GIS and Remote Sensing for Forest Monitoring Training Course

Participants can prepare spatial datasets using appropriate reference systems, structures and metadata.

• Coordinate reference systems selected for local measurement and reporting
• Vector geometry, topology and attribute domains checked before analysis
• Raster resolution, extent, cell size and resampling choices compared
• Geodatabases structured for boundaries, plots, observations and map outputs
• ISO 19115 metadata elements recorded for forest spatial datasets
• Data lineage documented from acquisition through processing and publication

Participants can integrate field observations and inventory plots with GIS layers.

• Plot designs compared for inventory, calibration and validation purposes
• Diameter, height and species observations linked to plot coordinates
• Field forms specify required attributes, codes and quality checks
• GPS positional error assessed against plot size and imagery resolution
• Joining plot records to spatial features using unique identifiers
• Sampling frames constructed from strata, accessibility and forest variability

Participants can select and interpret optical satellite imagery for forest monitoring tasks.

• Landsat and Sentinel-2 missions compared for forest monitoring requirements
• Spectral signatures distinguish vegetation, soil, water, shadow and built surfaces
• Near-infrared and shortwave infrared bands reveal vegetation condition
• Scene selection considers seasonality, cloud cover and phenological variation
• False-colour composites prepared for forest interpretation and feature review
• Image dates matched with field observations and reporting periods

Participants can interpret Sentinel-1 radar products for complementary forest observations.

• Synthetic aperture radar backscatter explained through surface and volume scattering
• Sentinel-1 polarisation channels interpreted for forest and nonforest separation
• Speckle filtering choices balanced against spatial-detail retention requirements
• Terrain correction addresses geometric distortion in mountainous forest regions
• Radar and optical imagery combined for persistent-cloud monitoring
• Backscatter limitations recognised for biomass estimation and degradation detection

Participants can prepare analysis-ready imagery and derived variables for classification.

• Atmospheric correction converts optical imagery into surface reflectance values
• Cloud, cirrus and shadow masks remove contaminated optical pixels
• Normalized Difference Vegetation Index calculated for vegetation greenness assessment
• Normalized Burn Ratio identifies burned vegetation and fire severity
• Texture measures derived for canopy structure and fragmentation interpretation
• Multi-date image stacks aligned before temporal classification analysis

Participants can design a class schema and training-data plan for forest mapping.

• Class definitions separate forest, plantations, shrubland, agriculture and water
• Anderson land-use classification principles adapted to project reporting needs
• Training samples distributed across spectral variation and geographic extent
• Reference labels checked using field data, imagery and local knowledge
• Class imbalance identified before supervised model development begins
• Decision rules documented for ambiguous edge and mixed-pixel classes

Participants can produce a supervised forest-cover classification from prepared inputs.

• Random Forest classification trained using labelled spectral and texture features
• Maximum Likelihood classification compared with Random Forest assumptions
• Feature selection removes redundant bands and unstable predictor variables
• Classification parameters recorded for repeatable annual map production
• Post-classification smoothing reduces isolated pixels without erasing boundaries
• Visual interpretation checks class outputs against high-resolution reference imagery

Participants can quantify map reliability and calculate area estimates from validation data.

• Confusion matrices summarise agreement between map labels and reference samples
• Producer's accuracy measures omission error for each mapped class
• User's accuracy measures commission error for each mapped class
• Stratified random sampling allocates validation points across mapped classes
• Area-adjusted estimates correct mapped totals using reference-sample results
• Confidence intervals communicate uncertainty around forest-area estimates

Participants can identify and characterise forest-cover change between reporting periods.

• Post-classification comparison identifies transitions between mapped land-cover classes
• Image differencing measures spectral change between aligned observation dates
• Change thresholds selected using reference data and error review
• Transition matrices quantify forest loss, gain and conversion pathways
• Minimum mapping units applied consistently across annual change products
• Change polygons attributed with date, driver, confidence and evidence source

Participants can use image time series to distinguish persistent change from seasonal variation.

• Seasonal composites reduce cloud contamination and short-term spectral noise
• Time-series charts reveal clearing, recovery and recurring burn patterns
• Break detection identifies abrupt vegetation changes within observation sequences
• Baseline periods established before setting disturbance-alert thresholds
• Phenology considered when interpreting vegetation-index fluctuations over time
• Alert prioritisation combines change magnitude, proximity and protection status

Participants can map indicators of degradation, fire impact and forest fragmentation.

• Canopy disturbance indicators distinguish degradation from complete forest conversion
• Burned-area mapping combines Normalized Burn Ratio and reference interpretation
• Fire severity classes calibrated against prefire and postfire imagery
• Edge density and core-area metrics quantify forest fragmentation
• Road proximity analysis identifies likely access-related disturbance pressure
• Degradation evidence documented with uncertainty and alternative explanations

Participants can develop spatial inputs supporting forest biomass and carbon estimates.

• Allometric equations convert tree measurements into aboveground biomass estimates
• IPCC Tier approaches compared for forest carbon estimation requirements
• Plot biomass values linked with spectral, radar and terrain predictors
• Regression models predict biomass surfaces from inventory calibration data
• Carbon stock maps report units, uncertainty and applicable assumptions
• Emission factors separated from activity data within forest-carbon accounting

Participants can align monitoring outputs with common forest and carbon reporting requirements.

• REDD+ activity data requirements linked to mapped forest-change categories
• National forest monitoring systems combine inventories, maps and quality assurance
• IPCC Good Practice Guidance informs transparent emissions estimation documentation
• Forest reference emission levels require consistent historical data series
• Quality assurance procedures assign review responsibilities and evidence records
• Reporting tables reconcile mapped areas, sampled estimates and uncertainty

Participants can derive spatial indicators for habitat condition and restoration progress.

• Habitat suitability layers combine forest structure, elevation and water proximity
• Connectivity analysis identifies corridors, barriers and isolated habitat patches
• Restoration baselines define measurable vegetation and survival indicators
• Reference ecosystems guide comparison of restored and natural forest condition
• Object-based image analysis distinguishes planted blocks from natural regeneration
• Monitoring dashboards link restoration sites with periodic spatial indicators

Participants can specify an operational workflow with roles, controls and reporting cycles.

• Standard operating procedures define inputs, steps, checks and approvals
• RACI matrices allocate monitoring responsibilities across technical and field teams
• Version control conventions protect reproducibility of maps and scripts
• Data-security classifications govern sensitive patrol and concession information
• Update schedules align imagery acquisition with management reporting cycles
• Escalation criteria trigger field verification after detected forest-loss alerts

Participants can communicate forest-monitoring results through clear maps, tables and briefings.

• Map layouts apply visual hierarchy, scale, legend and source conventions
• Uncertainty communicated using confidence statements, notes and map symbology
• Area statistics presented with units, dates, methods and error bounds
• Management briefs distinguish observed evidence from recommended actions clearly
• Web-map publication considerations include permissions, metadata and update dates
• Technical reports document methods sufficiently for independent review

Participants can build, review and present a complete documented forest-monitoring analysis for a defined management area.

• Project brief defines area, monitoring question, users and decisions
• Data inventory records imagery, field observations, boundaries and limitations
• Classification and change-detection methods selected for stated evidence requirements
• Accuracy assessment produces confusion matrix, area estimates and uncertainty statements
• Map package includes forest status, change, priority areas and metadata
• Peer review tests assumptions, traceability, interpretation and management recommendations