracy is assessed from a small (typically less than 30) set of locations and time periods by comparison with in situ or other suitable reference data.

• Stage 2 validation: Product accuracy is estimated over a significant set of locations and time periods by comparison with reference in situ or other suitable reference data. Spatial and temporal consistency of the product and consistency with similar products has been evaluated over globally representative locations and time periods. Results are published in the peer-reviewed literature.

• Stage 3 validation: Uncertainties in the product and its associated structure are well quantified from comparison with reference in situ or other suitable reference data. Uncertainties are characterised in a statistically robust way over multiple locations and time periods representing global conditions. Spatial and temporal consistency of the product and consistency with similar products has been evaluated over globally representative locations and periods. Results are published in the peer-reviewed literature.

• Stage 4 validation: Validation results for stage 3 are systematically updated when new product versions are released and as the time series expands.

The definitions for calibration, vicarious calibration and validation are adopted from the Committee on Earth Observation Satellites (CEOS) Working Group on Calibration and Validation (www.ceos.org/wgcv). The validation stages are from the guidelines for the CEOS Hierarchy of Validation (updated October 2009, available at http://lpvs.gsfc.nasa.gov/).

19


• Ensure that the field campaign implemented considers the seasons and overpasses of the satellite. Field measurements should at a minimum be twice per year, at the end of the wet and dry seasons, to correspond with maximum and minimum levels of cover.

Working group 4: Monitoring land management practices with remote sensing• Modify the Queensland SLATS generic site data form (Appendix E) to include land

management practices. • Seek to standardise practice terminology for consistency with other monitoring programs

(for example, windscreen/roadside surveys for erosion risk). Ideally this should align with the activities under the National Committee for Land Use and Management Information (NCLUMI)—namely the Australian Land Use and Management (ALUM) Classification and categories of the Land Use and Management Information System (LUMIS).

• Gather information required on the timing of land management practices throughout the year to assist in the interpretation of remotely sensed fractional cover time series. This could be achieved by:

– providing a confidence estimate for the area represented by generalised ‘calendars of operations’ collected for natural resource management regions by the NCLUMI through ACLUMP

– coordinating with other monitoring programs which collect information on land management practices multiple times per year (for example, windscreen surveys (SA, NSW, WA), Victorian Mallee survey) at known locations to be coincident with sensor overpass

– compiling ‘paddock diaries’ of field operations (including stocking rates and weather events) at sites with the assistance of land managers using standardised terminology

– visiting field sites at key changes in ground cover levels because of management practices to measure ground cover.

• Align management practices collected with those of the ABS Agricultural Resource Management Survey, with a focus on ground cover related practices.


20

Existing monitoring sites indicating those that have measured groundcover using the SLATS methodm

ap

5

ground cover measured using SLATS methodsite suitability characteristics

ground cover measured using other quantitative methodground cover not measured but possible

Source: Australian Collaborative Land Use and Management Program from data supplied by state and territory agencies.

21

7 Implementing the workshop recommendations

Tables 1 and 2 outline the steps and estimates of the effort required to implement the recom-mendations of the working groups for a nationally coordinated approach to ground cover mapping using remote sensing. With a limited budget, the Department of Agriculture, Fisher-ies and Forestry (DAFF) has identified the extensive grazing systems of the rangelands (Bastin et al. 2008) as the focus for initial monitoring (map 6). The rangelands contain most of the areas where wind erosion is widespread (see http://www.nrm.gov.au/business-plan/10-11/priorities/sustainable/pubs/map-2-4-2.pdf) and where data is required to improve estimates of hillslope erosion by water (see http://www.nrm.gov.au/business-plan/10-11/priorities/sustainable/pubs/map-2-4-4.pdf). Priority will be given to those tasks that will most improve the MODIS-based fractional cover product for monitoring within the rangelands. The mixed farming or intensive land use zone will be considered for limited sites only to assist in national validation of the frac-tional cover product. The Caring for our Country website (http://www.nrm.gov.au) will provide progress on the implementation of these recommendations, within the available resources.

Agricultural non-woody areas of the rangelands are where theMODIS-based fractional cover of Guerschman et al. (2009) is required �rst for monitoring

map

6

rangeland boundary

NRM regions

forests, woodlands and plantations

agricultural land uses

non-agricultural land uses


22

1 Ke

y ta

sks

to d

eliv

er a

n op

erat

iona

l rem

otel

y se

nsed

frac

tiona

l cov

er n

atio

nally

(201

0–20

13)

le

ad o

rgan

isat

ion

estim

ate

ofta

sk

and

part

ners

tim

efra

me

effor

t req

uire

d co

mm

ents

1. A

utom

ate

del

iver

y of

CSI

RO’s

MO

DIS

-bas

ed

GA

, CSI

RO (C

LW)

4–5

mon

ths

afte

r G

A 0

.5 F

TE

Exp

erim

enta

l pro

duc

t

frac

tion

al c

over

pro

duc

t (v

ersi

on 0

) a

p

roje

ct in

itia

tion

a

CSI

RO 0

.5 F

TE

Lim

ited

val

idat

ion

• usi

ng G

A’s

NBA

R d

ata

and

the

alg

orit

hm o

f

Del

iver

ed t

hrou

gh

GA

’s w

ebsi

te a

nd/

G

uers

chm

an e

t al

. (20

09)

or

TER

N/A

usC

over

• p

rod

uct

is u

pd

ated

eve

ry fo

ur d

ays

on 1

6-d

ay

U

sers

exp

erie

nces

and

cav

eats

to

be

500

-met

re c

omp

osit

es

p

rovi

ded

• pro

duc

t d

isti

ngui

shes

bar

e so

il, n

on-p

hoto

synt

heti

c

A

sses

smen

t of

GA

’s an

d C

SIRO

’s a

nd p

hoto

synt

heti

c ve

get

atio

n

q

ualit

y co

ntro

l of N

BAR

pro

duc

t an

d• i

nclu

des

flag

dat

a (in

dic

atio

n of

relia

bili

ty o

f cov

er

p

roce

ssin

g e

stim

ate)

• inc

lud

es m

etad

ata

and

doc

umen

tati

on

2. U

nder

take

ass

essm

ent

of C

SIRO

’s M

OD

IS-b

ased

G

A, C

SIRO

(CLW

) O

ngoi

ng

GA

0.2

5 FT

E C

onti

nuou

s im

pro

vem

ent

of M

OD

IS

fr

acti

onal

cov

er p

rod

uct

cons

ider

ing

: D

ERM

and

oth

er s

tate

(TER

N fu

nded

) fr

acti

onal

cov

er p

rod

uct

• rob

ustn

ess

of t

he a

lgor

ithm

– is

it a

pp

rop

riat

e to

ag

enci

es t

o p

rovi

de

C

SIRO

?

Stat

e ag

enci

es t

o p

rovi

de

exis

ting

ap

ply

nat

iona

lly o

r ar

e m

odifi

cati

ons

req

uire

d fo

r

field

dat

a

fie

ld d

ata

diff

eren

t la

ndsc

apes

?

DER

M t

o p

rovi

de

thei

r G

CI a

lgor

ithm

• sp

ectr

al c

hara

cter

isti

cs –

sou

ndne

ss o

f con

tras

t of

for

MO

DIS

imag

ery

ban

ds

6 an

d 7

(aff

ecte

d b

y so

il m

oist

ure

and

soi

l

Coo

rdin

atio

n ro

le fo

r A

CRI

S w

ith

col

our)

rang

elan

d ju

risd

icti

ons

• oth

er u

nmix

ing

mod

els

e.g.

DER

M’s

grou

nd c

over

ind

ex (G

CI)

• exi

stin

g d

atas

ets

avai

lab

le t

o te

st a

lgor

ithm

e.g.

GC

I and

SLA

TS d

ata

3. R

elea

se a

n im

pro

ved

ver

sion

of C

SIRO

’s

CSI

RO (C

LW),

GA

To

be

advi

sed

G

A 0

.25

FTE

M

inim

ise

unce

rtai

nty

in t

he 4

0–60

MO

DIS

-bas

ed fr

acti

onal

cov

er p

rod

uct

(ver

sion

1)

(d

epen

ds

on fi

eld

C

SIRO

?

per

cen

t gr

ound

cov

er le

vels

• usi

ng a

revi

sed

IDL

alg

orit

hm b

ased

on

the

resu

lts

dat

a co

llect

ion

U

se e

xist

ing

and

new

fiel

d d

ata

fr

om: –

cal

ibra

tion

sit

es

ca

mp

aign

in t

able

2)

co

llect

ed b

y th

e st

ate

agen

cies

,

– hi

gh/

med

ium

-res

olut

ion

dat

a w

ith

upsc

alin

g

C

SIRO

and

GA

–

valid

atio

n si

tes

• w

ith

doc

umen

tati

on t

o in

dic

ate

diff

eren

ces

bet

wee

n ve

rsio

ns

cont

inue

d....

23


1 Ke

y ta

sks

to d

eliv

er a

n op

erat

iona

l rem

otel

y se

nsed

frac

tiona

l cov

er n

atio

nally

(201

0–20

13)

cont

inue

d

le

ad o

rgan

isat

ion

estim

ate

ofta

sk

and

part

ners

tim

efra

me

effor

t req

uire

d co

mm

ents

4. P

rod

uce

a La

ndsa

t-b

ased

(30

m) f

ract

iona

l cov

er

DER

M, D

ECC

W, G

A,

To b

e ad

vise

d

To b

e ad

vise

d

TERN

look

ing

to

stan

dar

dis

e

pro

duc

t

CSI

RO (C

LW)

pro

cess

ing

of L

and

sat

imag

ery

to• u

nder

take

an

asse

ssm

ent

of a

pp

roac

hes,

incl

udin

g:

av

oid

dup

licat

ion

of e

ffor

t (F

ebru

ary

–

linea

r un

mix

ing

(DER

M/D

ECC

W)

20

10 w

orks

hop

)

– m

ulti

ple

line

ar re

gres

sion

(DER

M/D

ECC

W)

Ex

tens

ive

field

dat

a av

aila

ble

for

–

imag

e cl

assi

fied

unm

ixin

g (G

A)

Q

ueen

slan

d u

nder

pas

tora

l lan

d u

ses

• for

sel

ecte

d s

ites

Req

uire

s lo

ts o

f val

idat

ion

dat

a• f

or U

psc

alin

g t

o M

OD

IS-b

ased

frac

tion

al c

over

(6

00 s

ites

ava

ilab

le fo

r Q

ueen

slan

d

wit

h p

lans

for

200–

300

site

s fo

r N

SW)

Use

ful f

or t

arg

eted

regi

ons

rath

er

than

wal

l-to

-wal

l map

pin

g

a Ta

sk 1

cou

ld b

e d

efer

red

to la

ter i

n th

e p

roje

ct a

fter

dev

elop

men

t of v

ersi

on 1

of t

he M

OD

IS-b

ased

frac

tiona

l cov

er.


24

2 Ke

y ta

sks

in th

e va

lidat

ion

of re

mot

ely

sens

ed fr

actio

nal c

over

(201

0–20

13)

le

ad o

rgan

isat

ion

estim

ate

ofta

sk

and

part

ners

tim

efra

me

effor

t req

uire

d co

mm

ents

1. P

erso

nnel

tra

inin

g in

fiel

d d

ata

colle

ctio

n

DER

M/D

ECC

W

Man

ual –

dra

ft fo

r

To b

e ad

vise

d

Cur

rent

SLA

TS g

ener

ic s

ite

dat

a fo

rm• u

nder

pin

ned

by

a fie

ld m

anua

l whi

ch

and

all

stat

e ag

ency

co

mm

ence

men

t of

to b

e im

pro

ved

/mod

ified

(e.g

. to

inc

lud

es a

sit

e fo

rm

par

tner

s tr

aini

ng; f

inal

dra

ft

in

clud

e ex

isti

ng n

atio

nal s

tand

ard

s• m

easu

rem

ent

of v

eget

atio

n fr

acti

ons

whi

ch u

ses

the:

at c

omp

leti

on o

f

for

soil

and

land

sur

veys

(NC

ST 2

009)

– s

tar-

tran

sect

met

hod

for

pas

tora

l env

ironm

ents

trai

ning

for

surf

ace

cond

itio

n an

d la

nd –

cro

ss-t

rans

ect

met

hod

for

linea

rly

sow

n cr

ops

Tr

aini

ng t

o oc

cur

m

anag

emen

t p

ract

ices

(nee

d fo

r• m

easu

rem

ent

of s

oil s

pec

tral

refle

ctan

ce

ov

er 6

–12

mon

ths

st

and

ard

ter

min

olog

y))

afte

r p

roje

ct in

itia

tion

Pers

onne

l tra

inin

g t

o oc

cur

in

diff

eren

t en

viro

nmen

ts t

hrou

gho

ut

Aus

tral

ia (a

t le

ast

one

rang

elan

d s

ite

an

d o

ne c

rop

pin

g s

ite

per

sta

te).

This

tr

aini

ng e

xerc

ise

can

be

used

to

co

mp

are

the

SLAT

S m

etho

dol

ogy

w

ith

exis

ting

sta

te-b

ased

pro

gram

s

Fi

eld

dat

a to

be

colle

cted

for

a

100

met

re x

100

met

re s

ite

(sui

tab

le

for

Land

sat

and

up

scal

ing

to

MO

DIS

)

2. S

amp

ling

fram

ewor

k A

BARE

S an

d C

SIRO

2–

3 m

onth

s fo

llow

ing

To

be

advi

sed

C

onsi

der

leve

l of a

ccur

acy

req

uire

d• s

tati

stic

ally

rob

ust

(C

MIS

) an

exp

erts

wor

ksho

p

an

d t

he im

age

sens

or p

rod

uct

bei

ng• c

onsi

der

s se

ason

alit

y, la

nd u

ses,

soi

l col

our

afte

r p

roje

ct in

itia

tion

valid

ated

(e.g

. MO

DIS

ver

sus

Land

sat)

• aim

s to

tim

e sa

mp

ling

to

coin

cid

e w

ith

Con

sid

er e

xist

ing

str

atifi

cati

ons

used

s

atel

lite

over

pas

s

for

mon

itor

ing

nat

iona

lly a

nd w

ithi

n• s

eeks

a s

tag

ed a

pp

roac

h to

val

idat

ion

of t

he

st

ates

(e.g

. soi

l con

dit

ion

mon

itor

ing,

f

ract

iona

l cov

er p

rod

uct

Vi

ctor

ia’s

pri

mar

y p

rod

ucti

on• u

ses

MC

AS-

S to

pri

orit

ise

site

loca

tion

s b

ased

land

scap

es)

on

fact

ors

such

as

eros

ion

risk

, fla

g a

naly

sis

etc

cont

inue

d....

25


2 Ke

y ta

sks

in th

e va

lidat

ion

of re

mot

ely

sens

ed fr

actio

nal c

over

(201

0–20

13)

cont

inue

d

le

ad o

rgan

isat

ion

estim

ate

ofta

sk

and

part

ners

tim

efra

me

effor

t req

uire

d co

mm

ents

3. S

pat

ially

ena

ble

d re

lati

onal

dat

abas

e fo

r fie

ld d

ata

ABA

RES

Ong

oing

com

men

cing

To

be

advi

sed

Ex

plo

re o

pti

ons

for

elec

tron

ic e

ntry

• com

pat

ible

wit

h cu

rren

t d

atab

ase

whi

ch u

ses

once

dat

a is

rece

ived

of d

ata

wit

hin

the

field

POST

GIS

from

tra

inin

g (

Task

1)

4. F

ield

dat

a co

llect

ion

cam

pai

gn

All

stat

e ag

ency

To

com

men

ce

0.01

FTE

per

day

2–

3 p

eop

le re

qui

red

for

each

sit

e vi

sit

• und

erp

inne

d b

y sa

mp

ling

fram

ewor

k p

artn

ers

fo

llow

ing

tra

inin

g

in t

he fi

eld

Ti

me

at e

ach

site

usi

ng D

ERM

’s SL

ATS

• sta

ged

ap

pro

ach

an

d t

o co

ntin

ue

(thi

s eq

uate

s to

m

etho

d:

• dat

a co

llect

ed t

wic

e ye

arly

as

a m

inim

um

to

May

201

3 ap

pro

xim

atel

y • ~

90 m

inut

es fo

r w

ood

land

s (

wet

and

dry

sea

son)

$1

000

per

sit

e

• ~40

min

utes

for

rang

elan

ds

(no

p

er v

isit

incl

udin

g

tr

ees)

tr

avel

exp

ense

s)

• ~30

min

utes

for

crop

pin

g

Alig

n ca

mp

aign

wit

h in

tere

sted

ot

hers

(e.g

. GA

, CSI

RO, G

RDC

, CM

As)

.

5. E

stab

lishm

ent

of s

uper

site

s

ABA

RES/

CSI

RO

Sele

ct s

ites

to

cove

r ra

nge

of• s

elec

ted

tar

get

ed s

ites

whe

re d

etai

led

fiel

d

(CLW

) wit

h st

ate

en

viro

nmen

ts, p

arti

cula

rly

thos

e m

easu

rem

ents

are

mad

e of

gro

und

cov

er,

agen

cy p

artn

ers

w

here

the

frac

tion

al c

over

pro

duc

t s

pec

tral

refle

ctan

ce, l

and

man

agem

ent

pra

ctic

es

ne

eds

mos

t im

pro

vem

ent

• tri

al U

psc

alin

g t

echn

ique

s• t

rial

met

rics

for

det

ecti

on o

f sel

ecte

d la

nd

man

agem

ent

pra

ctic

es in

mix

ed fa

rmin

g/c

rop

pin

g

26

Aappendix

26

presentation title presenter

Ground cover monitoring for soil condition reporting

Mr Ian ThompsonExecutive Manager, Sustainable Resources Management, Department of Agriculture, Fisheries and Forestry

Ground cover inputs for wind and water monitoring and modelling

Dr Michele Barson1 and Dr John Leys2

1Science Advisor, Australian Government Land & Coasts, Department of Agriculture, Fisheries and Forestry2Principal Research Scientist, Scientific Services Division, NSW Department of Environment, Climate Change and Water

Fractional cover for Australia: scaling from the field to MODIS

Dr Juan GuerschmanResearch Scientist, Environmental Earth Observation Research Group, CSIRO Land and Water

Towards image-based monitoring of soil erosion risk in South Australian agricultural land

Dr Ken ClarkeSchool of Earth and Environmental Sciences, University of Adelaide

A national dynamic land cover capability Dr Leo LymburnerRemote Sensing Applications Specialist, National Earth Observation Group, Geospatial and Earth Monitoring Division, Geoscience Australia

Ground cover in cropping environments using Landsat and MODIS

Dr Michael SchmidtScientist, Remote Sensing Centre, Queensland Department of Environment and Resource Management

Mapping and monitoring of the soil surface using remote hyperspectral imaging

Dr Thomas CudahyGroup Leader, Minerals and Environmental Sensing, CSIRO Exploration & Mining

Bushfire CRC project: improved methods for assessment of grassland curing

Dr Ian GrantSpace Based Observation Section, Bureau of Meteorology

Calibration and validation – analysis of CSIRO fractional cover product

Dr Lucy RandallSenior Scientist, Land and Forest Sciences, Australian Bureau of Agricultural and Resource Economics and Sciences

A very brief update on ground cover mapping in NSW

Mr Tim DanaherRemote Sensing Section, Department of Environment, Climate Change and Water

Wireless sensor networks for monitoring ground cover change – a case study

Dr Andre ZergerCSIRO Sustainable Ecosystems

Wrap-up and options Dr John LeysPrincipal Research Scientist, Scientific Services Division, NSW Department of Environment, Climate Change and Water

Calibration and validation for remotely sensed ground cover: what is required?

Dr Tim MalthusResearch Group Leader, Environmental Earth Observation, CSIRO Land and Water

Note: These presentations are available to workshop participants and interested others via the ground cover monitoring GovDex site. Alternatively, contact the authors for access to the above presentations.

Workshop presentations

27

appe

ndix

27

Bname state organisation

Jason Barnetson e4 NTDepartment of Natural Resources, Environment, The Arts and Sports

Simon Barry Aus CSIRO Mathematical and Information Sciences

Michele Barson a Aus Department of Agriculture, Fisheries and Forestry

Gary Bastin e1 Aus CSIRO Sustainable Ecosystems

Mark Berman Aus CSIRO Mathematical and Information Sciences

Vivienne Bordas AusAustralian Bureau of Agricultural and Resource Economics and Sciences

Dan Carter e2 WA Department of Agriculture and Food

Greg Chapman e2 NSW Department of Environment, Climate Change and Water

Rob Clark e2,4 Vic Department of Primary Industries

Ken Clarke ae1 SA University of Adelaide

Lex Cogle c Aus Murray–Darling Basin Authority

Tom Cudahy a Aus CSIRO Exploration & Mining

Tim Danaher ae2,3 NSW Department of Environment, Climate Change and Water

Anna Dutkiewicz e2 SA Department of Environment and Natural Resources

Alan Forghani d Aus Murray–Darling Basin Authority

Giles Forward e4 SA Department of Environment and Natural Resources

Ian Grant a Aus Bureau of Meteorology

Juan Guerschman ae1,3 Aus CSIRO Land and Water

Peter Hairsine bc Aus CSIRO Land and Water

Brian Harrild d Aus Australian Bureau of Statistics

Alex Held b Aus CSIRO Office of Space Science & Applications

Bob Karfs e2 QldDepartment of Employment, Economic Development and Innovation

Stuart Kearns e4 RDC Grains Research and Development Corporation

Darren Kidd e2,4 Tas Department of Primary Industries, Parks, Water and Environment

Paul Lawrence bc Qld Department of Environment and Resource Management

Adam Lewis Aus Geoscience Australia

John Leys ae1,4 NSW Department of Environment, Climate Change and Water

Leo Lymburner ae1,3 Aus Geoscience Australia

Tim Malthus ae2 Aus CSIRO Land and Water

Elizabeth McDonald Aus Bureau of Meteorology

Tim McVicar be3,4 Aus CSIRO Land and Water

Simon Oliver AusDepartment of Sustainability, Environment, Water, Population and Communities

Annette Philippa e2 AusDepartment of Sustainability, Environment, Water, Population and Communities

Lucy Randall ae1,3 AusAustralian Bureau of Agricultural and Resource Economics and Sciences

Workshop participants


28

name state organisation

Jasmine Rickards e2,4 AusAustralian Bureau of Agricultural and Resource Economics and Sciences

Michael Schmidt ae3 Qld Department of Environment and Resource Management

Jane Stewart e2,4 AusAustralian Bureau of Agricultural and Resource Economics and Sciences

Glenn Stone Aus CSIRO Mathematical and Information Sciences

Richard Thackway e2 AusAustralian Bureau of Agricultural and Resource Economics and Sciences

Medhavy Thankappan Aus Geoscience Australia

Stephen Thomas RDC Grains Research and Development Corporation

Ian Thompson ac Aus Department of Agriculture, Fisheries and Forestry

Alys Wall Aus Murray–Darling Basin Authority

Peter Wilson e2,4 Aus CSIRO Land and Water

Christian Witte e4 Qld Department of Environment and Resource Management

Andre Zerger a Aus CSIRO Sustainable Ecosystems

a Workshop presenter. b Workshop chair. c Attended first day of workshop only. d Attended second day of workshop onlye Working group member. 1 Implementing MODIS-based fractional cover mapping operationally for the nation. 2 Field data collection. 3 Upscaling. 4 Monitoring land management practices with remote sensing.

29

appe

ndix

29

CIntroduction The sustainable practices target for Caring for our Country focuses on improving the condition of the soil resource through the adoption of better land management practices. Four indicators of resource condition, identified through the work of the National Land and Water Resources Audit (and endorsed by the Audit Advisory Council)—wind erosion, water erosion, soil acidification and soil carbon—will be used to help report on the condition of Australia’s soil resources and the longer term effect of land management practices on condition.

Caring for our Country, working in collaboration with state agencies, CSIRO and other research organisations, has invested in a series of projects aimed at improving Australia’s capacity to monitor wind erosion (Forward 2009; Leys et al. 2009, 2010; McTainsh et al. 2010; Mitchell et al. 2010), water erosion (Leys et al. 2009; Bui et al. 2010), soil acidification and soil carbon (Baldock et al. 2009; Ringrose-Voase in preparation). These projects have recommended bringing together data from a number of sources to monitor soil condition indicators by combining ground-based sampling, remote sensing and spatially explicit modelling. A number of the data inputs required for this monitoring will need to be developed or improved—proposals for this work are currently being considered.

High frequency (monthly) spatially explicit ground cover (vegetation, biological crusts and stone that is in contact with the soil surface) data are needed to improve wind and water erosion modelling and monitoring. Most methods used to estimate cover values have not been able to account for dead and dry vegetation, which has frequently been classified as bare ground, leading to overestimates of the rates of erosion (Hairsine et al. 2009).

Spatially explicit ground cover data has also been identified as an input needed to improve soil carbon and pH modelling proposed under the national soil condition modelling project. Additionally, monthly ground cover data will contribute to understanding the impact of climate variability on pasture condition, particularly in the rangelands.

Wind erosion modelling for resource condition reportingThe CEMSYS (Computational Environmental Management System; Shao et al. 2007) model is recommended (Leys et al. 2009) for evaluating the frequency and intensity of wind erosion across Australia, and has been used by Caring for our Country to identify where improving soil and land management practices to reduce soil loss from wind erosion will provide the biggest benefits (Smith and Leys 2009). CEMSYS estimates the onsite soil loss and the offsite dust concentration at national, state and (some) regional scales and provides monthly maps of wind erosion rates at 50-kilometre resolution for the Australian continent and 10-kilometre resolution for New South Wales. The spatial resolution for wind erosion modelling is limited by the resolution of the GIS data (soil texture, soil type, vegetation, roughness) and that of

Remote sensing specifications for monitoring ground cover nationally


30

the atmospheric model (currently 10-kilometre resolution) (Leys et al. 2009, pp. 114–119). A number of improvements to CEMSYS are proposed to enable its use in modelling trends in wind erosion. These include improving the reliability of the cover estimates by incorporating fractional cover data.

Spatial and temporal scalesFor ground cover (woody and non-woody), the following specifications are required of data layers for use in modelling wind erosion (using CEMSYS) at the national scale (with applicability at the state and regional scale):

• spatial resolution: 500 to 1000-metre resolution• temporal resolution: monthly• domain would be: Australia-wide (lower left corner 112 east to 45 south, upper right corner

155 east to 10 south)• preferred file format would be ARC-ACSII format.

Non-woody coverIdeally fractional non-woody cover estimates would be classed as follows:

• physical soil crusts• disturbed/loose soil• cryptogamic soil crusts• live (that is, photosynthesising)• dead vegetation cover (that is, non-photosynthesising vegetation)• litter• rock.

Interim non-woody fractional cover estimates could be classed as follows:

• photosynthetic vegetation (PV)• non-photosynthetic vegetation (NPV)• bare ground (BG).

Woody coverFractional woody cover (proportion of photosynthetic vegetation contributing to the overstorey) is also required for:

• spatial resolution: 500 to 1000-metre resolution• temporal resolution: annual• domain would be: Australia-wide (lower left corner 112 east to 45 south, upper right corner


31


Water erosion modelling for resource condition monitoringCurrently SedNet (Sediment and nutrient budget model (Wilkinson et al. 2004)), which includes the Revised Universal Soil Loss Equation (RUSLE), is recommended for modelling water erosion nationally down to the regional scale (Leys et al. 2009). WaterCAST, being developed at the eWater Cooperative Research Centre, is viewed as having the potential to replace SedNet in the future for assessment at catchment level and possibly state and national level (Leys et al. 2009) but at present data are inadequate for model parameterisation.

SedNet models sediment loads from hillslope, gully and streambank erosion and models nutrient loads providing long-term annual average estimates at 5-kilometre resolution. Improved differentiation of bare ground and dead or bleached vegetation in the satellite image interpretation used as measures of cover in the RUSLE calculations would improve model outputs (Hairsine et al. 2009). Higher resolution inputs such as for land use and vegetation, the introduction of a land management term in addition to a cover term and improved soils data have also been identified as priority improvements needed in the assessment of soil erosion by water (following NLWRA program of 2001–2003) (see Leys et al. 2009, p. 43). Some of these issues have been addressed for case studies but have yet to be applied nationwide.

Spatial and temporal scalesFor ground cover, the following specifications are required of data layers for use in modelling water erosion (using SedNet) at the national to regional scale:

• spatial resolution: 500 to 1000-metre resolution• temporal resolution: monthly• domain would be: Australia-wide (lower left corner 112 east to 45 south, upper right corner


Non-woody coverFractional non-woody cover estimates are required for:

• photosynthetic vegetation (PV)• non-photosynthetic vegetation (NPV) • bare ground (BG)• rock.

Interpretation of fractional cover into management practice classes (such as crop versus fallow) would improve cover estimate in RUSLE.

Monitoring agricultural land management practicesCaring for our Country sustainable practices funding is focused on improving the uptake of practices which will reduce soil loss through wind and water erosion, manage the soil


32

acidification risk and increase soil carbon storage. Practices of particular interest include those which increase the proportion of ground cover present year round.

The biennial ABS Agricultural Resource Management Survey provides information reported by farmers on the uptake of land management practices at NRM region level for Caring for our Country reporting. There may be opportunities for providing an additional line of evidence on the impact of certain management practices on ground cover at a small number of sites within the intensively used agricultural zone.

Spatial and temporal scalesFor national, state and regional assessment of ground cover levels to support Caring for our Country reporting, the following are required:

• spatial resolution: generally 30-metre but up to 500-metre resolution (influenced by paddock size—resolution less if want to assess within paddock variation)

• temporal resolution: annually (at minimum cover or at time of most risk) to monthly— time-series required

• coverage: rangelands and intensive land use zones of Australia.

Non-woody coverFractional non-woody cover estimates are required for:

• photosynthetic vegetation (PV)• non-photosynthetic vegetation (NPV)• bare ground (BG).

Within the intensively used agricultural zone, methods to measure fractional cover and to interpret remotely sensed fractional cover estimates into key management practices related to ground cover maintenance should be trialled at selected sites. Such management practices include:

• cultivated fallow (bare soil, timing and length)• crop residue (left or removed; timing and length)• grazing pressure (ground cover (PV+NPV) > 50 per cent).

Stewart and Rickards (2010) indicate potential options for monitoring land management practices using remote sensing in the intensively managed agricultural zone.

Calibration and validation of remotely sensed ground cover productsA remotely sensed ground cover product is underpinned by on-ground measurement of the features being estimated. To implement production nationally, agreement must be reached

33


on the location of ground control sites and what is measured at each site. A separate issue is agreement on the processing steps to produce the remotely sensed ground cover product.

To establish a national network of control sites for remote sensing of ground cover, the following steps are suggested:

1 Develop and evaluate a core set of attributes for describing and mapping ground control sites.

2 Review and test methods needed for field-based collection of new ground control sites to ensure suitability of use in all jurisdictions for all agricultural systems (including the rangelands).

3 Review existing sites against the agreed core set of attributes.

4 Undertake a statistical analysis to identify the numbers and likely locations of sites needed to adequately represent spatial variability of all agricultural systems.

5 Compare existing sites with sites needed according to the statistical analysis to identify gaps in the national ground reference sites database.

6 Develop an investment plan (taking into account available budget) to collect the data needed for additional ground control sites.

7 Agree arrangements, methods, costings and timelines with agencies participating in the on-ground collection.

8 Collate ground control site data.

9 Make ground control site dataset publicly available via the web.

Methods for on-ground measurement of ground cover fractions should be assessed on the usefulness of the information to inform reporting, capacity to migrate methods across environments, and suitability for multiple sensors (for example, MODIS and Landsat). These control sites will need to meet the needs of image-based ground cover mapping at national, state and regional scales. Establishment of new control sites to improve the reliability of the remotely sensed outputs must be prioritised.

34

appe

ndix

34

DExample 1: Leaf area indexFigure 5 shows how an upscaling methodology can be used to extend the spatial domain of measurements initially made over a much smaller area than the final relationships are used for. In this example, 1 m2 field observations of leaf area index (LAI) in homogeneous herbaceous (that is, cropping and grazing) fields were destructively sampled, and step one (McVicar et al. 1996a) involved relating this site data to Landsat TM data, enabling production of a Landsat TM based LAI image. Step two (McVicar et al. 1996b) involved developing relationships between the Advanced Very High Resolution Radiometer (AVHRR) data and the Landsat TM based LAI image. The AVHRR LAI estimate was then estimated for the entire Murray–Darling Basin.

Example 2: Ground cover and biomassMilne at al. (2007) evaluated MODIS for estimating ground cover and biomass/feed availability in mixed tree-grass grazing systems (for example, tropical savannah). Foliage projective cover and biomass was collected at 100 by 100-metre field sites. Field measurements were related to the 1-kilometre MODIS cells in which they reside by weighting an observation based on a ratio between the scaled up Landsat TM (30-metre cells) bare ground data at 100 by 100-metre

Remotely sensed examples of upscaling

+G G+

G++D

+D

+G G+

G++D

+D

A remotely sensed upscaling method to spatially extend data beyond the isolated �eld measurement sites

1000 km

1000 km

180 km

5AVHRR data for 1 000 000 km2

at 1 km resolution

ground measurements fromalready existing �eld stationsof soil and water parameters

Landsat TM data for 32400 km2

at 30 m2 resolution+G G+

G++D

+D180 km

35


(the field site area) and the scaled up Landsat TM bare ground at the MODIS 1-kilometre pixel resolution. The project centred on Charters Towers and the Brisbane Valley but the methodology (or upscaling technique) could be equally applied across all of Queensland.

Example 3: Fractional coverGuerschman et al. (2009) linked field-level spectral radiometric observations (made with a portable spectroradiometer from Analytical Spectral Devices, Inc) to Hyperion (30-metre resolution) hyperspectral satellite data. End-members relating to the pure components of photosynthetic vegetation, non-photosynthetic (senescent) vegetation and bare soil could be distinguished in both sets of hyperspectral data. To upscale to the MODIS resolution (a broadband instrument, like Landsat, which does not have all the hyperspectral bands of the spectroradiometer and Hyperion), they developed spectral surrogates using an optimally selected set of MODIS bands to capture the information content in the hyperspectral imagery. This enabled the unmixing of two MODIS indices to map the fraction of photosynthetic vegetation, senescent vegetation and bare soil at the MODIS resolution.

36

appe

ndix

36

EGeneric_Site_20060221_RCH

- 1 -6

Site Description

Site Number:_______________ Date:___________

Zone:____ Datum:_______ Easting:____________ Northing:____________

Bearing (to GPS):_______ Distance (to GPS):_______

Film No:_____ Photo Nos:________________________________

Description:__________________________________________________________________________

__________________________________________________________________________

__________________________________________________________________________

Landform (Crest; Upper; Mid; Lower; Flat; Closed; Open):__________

Slope %:_____ Aspect:_____

Soil Colour:

Crust Dry Dist. Dry Adtnl. Dry Crust Wet Dist. Wet Adtnl. Wet Soil Hue Soil Value SoilChroma

Cryptogam %:_____ Cryptogam colour:__________

Rock/Lag Colour:

1st Dominant 2nd Dominant 3rd Dominant 4th Dominant 5th Dominant Rock Hue Rock Value RockChroma

Rock/Lag % est.:______ Biomass est. (kg/ha):___________

Grass height (cm):_____ Fire:___________

Faunal activity:_____ Faunal type (Cattle; Sheep; Macropod; Termites; Ants; Locust):_____

Crust Brokenness (Extensively broken; Moderately broken; Slightly broken; Intact):_____

Erosion Features (Rills; Terracettes; Sheeting; Scalding; Hummocking; Pedestalling):_____

Deposited Material (Extensive; Moderate; Slight; Insignificant):_____

Soil Microtopography (Smooth; Shallow depressions; Deeper depressions; Deep formations; Very deep and extensive):_____

Surface Nature (Non-brittle; Very hard; Moderately hard; Easily broken; Loose-sandy):_____

SLATS generic site data form

37


Generic_Site_20060221_RCH

- 2 -6

Vegetation description:

1st Dom by biomass % 2nd Dom by biomass % 3rd Dom by biomass %

Overstorey

Understorey

Grasses/Forb

Overstory height: Observer:________ % Slope at 20m:________

TBA:

Prism factor Observer Live Dead Killed by fire Converted

Centre 1

North 2

NE 3

SE 4

South 5

SW 6

NW 7

Total Live TBA (Sum Converted TBA / 7):______m2/ha

Weights: No. Q’s:_____ Total Wet Weight :__________g Sub Sample Wet Weight :__________g

Sub Sample Dry Weight :___________g TSDM :___________kg/ha


38


- 3 -6

Site TransectSite Number:_______________ Sheet Number:_____ Date:__________ Bearing:_______

Zone:_____ Datum:__________ Easting:__________ Northing:__________

CRUST DIST ROCKGREEN

LF DEAD

LF LITTER CRYPT MIDGR

MIDDE

MIDBR

INCRWN

OVER GR

OVER DE

OVER BR CRUST DIST ROCK

GREEN LF

DEAD LF LITTER CRYPT

MIDGR

MIDDE

MIDBR

INCRWN

OVER GR

OVER DE

OVER BR

1 51

2 52

3 53

4 54

5 55

6 56

7 57

8 58

9 59

10 60

11 61

12 62

13 63

14 64

15 65

16 66

17 67

18 68

19 69

20 70

21 71

22 72

23 73

24 74

25 75

26 76

27 77

28 78

29 79

30 80

31 81

32 82

33 83

34 84

35 85

36 86

37 87

38 88

39 89

40 90

41 91

42 92

43 93

44 94

45 95

46 96

47 97

48 98

49 99

50 100

SUB

TTL

39

Ground cover monitoring for Australia ABARES Workshop proceedings Generic_Site_20060221_RCH

- 4 -6

Site Number:_______________ Sheet Number:_____ Date:__________ Bearing:_______



LF DEAD


MIDDE

MIDBR

INCRWN

OVER GR

OVER DE


GREEN LF


MIDGR

MIDDE

MIDBR

INCRWN

OVER GR

OVER DE

OVER BR

1 51

2 52

3 53

4 54

5 55

6 56

7 57

8 58

9 59

10 60

11 61

12 62

13 63

14 64

15 65

16 66

17 67

18 68

19 69

20 70

21 71

22 72

23 73

24 74

25 75

26 76

27 77

28 78

29 79

30 80

31 81

32 82

33 83

34 84

35 85

36 86

37 87

38 88

39 89

40 90

41 91

42 92

43 93

44 94

45 95

46 96

47 97

48 98

49 99

50 100

SUB

TTL


40


- 5 -6

Site Number:_______________ Sheet Number:_____ Date:__________ Bearing:_______



LF DEAD


MIDDE

MIDBR

INCRWN

OVER GR

OVER DE


GREEN LF


MIDGR

MIDDE

MIDBR

INCRWN

OVER GR

OVER DE

OVER BR

1 51

2 52

3 53

4 54

5 55

6 56

7 57

8 58

9 59

10 60

11 61

12 62

13 63

14 64

15 65

16 66

17 67

18 68

19 69

20 70

21 71

22 72

23 73

24 74

25 75

26 76

27 77

28 78

29 79

30 80

31 81

32 82

33 83

34 84

35 85

36 86

37 87

38 88

39 89

40 90

41 91

42 92

43 93

44 94

45 95

46 96

47 97

48 98

49 99

50 100

SUB

TTL

41



- 6 -6

Site selection criteria

• Sample the full range of cover from 0 to 100%. • Sample a range of soil colours and textures. • Sample a range of TBA from 0 to approx 10m2/ha (although upper limit yet to be determined) • Sample a range of understorey cover amounts. • Ideally locate paired sites across fence lines sampling a contrasting cover on either side. • Locate in areas of uniform density and species mix for a minimum patch size of 300 x 300m if possible (larger if non DGPS). • Locate a minimum of 100 metres from roads or other features not characteristic of the vegetation being measured (from the edge of site). • Locate away from water run-on areas if possible. • On level or near level ground. • If a sloped site is necessary, avoid western and southern slopes (these are affected by shadow due to winter and morning sun angles).

Transect recording

• Minimum 300 sample points (100m transect tape x 3). • Transect tapes laid strictly in a 0, 60 and 120 deg star pattern. The site centre is located at the intersection of the star. • Measure in order of 0 – 180 deg, 60 – 240 deg and 120 – 300 deg for consistency. • Averaged DGPS or GPS position at site centre (minimum 3-5min averaging for DGPS, longer for non DGPS). Recorded as UTM (Easting and

Northing) with associated Zone and using default Datum WGS84. (Use UTM, Datum GDA94, Spheroid GRS1980 if available on GPS). • Care should be taken to lay the tape out in a straight line - not dodging trees or shrubs. • Care should be taken when placing the pole at each metre mark - read the ground cover first by looking vertically (or using a thin steel wire rod) above

the tape thus avoiding crushing or moving a prospective grass/litter attribute. A laser pointer taped to the pole is more useful in areas with significant understorey e.g. heath.

• Before recording make sure both the observer and recorder are clear as to the classes being measured. Call in order of ground layer, mid layer, overstorey e.g. crust; mid (storey) green (leaf); crown; over (storey) branch.

• If recording hard copy sheets, care must be taken when reducing the measurements - have someone check the adding up and also the logic.• If using a GRS Densitometer, both spirit levels must be centred and the recording spot centred in target circle.• If using a gimballed sighting tube, the tube MUST BE FREE to ROTATE on the gimbal, otherwise operator bias WILL BE INTRODUCED.

Optical Wedges/Prisms

• When two people first go into the field they should both take BA wedge measurements at several sites to check if one or the other is systematically measuring more or less than the other [strict blind testing procedure advised].

• The possible reasons for bias are: o instead of counting every second "split" tree as “in”, the operator leaves all “out” or puts all “in”...(error = 0.5xthe number of "split" trees) o not keeping the wedge at the site centre – i.e. rotating the wedge around the body at arms length (error is most significant where site

position is amongst bushes that may be counted as solid trunks if the wedge is close enough)o not looking behind large trees near the observer – can move off centre to check, keeping perpendicular...(error nil to quite large)o using the wrong thickness wedge relative to the tree/bush trunk thicknesses (error can vary depending upon site and or the operator) o poor eyesight - not being able to clearly see distant trees - thus leading to a preference for using a wedge that has a shorter

measurement range (error generally in reading less trees than exist at the site) o not measuring/looking at the tree trunk at breast height – 1.3 m (error depends on tree thickness but if looking low the counts will be

higher than they should be)

The following table should be used to help select a suitable prism for a basal area measurement. Try to choose a prism which will give a plot radius of 15-20m. If the plot radius is larger make sure that its in a uniform area.

Plot radii (metres) as a function of prism factor and tree diameter

Tree Diameter (m)

Prism Factor 0.1 0.3 0.5 0.7 1.0 1.5

0.2 11 34 56 78 112 168 0.25 10 30 50 70 100 150 0.7 6 18 30 42 60 90 0.8 6 17 28 39 56 84 0.9 5 16 26 37 53 79 1.0 5 15 25 35 50 75 2.0 4 11 18 25 35 53 2.25 3 10 17 23 33 50 2.5 3 10 16 22 32 47

Optical Wedge Calibration

1. Set up a target such as a white board or wall at least 5 metres from where the wedge is held. Use up to 10 metres if you can see well as a greater distance is better.

2. Place a vertical line on the target using a marker pen. Hold the wedge at the measured distance (d) from the target and look through it at the line. Get a second person to draw another line, marking the displacement seen through the wedge.

3. Measure this distance (w) in the same units as (d).

4. Calculate the basal area factor as follows: Basal Area Factor (BAF) = 1000/ (1+4*(d/w)2)

where d = distance to target; w = displacement width e.g. d = 5m, w = 0.1m; BAF = 1.0

Groundcover

Tree & G. C.

42

ReferencesABS (Australian Bureau of Statistics) 2009, Land management and farming in Australia 2007–08

(4627.0), viewed 29 December 2010, www.ausstats.abs.gov.au/Ausstats/subscriber.nsf/0/7CAF008176640264CA2575C40017E313/$File/46270_2007-08.pdf.

Baldock, JA, Grundy, MJ, Griffin, EA, Webb, MJ, Wong, MTF and Broos, K 2009, Building a foundation for soil condition assessment, report to the Australian Government’s Caring for our Country, CSIRO Land and Water Science Report, viewed 29 December 2010, www.nrm.gov.au/publications/books/pubs/foundation-soil-condition.pdf.

Bastin, G and the ACRIS (Australian Collaborative Rangelands Information System) Management Committee 2008, Rangelands 2008 – Taking the pulse, National Land and Water Resources Audit, Canberra.

Bui, EN, Hancock, GJ, Chappell, A and Gregory, LJ 2010, Evaluation of tolerable erosion rates and time to critical topsoil loss in Australia, report to the Australian Government’s Caring for our Country, CSIRO National Research Flagships Sustainable Agriculture, viewed 29 December 2010, www.nrm.gov.au/publications/books/pubs/tolerable-erosion.pdf.

BRS (Bureau of Rural Sciences) 2009, MCAS-S: Multi-Criteria Analysis Shell for Spatial Decision Support Version 2.1, User Guide, Commonwealth of Australia, viewed 4 January 2011, www.adl.brs.gov.au/mcass/docs/User_Guide.pdf.

Clarke, K, Lewis, M, Ostendorf, B and Dutkiewicz, A 2010, ‘Towards broadscale monitoring of soil erosion risk in southern Australia through relative spectral mixture analysis of MODIS imagery’, Proceedings of the 15th Australasian Remote Sensing & Photogrammetry Conference, viewed 4 January 2011, www.scribd.com/doc/37527339/15arspc-Submission-6.

Forward, G 2009, Manual of proposed national minimum standards for roadside erosion survey, report to the Australian Government’s Caring for our Country, DWLBC Report 2009/24, Government of South Australia, through the Department of Water, Land and Biodiversity Conservation, Adelaide, viewed 29 December 2010, www.nrm.gov.au/publications/books/pubs/standards-roadside-erosion.pdf.

Guerschman, JP, Hill, MJ, Renzullo, LJ, Barrett, DJ, Marks, AS and Botha, EJ 2009, ‘Estimating fractional cover of photosynthetic vegetation, non-photosynthetic vegetation and bare soil in the Australian tropical savanna region upscaling the EO-1 Hyperion and MODIS sensors’, Remote Sensing of Environment, vol. 113, pp. 928–45, viewed 9 June 2010, www.sciencedirect.com/science?_ob=MImg&_imagekey=B6V6V-4VPV8TF-1-1M&_cdi=5824&_user=768459&_pii=S0034425709000121&_orig=search&_coverDate=05%2F15%2F2009&_sk=998869994&view=c&wchp=dGLbVzz-zSkzS&md5=6f54f3d34442f91457bd6140e1d9876c&ie=/sdarticle.pdf.

Hairsine, PB, Barson, M, Randall, L and Wilkinson, SN 2009, The 2009 expert panel process in support of the spatial targeting of the Caring for our Country investment concerning hillslope erosion by water, report to the Australian Government’s Caring for our Country, CSIRO Land and Water Science Report 45/09, viewed 15 January 2010, www.nrm.gov.au/business-plan/10-11/priorities/sustainable/pubs/report2-wind-erosion.pdf.

Lymburner, L, Tan, P, Mueller, N, Thackway, R, Thankappan, M, Islam, A, Randall, L and Lewis, A 2010, ‘Australia’s first dynamic land cover map’, Proceedings of the 15th Australasian Remote Sensing & Photogrammetry Conference, 13–17 September 2010, Alice Springs, viewed 5 January 2010, www.scribd.com/doc/37457498/15arspc-Submission-96.

43


Leys, J, Butler, H, Yang, X and Heidenreich, S 2010, CEMSYS modelled wind erosion, report to the Australian Government’s Caring for our Country, Department of Environment, Climate Change and Water NSW, viewed 29 December 2010, www.nrm.gov.au/publications/books/pubs/cemsys-wind-erosion.pdf.

Leys, J, Smith, J, MacRae, C, Rickards, J, Yang, X, Randall, L, Hairsine, P, Dixon, J and McTainsh, G 2009, Improving the Capacity to Monitor Wind and Water Erosion: A Review, Department of Agriculture, Fisheries and Forestry, Commonwealth of Australia, viewed 15 January 2010, www.nrm.gov.au/publications/books/pubs/wind-water-erosion.pdf.

McTainsh, G, Strong, C and Leys, J 2010, Wind erosion histories, model input data and community DustWatch, report to the Australian Government’s Caring for our Country, Atmospheric Environment Research Centre, Griffith University, Brisbane, viewed 29 December 2010, www.nrm.gov.au/publications/books/pubs/wind-erosion-histories.pdf.

McVicar, TR, Jupp, DLB, Reece, PH and Williams, NA 1996a, Relating LANDSAT TM vegetation indices to in situ leaf area index measurements, Technical Memorandum 96.14, CSIRO Division of Water Resources, Canberra, ACT, viewed 15 January 2010, www.clw.csiro.au/publications/technical96/DWR_TR96-14.pdf.

McVicar, TR, Jupp, DLB and Williams, NA 1996b, Relating AVHRR vegetation indices to LANDSAT TM leaf area index estimates, Technical Memorandum 96.15, CSIRO Division of Water Resources, Canberra, ACT, viewed 15 January 2010, www.clw.csiro.au/publications/technical96/DWR_TR-96-15.pdf.

Milne, J, Danaher, TJ, Scarth, P, Carter, JO, Armston, J, Henry, B, Cronin, N, Hassett, R, Stone, G, Williams, P, Denham, R, Byrne, M and Gillingham, S 2007, Evaluation of MODIS for groundcover and biomass/feed availability estimates in tropical savannah systems, final report to Meat & Livestock Australia (Project Code NBP.330), Queensland Department of Natural Resources and Water, Climate Impacts and Natural Resource Systems, Indooroopilly, Brisbane.

Mitchell, RM, Qin, Y and Campbell, SK 2010, Dust sentinel pilot: capacity of three polar orbiting satellite instruments to map dust aerosol over the Australian continent, report to the Australian Government’s Caring for our Country, CSIRO Marine and Atmospheric Research, viewed 29 December 2010, www.nrm.gov.au/publications/books/pubs/dust-sentinal-pilot.pdf.

NCST (The National Committee on Soil and Terrain) 2009, Australian Soil and Land Survey Field Handbook, 3rd edn, CSIRO Publishing, Collingwood, Victoria.

Ringrose-Voase, A in preparation, Establishing the coordinated national soil condition monitoring project, report to the Australian Government’s Caring for our Country.

Scarth, P, Byrne, M, Danaher, T, Hasset, R, Carter, J and Timmers, P 2006, State of the Paddock: Monitoring condition and trend in groundcover across Queensland, Queensland Department of Natural Resources and Mines, Indooroopilly, Brisbane.

Scarth, P, Roder, A and Schmidt, M 2010, ‘Tracking grazing pressure and climate interaction – the role of Landsat fractional cover in time series analysis’, Proceedings of the 15th Australasian Remote Sensing & Photogrammetry Conference, viewed 4 January 2011, www.scribd.com/doc/37455672/15arspc-Submission-140.

Schmidt, M, Denham, R and Scarth, P 2010a, ‘Fractional ground cover monitoring of pastures and agricultural areas in Queensland’, Proceedings of the 15th Australasian Remote Sensing & Photogrammetry Conference, viewed 4 January 2011, www.scribd.com/doc/37455826/15arspc-Submission-119.


44

Schmidt, M, Tindall, D, Speller, K, Scarth, P and Dougall, C 2010b, Ground cover monitoring with satellite imagery in agricultural areas and improved pastures, final report to the Bureau of Rural Sciences, State of Queensland (Department of Environment and Resource Management).

Shao, Y, Leys, J, McTainsh, G and Tews, K 2007, ‘Numerical simulation of the October 2002 dust event in Australia’, Journal of Geophysical Research, vol. 112, pp. 1–12.

Smith, J and Leys, J 2009, Wind erosion priority areas for Caring for our Country expert panel prioritisation process, report to the Australian Government’s Caring for our Country, Bureau of Rural Sciences, Canberra, viewed 15 January 2010, www.nrm.gov.au/business-plan/10-11/priorities/sustainable/pubs/report2-wind-erosion.pdf.

Stewart, JB and Rickards, JE 2010, Developing national protocols to map ground cover management practices in cropping and grazing systems: Report on an Australian Collaborative Land Use and Management Program workshop, 12–13 May 2009, Brisbane, Bureau of Rural Sciences, Canberra, viewed 29 December 2010, http://adl.brs.gov.au/data/warehouse/pe_brs90000004194/ACLUMP_workshopReport_20100301_ap14.pdf.

Wilkinson, S, Henderson, A, Chen, Y and Sherman, B 2004, SedNet user guide, client report to the Cooperative Research Centre for Catchment Hydrology, CSIRO Land and Water, Canberra, viewed 15 January 2010, www.toolkit.net.au/Tools/SedNet/documentation.

australian government - data.daff.gov.audata.daff.gov.au/data/...mapping-workshop_proc_11.pdf ·...

Documents