Linking RS to Land Management & Condition

Get Started. It's Free
or sign up with your email address
Linking RS to Land Management & Condition by Mind Map: Linking RS to Land Management &  Condition

1. Understanding the Users

1.1. Who are they?

1.1.1. Landholders

1.1.2. Regional Bodies

1.1.3. Policymakers

1.1.4. Landholders

1.1.5. National Reporting

1.1.6. Woody Carbon

1.2. Needs

1.2.1. Temporal

1.2.2. Spatial

1.2.3. Accuracy

1.3. Providing Data and Getting Feedback

1.3.1. GeoWIKI

1.3.2. FORAGE

1.3.3. Auscover

1.3.4. Rangelands Alliance

1.4. vLEARM

1.4.1. Virtual laboratory Deliver improvements to science Transform use and intergration of remote sensning data

1.4.2. Needs Measure Impact of investment Cannot consistently report on extent and condition of assets Remove barriers to adoption Simplify access without black box

1.4.3. Objective Simplify user discover and access Simplify tools

1.4.4. Proposal Consortium approach Clearing house Storage architecture for full time series of data Series of workbenches (tools and models Land Condition Productivity Environmental accounting

1.4.5. Existing infrastructure AusCover Giovanni AuScope Virtual geophysics lab Web processing Google Earth Engine Maps Engine

1.4.6. System Architecture Ingest & aggregate Transform to grid Query Filter & Statistics Visualise

1.4.7. Ground cover and land condition assessment use case example

1.4.8. Next Steps Vision Governance Identify use cases Land condition Water balance models Carbon Models Road Map

2. Measuring & Mapping Land Condition

2.1. Land Management

2.1.1. Types of land management

2.1.2. Incentives for change

2.1.3. Building guidelines

2.2. Rangeland Monitoring

2.2.1. State Based Methods NSW RAP Sites NT 2 Tier system Pastoral monitoring program Tier 1 Subjective Tier 2 Quantitative WARMS Measure regulary SA PMS Up to 14 years between assessments Qld TRAPS QGRAZE Delbessie Remote Sensing

2.2.2. ACRIS Objective understanding of management effects Present using Change matrix Seasonal conditions vs Landscape function change Components of total grazing pressure Land Use Pastoral land values Integrate all the above information Drivers acting on... Outcomes

2.3. Operational Products

2.3.1. Forage

2.3.2. AussieGRASS

2.3.3. Phil's rangelands Initiative

2.3.4. Pastures from Space

2.4. Measuring

2.4.1. Delbessie

2.4.2. NT Tier 1 & 2

2.4.3. RAP Sites

2.4.4. Step points SA Vic

2.4.5. Windshield Standing Cover Horizontal Cover Height Detachment

2.4.6. Rob Hassets Biomass Data?

2.5. Climate Effects

2.5.1. SILO

2.5.2. AWAP

2.5.3. BOM

2.6. Historical context

2.6.1. Need to go back to 70's for historical context

2.6.2. post-2000 only captures one el nino effect

2.7. Beef R&D

2.7.1. Beef RD&E Strategy Qld has half national herd LNP Agriculture Strategy Northern Beef research aliance

2.7.2. Impediments to productivity growth Reproduction inefficiency Buisness skills Stocking rate decisions Climate change adaption Carbon economy Skills shortages Disease & parasites Shrinking RD&E pool Variable eating quality

2.7.3. Oppertunities Reproduction Herd management Nutrition Matching stocking rate to carrying capacity Precision grazing Communication Extension is the key Naturally fed etc. Collaboration is essential Train the next generation

2.7.4. Improving tool to improve productivity Economic analysis essential

2.7.5. Grazing pressure/Utilization Biomass estimation is essential ABCD Land Condition Woody thickening of weedy shrubs mask RS cover estimates

2.7.6. SPYGlass For integrated research extension

2.7.7. Vegmachine Tool to summarise time series satellite data Hands on communication tool

2.7.8. Forage Email delivery

2.7.9. FutureBeef "Supercharged Extension"

2.7.10. Grazing BMP Voluntary Online self assessment tool Outcomes Producer Industry

2.7.11. Remote Sensing Extension R&D Producers are busy Land condition affects nutrition

2.8. Dustwatch

2.8.1. Modelling

2.8.2. Measurements

2.8.3. OEH Remote Sensing Program Local validation of RS products Using seasonal cover

2.8.4. Logic Where is the dust coming from From monitoring units at 44 sites Dust Rose compared with Wind Rose Integrate with MODIS data

2.8.5. Sensors MODIS best for responsive strategic decisions Landsat - best for monitoring investments

2.8.6. Types of Cover Horizontal Vertical Lateral Cover Major issue for wind erosion Frontal Area Index

2.8.7. Untitled

3. Calibration & Validation of cover & "condition"

3.1. Sampling guidelines

3.1.1. Local

3.1.2. National

3.1.3. What data is fit for purpose?

3.1.4. Considerations Soils Climate Vegetation Fire Paired Sites Access Political Untitled

3.2. Groups

3.2.1. ABARES Site Selection National sampling strategy Site Characterisation Field procedure Data collation 524 Funded sites 1139 Observations 90 Sites with multiple observations On AEKOS & NCI & Geodatabase Data analysis Flag analysis Majority of sites high cover Majority of values from red soils Qld data no colour data Soil moisture affects NPV Future Investment priorities Improve spatial and temporal accuracy Improve fractional cover model performance Provide product continuity and real time capability

3.2.2. Auscover

3.2.3. Ausplots

3.2.4. States

3.2.5. Regional Groups

3.2.6. The public

3.3. Observer experience levels

3.3.1. Based on Star transects

3.3.2. Why variability in data Sampling method Observer Incorrect application of method

3.3.3. Experiment 4 Experienced 4 inexperienced 10 Simplified 100m transects Statistical tests Wilcoxon rank sum Residual analysis Generalised Linear Mixed Model Logistic regression Mixed model Binomial model Findings No systematic bias between groups Average cover for each experience group was the same Not all variation explained by sampling Results more variable at complex sites Experienced group more precise

3.4. Data collection

3.4.1. Methods

3.4.2. Standards

3.4.3. User errors

3.4.4. Metadata

3.4.5. Storage

3.5. Spectral Data Exchange

3.5.1. Huge variety of Field spectroscopy methods ACEAS workshop DC10 Project Speccio TERN/Auscover

3.5.2. Context Challenge is to make good measurements Discoverable data Storage and reuse of data

3.5.3. Metadata is critical

3.5.4. National ACEAS Workshop Drive best practice SPECCIO proposed as international tool Draft metadata requirements Tools to summarise completeness and quality of spectral datasets Quality analysis algorithms? Best practice guidelines to improve data collection

3.5.5. SPECCIO v3 Web server backbone Testing deployment in June 2013 Moving to open source

4. Fractional Cover Methods & Comparison

4.1. Methods

4.1.1. Landsat Unmixing Policy Drivers Original Bare Ground Index Fractional Landsat Cover Inversion of regression coefficients to build endmembers

4.1.2. CSIRO 2009 Algorithm semi-operational product Cellulose and Lignin Feature (CAI) Photosynthetic Feature (NDVI) Applied to Hyperion Applied to MODIS Validation in 2012 Hyperspectral Models the data well Spectral aggregation did not degrade the model much SLATS Sites Site heterogenaity Model was recalibrated to elimate bias in three fractions (version 2.2)

4.1.3. Soil Exposure in SA Drivers NRM Act 2004 Soil Conservation essential Cereal cropping essential to economy DEWNR Monitors soil erosion Rapid field survey (RFS) methodology Ken - 2 MODIS indicies Early NDVI methods LCI1 MODIS 6 and 7 R2 of 0.33 PD14 R2 or 0.34 More validation sites at MODIS Scale NDVI v Soil R2 0.21 LCI v Soil R2 0.44 GREG's stuff Links to reporting requirements Time Series Plots Time series summaries Area vulnerable to erosion Looking at Operation RS Magnitude and duration of soil exposure Concerned about data continuity

4.1.4. Dynamic Land Cover Uses EVI Time series Version 1 eleased in 2011 34 ISO Classes via SVM Version 2 Areas flagged in version 1 reviewed Dynamic Markov Chain Map product now every 2 years Dynamic approach allow you to look at change Validation Rangeland applications Changes in shrubland extent Fires Flooding Provides context for Fractional Cover Applications Continuous field for interpretation of data Trends Untitled Future applications ULA 25m NBAR Data time series Now have fractional cover

4.1.5. Greg Okin Interested in Wind erosion Spectral MIxture overview Most of the spectral variation is in the soil SWIR only place where Soil and Brown differ Even worse in MODIS bands Problems Simple Spectral MIxture Analysis Problem is that vegtation is a mix of green and brown Multiple endmember spectral mixture analysis Originally using AVIRIS data Problem is that you get spectral coupling - combinations of endmembers look like other combinations RSMA Reformulate SMA All retrievals are then relative - X<0 Less than t0, X>0 Greater than t0 Vegetation anomaly product In some areas time series give cyclic behaviour, others episodic Most of the variation is in "brown" Plots of phase difference between green and brown Evaluation in South Australia Correlation and regression analysis of EVI, RMSA, SMA, MESMA and Normalised (RMSA, SMA) See presentation forresults. RSMA did very well

4.2. Scale Issues

4.2.1. Spectral

4.2.2. Temporal

4.2.3. Sparial

4.2.4. Radiometric

4.3. Geoscience Australia ULA

4.3.1. Aus Space Research Program 3.5M funding for Project Automated processing National Nested Grid Specification Pixel level time series information Move from tape to high performance data NCI Spinning disk

4.3.2. Consortium GA VPAC NCI CRC-SI Lockheed Martin

4.3.3. NBAR Processing post 2000 data processed

4.3.4. Use Auscover JRSRP method Fractional Cover

4.3.5. To better inform policy

4.3.6. Hypertemporal datacube How to structure Need tracability of data What ICT platform How to nest data of different types How to deal with irregular temporal acquistion Need to access in space and time What type of query interface? People want total flexibility Storage type still being determined User survey What type of applications Untitled

4.4. Operational Processing

4.4.1. USGS

4.4.2. Google

4.4.3. Continuity MODIS Landsat 8 Sentinel

4.5. Comparison

4.5.1. Why To improve Future site selection Inform funding bodies

4.5.2. Methods Accuracy Precision Scale

4.5.3. Interpretation Visual Landholder

4.6. Blending MODIS / Landsat

4.6.1. Simulate Landsat at MODIS resolution

4.6.2. 2 Study Sites Colembally 17 Dates Gwidir Test to predict flood extent

4.6.3. Algorithms STARFM ESTARFM Outperformed by STARFM for flood extent Linear Interpolation GEIFM - Global Empirical Image Fusion Model

4.6.4. Future - Australian Landsat Modis Blending Infrastructure

4.7. Derived Products

4.7.1. Composites 8/16 Day Seasonal

4.7.2. Time Series Statistics Decomposition Change

4.7.3. Land Management? DCRM Baby's head Percentiles Untitled

5. Outputs - where to next?

5.1. Wiki

5.2. Paper

5.3. Knowledge Gaps

5.4. Map products

5.5. Final Report

5.6. Future Sensors

6. What are the major achievements?

6.1. The Paper

6.1.1. Tim needs Good Figures

6.1.2. Write it in a week

6.2. Improving field data collection

6.2.1. Cross calibration between groups

6.3. Extension tools?

6.4. Frontal Area Index